diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,172693 @@
+{
+  "best_metric": 0.2483615279197693,
+  "best_model_checkpoint": "./checkpoints/ckpt_8_6_r32_claude_all_fp16_lr5e-05_batch8/checkpoint-141500",
+  "epoch": 4.995953980930937,
+  "eval_steps": 500,
+  "global_step": 142000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.863039399624766e-08,
+      "loss": 0.796,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1726078799249532e-07,
+      "loss": 0.7443,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7589118198874298e-07,
+      "loss": 0.8128,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3452157598499064e-07,
+      "loss": 0.8111,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.931519699812383e-07,
+      "loss": 0.6772,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5178236397748596e-07,
+      "loss": 0.8012,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1041275797373365e-07,
+      "loss": 0.7822,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.690431519699813e-07,
+      "loss": 0.7554,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.276735459662289e-07,
+      "loss": 0.7366,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.863039399624765e-07,
+      "loss": 0.7378,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.449343339587242e-07,
+      "loss": 0.8267,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.035647279549719e-07,
+      "loss": 0.8,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.621951219512196e-07,
+      "loss": 0.8192,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.208255159474673e-07,
+      "loss": 0.8117,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.794559099437148e-07,
+      "loss": 0.7438,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.380863039399626e-07,
+      "loss": 0.7744,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.9671669793621e-07,
+      "loss": 0.7994,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0553470919324578e-06,
+      "loss": 0.7964,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1139774859287056e-06,
+      "loss": 0.7616,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.172607879924953e-06,
+      "loss": 0.7486,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2312382739212008e-06,
+      "loss": 0.7867,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2898686679174484e-06,
+      "loss": 0.7258,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.348499061913696e-06,
+      "loss": 0.7761,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4071294559099438e-06,
+      "loss": 0.7119,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4657598499061916e-06,
+      "loss": 0.7707,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5243902439024391e-06,
+      "loss": 0.7129,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5830206378986866e-06,
+      "loss": 0.7411,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6416510318949346e-06,
+      "loss": 0.781,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7002814258911821e-06,
+      "loss": 0.7319,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7589118198874296e-06,
+      "loss": 0.7559,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8175422138836774e-06,
+      "loss": 0.8009,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8761726078799251e-06,
+      "loss": 0.7054,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.934803001876173e-06,
+      "loss": 0.6913,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.99343339587242e-06,
+      "loss": 0.6887,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.052063789868668e-06,
+      "loss": 0.6825,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1106941838649157e-06,
+      "loss": 0.7041,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1693245778611634e-06,
+      "loss": 0.686,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.227954971857411e-06,
+      "loss": 0.7251,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2865853658536584e-06,
+      "loss": 0.6662,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.345215759849906e-06,
+      "loss": 0.7285,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.403846153846154e-06,
+      "loss": 0.6286,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4624765478424017e-06,
+      "loss": 0.6775,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5211069418386494e-06,
+      "loss": 0.6458,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5797373358348967e-06,
+      "loss": 0.6852,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.638367729831145e-06,
+      "loss": 0.736,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.696998123827392e-06,
+      "loss": 0.6177,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.75562851782364e-06,
+      "loss": 0.6665,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8142589118198877e-06,
+      "loss": 0.6038,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.872889305816135e-06,
+      "loss": 0.6792,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.931519699812383e-06,
+      "loss": 0.6346,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9901500938086305e-06,
+      "loss": 0.5861,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0487804878048782e-06,
+      "loss": 0.6147,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1074108818011255e-06,
+      "loss": 0.5976,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1660412757973733e-06,
+      "loss": 0.6033,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.224671669793621e-06,
+      "loss": 0.6113,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.283302063789869e-06,
+      "loss": 0.5681,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3419324577861165e-06,
+      "loss": 0.5717,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4005628517823642e-06,
+      "loss": 0.5843,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4591932457786115e-06,
+      "loss": 0.5899,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5178236397748593e-06,
+      "loss": 0.5319,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5764540337711074e-06,
+      "loss": 0.6013,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6350844277673548e-06,
+      "loss": 0.6063,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6937148217636025e-06,
+      "loss": 0.5724,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7523452157598502e-06,
+      "loss": 0.5849,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8109756097560976e-06,
+      "loss": 0.5275,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.869606003752346e-06,
+      "loss": 0.5455,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.928236397748593e-06,
+      "loss": 0.518,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.98686679174484e-06,
+      "loss": 0.5309,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0454971857410885e-06,
+      "loss": 0.5405,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.104127579737336e-06,
+      "loss": 0.4894,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.162757973733584e-06,
+      "loss": 0.5709,
+      "step": 355
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.221388367729831e-06,
+      "loss": 0.5289,
+      "step": 360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.280018761726079e-06,
+      "loss": 0.5217,
+      "step": 365
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.338649155722327e-06,
+      "loss": 0.5165,
+      "step": 370
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.397279549718574e-06,
+      "loss": 0.4879,
+      "step": 375
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455909943714822e-06,
+      "loss": 0.5202,
+      "step": 380
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.51454033771107e-06,
+      "loss": 0.4872,
+      "step": 385
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.573170731707317e-06,
+      "loss": 0.493,
+      "step": 390
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.631801125703565e-06,
+      "loss": 0.4825,
+      "step": 395
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.690431519699812e-06,
+      "loss": 0.456,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7490619136960605e-06,
+      "loss": 0.4817,
+      "step": 405
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.807692307692308e-06,
+      "loss": 0.4762,
+      "step": 410
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.866322701688555e-06,
+      "loss": 0.4508,
+      "step": 415
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.924953095684803e-06,
+      "loss": 0.4522,
+      "step": 420
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.983583489681051e-06,
+      "loss": 0.4894,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.042213883677299e-06,
+      "loss": 0.4979,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.100844277673546e-06,
+      "loss": 0.4948,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1594746716697934e-06,
+      "loss": 0.4799,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.218105065666042e-06,
+      "loss": 0.4551,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.27673545966229e-06,
+      "loss": 0.4232,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.335365853658537e-06,
+      "loss": 0.4446,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.393996247654784e-06,
+      "loss": 0.4542,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.452626641651032e-06,
+      "loss": 0.4734,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.51125703564728e-06,
+      "loss": 0.4564,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.569887429643528e-06,
+      "loss": 0.4991,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.628517823639775e-06,
+      "loss": 0.4756,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.687148217636023e-06,
+      "loss": 0.474,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.74577861163227e-06,
+      "loss": 0.4501,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.804409005628518e-06,
+      "loss": 0.4726,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.863039399624766e-06,
+      "loss": 0.4845,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 0.45026421546936035,
+      "eval_runtime": 10.5589,
+      "eval_samples_per_second": 9.471,
+      "eval_steps_per_second": 9.471,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.921669793621013e-06,
+      "loss": 0.4809,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.980300187617261e-06,
+      "loss": 0.4719,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.038930581613509e-06,
+      "loss": 0.4421,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 0.4295,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.156191369606005e-06,
+      "loss": 0.4448,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.214821763602251e-06,
+      "loss": 0.4497,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2734521575985e-06,
+      "loss": 0.4438,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3320825515947465e-06,
+      "loss": 0.4549,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.390712945590995e-06,
+      "loss": 0.4767,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.449343339587242e-06,
+      "loss": 0.4291,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.50797373358349e-06,
+      "loss": 0.4421,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.566604127579738e-06,
+      "loss": 0.4297,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.625234521575985e-06,
+      "loss": 0.4256,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.683864915572233e-06,
+      "loss": 0.4481,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.74249530956848e-06,
+      "loss": 0.4512,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.8011257035647285e-06,
+      "loss": 0.4377,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.859756097560977e-06,
+      "loss": 0.4654,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.918386491557223e-06,
+      "loss": 0.4026,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.977016885553471e-06,
+      "loss": 0.4547,
+      "step": 595
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.0356472795497186e-06,
+      "loss": 0.4349,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.094277673545967e-06,
+      "loss": 0.4776,
+      "step": 605
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.152908067542215e-06,
+      "loss": 0.4121,
+      "step": 610
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.211538461538461e-06,
+      "loss": 0.4229,
+      "step": 615
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.2701688555347095e-06,
+      "loss": 0.445,
+      "step": 620
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.328799249530957e-06,
+      "loss": 0.4528,
+      "step": 625
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.387429643527205e-06,
+      "loss": 0.4382,
+      "step": 630
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.446060037523453e-06,
+      "loss": 0.4237,
+      "step": 635
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5046904315197005e-06,
+      "loss": 0.4303,
+      "step": 640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.563320825515948e-06,
+      "loss": 0.4365,
+      "step": 645
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.621951219512195e-06,
+      "loss": 0.4397,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.680581613508442e-06,
+      "loss": 0.4157,
+      "step": 655
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.739212007504691e-06,
+      "loss": 0.4351,
+      "step": 660
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.797842401500939e-06,
+      "loss": 0.4226,
+      "step": 665
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.856472795497186e-06,
+      "loss": 0.4086,
+      "step": 670
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.915103189493433e-06,
+      "loss": 0.4167,
+      "step": 675
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.97373358348968e-06,
+      "loss": 0.436,
+      "step": 680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.03236397748593e-06,
+      "loss": 0.4404,
+      "step": 685
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.090994371482177e-06,
+      "loss": 0.4268,
+      "step": 690
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.149624765478424e-06,
+      "loss": 0.4325,
+      "step": 695
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.208255159474672e-06,
+      "loss": 0.41,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.266885553470919e-06,
+      "loss": 0.4027,
+      "step": 705
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.325515947467168e-06,
+      "loss": 0.4532,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.384146341463415e-06,
+      "loss": 0.4342,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.442776735459663e-06,
+      "loss": 0.4072,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.50140712945591e-06,
+      "loss": 0.3962,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.560037523452157e-06,
+      "loss": 0.4224,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.618667917448406e-06,
+      "loss": 0.443,
+      "step": 735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.677298311444654e-06,
+      "loss": 0.4262,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.735928705440901e-06,
+      "loss": 0.4276,
+      "step": 745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.794559099437148e-06,
+      "loss": 0.4333,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.853189493433396e-06,
+      "loss": 0.4199,
+      "step": 755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.911819887429645e-06,
+      "loss": 0.4505,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.970450281425892e-06,
+      "loss": 0.4286,
+      "step": 765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.02908067542214e-06,
+      "loss": 0.4126,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.087711069418386e-06,
+      "loss": 0.3909,
+      "step": 775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.146341463414634e-06,
+      "loss": 0.4132,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.204971857410883e-06,
+      "loss": 0.3725,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.26360225140713e-06,
+      "loss": 0.3996,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.322232645403377e-06,
+      "loss": 0.434,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.380863039399625e-06,
+      "loss": 0.432,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.439493433395872e-06,
+      "loss": 0.407,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.498123827392121e-06,
+      "loss": 0.3913,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.556754221388368e-06,
+      "loss": 0.4433,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 0.4153,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.674015009380863e-06,
+      "loss": 0.4182,
+      "step": 825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.73264540337711e-06,
+      "loss": 0.4045,
+      "step": 830
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.79127579737336e-06,
+      "loss": 0.4041,
+      "step": 835
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.849906191369607e-06,
+      "loss": 0.3941,
+      "step": 840
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.908536585365854e-06,
+      "loss": 0.3963,
+      "step": 845
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.967166979362101e-06,
+      "loss": 0.432,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0025797373358349e-05,
+      "loss": 0.4181,
+      "step": 855
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0084427767354598e-05,
+      "loss": 0.4433,
+      "step": 860
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0143058161350845e-05,
+      "loss": 0.401,
+      "step": 865
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0201688555347092e-05,
+      "loss": 0.3956,
+      "step": 870
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0260318949343341e-05,
+      "loss": 0.4052,
+      "step": 875
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0318949343339587e-05,
+      "loss": 0.4146,
+      "step": 880
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0377579737335836e-05,
+      "loss": 0.4049,
+      "step": 885
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0436210131332083e-05,
+      "loss": 0.4177,
+      "step": 890
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.049484052532833e-05,
+      "loss": 0.4537,
+      "step": 895
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.055347091932458e-05,
+      "loss": 0.4175,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0612101313320825e-05,
+      "loss": 0.3695,
+      "step": 905
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0670731707317074e-05,
+      "loss": 0.4037,
+      "step": 910
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0729362101313321e-05,
+      "loss": 0.4141,
+      "step": 915
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0787992495309569e-05,
+      "loss": 0.4023,
+      "step": 920
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0846622889305818e-05,
+      "loss": 0.3647,
+      "step": 925
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0905253283302063e-05,
+      "loss": 0.4172,
+      "step": 930
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0963883677298312e-05,
+      "loss": 0.3895,
+      "step": 935
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.102251407129456e-05,
+      "loss": 0.3876,
+      "step": 940
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1081144465290807e-05,
+      "loss": 0.3759,
+      "step": 945
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1139774859287056e-05,
+      "loss": 0.4043,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1198405253283302e-05,
+      "loss": 0.3948,
+      "step": 955
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.125703564727955e-05,
+      "loss": 0.4055,
+      "step": 960
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1315666041275798e-05,
+      "loss": 0.4142,
+      "step": 965
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1374296435272045e-05,
+      "loss": 0.4328,
+      "step": 970
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1432926829268294e-05,
+      "loss": 0.4047,
+      "step": 975
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.149155722326454e-05,
+      "loss": 0.408,
+      "step": 980
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1550187617260787e-05,
+      "loss": 0.405,
+      "step": 985
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.1608818011257036e-05,
+      "loss": 0.4044,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1667448405253284e-05,
+      "loss": 0.4023,
+      "step": 995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1726078799249533e-05,
+      "loss": 0.3951,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.3985450863838196,
+      "eval_runtime": 10.5677,
+      "eval_samples_per_second": 9.463,
+      "eval_steps_per_second": 9.463,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.178470919324578e-05,
+      "loss": 0.3828,
+      "step": 1005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1843339587242026e-05,
+      "loss": 0.3952,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1901969981238275e-05,
+      "loss": 0.4074,
+      "step": 1015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.1960600375234522e-05,
+      "loss": 0.3446,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2019230769230771e-05,
+      "loss": 0.3972,
+      "step": 1025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2077861163227018e-05,
+      "loss": 0.3501,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2136491557223264e-05,
+      "loss": 0.4003,
+      "step": 1035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.3998,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.225375234521576e-05,
+      "loss": 0.3966,
+      "step": 1045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.231238273921201e-05,
+      "loss": 0.4103,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2371013133208257e-05,
+      "loss": 0.4126,
+      "step": 1055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2429643527204502e-05,
+      "loss": 0.4029,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2488273921200751e-05,
+      "loss": 0.4234,
+      "step": 1065
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2546904315197e-05,
+      "loss": 0.4048,
+      "step": 1070
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2605534709193246e-05,
+      "loss": 0.4306,
+      "step": 1075
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2664165103189493e-05,
+      "loss": 0.4036,
+      "step": 1080
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.272279549718574e-05,
+      "loss": 0.3782,
+      "step": 1085
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.278142589118199e-05,
+      "loss": 0.408,
+      "step": 1090
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2840056285178237e-05,
+      "loss": 0.4104,
+      "step": 1095
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2898686679174484e-05,
+      "loss": 0.3797,
+      "step": 1100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2957317073170733e-05,
+      "loss": 0.4013,
+      "step": 1105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.301594746716698e-05,
+      "loss": 0.401,
+      "step": 1110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.307457786116323e-05,
+      "loss": 0.4132,
+      "step": 1115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3133208255159477e-05,
+      "loss": 0.3838,
+      "step": 1120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3191838649155722e-05,
+      "loss": 0.4303,
+      "step": 1125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.325046904315197e-05,
+      "loss": 0.4177,
+      "step": 1130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3309099437148217e-05,
+      "loss": 0.3646,
+      "step": 1135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3367729831144466e-05,
+      "loss": 0.3996,
+      "step": 1140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3426360225140713e-05,
+      "loss": 0.3913,
+      "step": 1145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.348499061913696e-05,
+      "loss": 0.3774,
+      "step": 1150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.354362101313321e-05,
+      "loss": 0.4188,
+      "step": 1155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3602251407129457e-05,
+      "loss": 0.3658,
+      "step": 1160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3660881801125704e-05,
+      "loss": 0.387,
+      "step": 1165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3719512195121953e-05,
+      "loss": 0.3991,
+      "step": 1170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.37781425891182e-05,
+      "loss": 0.3824,
+      "step": 1175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3836772983114446e-05,
+      "loss": 0.4034,
+      "step": 1180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3895403377110693e-05,
+      "loss": 0.3582,
+      "step": 1185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.3954033771106943e-05,
+      "loss": 0.3816,
+      "step": 1190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.401266416510319e-05,
+      "loss": 0.38,
+      "step": 1195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4071294559099437e-05,
+      "loss": 0.3887,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4129924953095686e-05,
+      "loss": 0.4033,
+      "step": 1205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4188555347091933e-05,
+      "loss": 0.3784,
+      "step": 1210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.424718574108818e-05,
+      "loss": 0.3889,
+      "step": 1215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.430581613508443e-05,
+      "loss": 0.3671,
+      "step": 1220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4364446529080677e-05,
+      "loss": 0.3575,
+      "step": 1225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4423076923076923e-05,
+      "loss": 0.3871,
+      "step": 1230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.448170731707317e-05,
+      "loss": 0.4034,
+      "step": 1235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4540337711069419e-05,
+      "loss": 0.396,
+      "step": 1240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4598968105065666e-05,
+      "loss": 0.3741,
+      "step": 1245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4657598499061914e-05,
+      "loss": 0.3861,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4716228893058163e-05,
+      "loss": 0.4007,
+      "step": 1255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.477485928705441e-05,
+      "loss": 0.367,
+      "step": 1260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4833489681050657e-05,
+      "loss": 0.355,
+      "step": 1265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4892120075046906e-05,
+      "loss": 0.4029,
+      "step": 1270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.4950750469043154e-05,
+      "loss": 0.3722,
+      "step": 1275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5009380863039401e-05,
+      "loss": 0.3901,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5068011257035647e-05,
+      "loss": 0.4027,
+      "step": 1285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5126641651031896e-05,
+      "loss": 0.3618,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5185272045028143e-05,
+      "loss": 0.3801,
+      "step": 1295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.524390243902439e-05,
+      "loss": 0.4104,
+      "step": 1300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.530253283302064e-05,
+      "loss": 0.4028,
+      "step": 1305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5361163227016885e-05,
+      "loss": 0.3972,
+      "step": 1310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5419793621013134e-05,
+      "loss": 0.4118,
+      "step": 1315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5478424015009383e-05,
+      "loss": 0.3466,
+      "step": 1320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5537054409005632e-05,
+      "loss": 0.4084,
+      "step": 1325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5595684803001878e-05,
+      "loss": 0.3734,
+      "step": 1330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5654315196998123e-05,
+      "loss": 0.4115,
+      "step": 1335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5712945590994372e-05,
+      "loss": 0.382,
+      "step": 1340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5771575984990618e-05,
+      "loss": 0.361,
+      "step": 1345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5830206378986867e-05,
+      "loss": 0.413,
+      "step": 1350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.5888836772983116e-05,
+      "loss": 0.3922,
+      "step": 1355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.594746716697936e-05,
+      "loss": 0.3477,
+      "step": 1360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.600609756097561e-05,
+      "loss": 0.386,
+      "step": 1365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.606472795497186e-05,
+      "loss": 0.3448,
+      "step": 1370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6123358348968105e-05,
+      "loss": 0.3657,
+      "step": 1375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6181988742964354e-05,
+      "loss": 0.3743,
+      "step": 1380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.62406191369606e-05,
+      "loss": 0.392,
+      "step": 1385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.629924953095685e-05,
+      "loss": 0.3744,
+      "step": 1390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6357879924953094e-05,
+      "loss": 0.383,
+      "step": 1395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6416510318949343e-05,
+      "loss": 0.3739,
+      "step": 1400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6475140712945592e-05,
+      "loss": 0.3644,
+      "step": 1405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6533771106941838e-05,
+      "loss": 0.3429,
+      "step": 1410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6592401500938087e-05,
+      "loss": 0.3876,
+      "step": 1415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6651031894934336e-05,
+      "loss": 0.3877,
+      "step": 1420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.670966228893058e-05,
+      "loss": 0.4003,
+      "step": 1425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.676829268292683e-05,
+      "loss": 0.3824,
+      "step": 1430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.682692307692308e-05,
+      "loss": 0.375,
+      "step": 1435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.6885553470919325e-05,
+      "loss": 0.3732,
+      "step": 1440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.694418386491557e-05,
+      "loss": 0.3874,
+      "step": 1445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.700281425891182e-05,
+      "loss": 0.3687,
+      "step": 1450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.706144465290807e-05,
+      "loss": 0.3991,
+      "step": 1455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7120075046904315e-05,
+      "loss": 0.3493,
+      "step": 1460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7178705440900564e-05,
+      "loss": 0.3338,
+      "step": 1465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7237335834896813e-05,
+      "loss": 0.3733,
+      "step": 1470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7295966228893058e-05,
+      "loss": 0.3407,
+      "step": 1475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7354596622889307e-05,
+      "loss": 0.3743,
+      "step": 1480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7413227016885556e-05,
+      "loss": 0.3911,
+      "step": 1485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7471857410881802e-05,
+      "loss": 0.3591,
+      "step": 1490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7530487804878047e-05,
+      "loss": 0.3793,
+      "step": 1495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7589118198874296e-05,
+      "loss": 0.3851,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.3775494694709778,
+      "eval_runtime": 10.5367,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 1500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7647748592870545e-05,
+      "loss": 0.3634,
+      "step": 1505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.770637898686679e-05,
+      "loss": 0.4114,
+      "step": 1510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.776500938086304e-05,
+      "loss": 0.378,
+      "step": 1515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.782363977485929e-05,
+      "loss": 0.3722,
+      "step": 1520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7882270168855535e-05,
+      "loss": 0.3827,
+      "step": 1525
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7940900562851784e-05,
+      "loss": 0.383,
+      "step": 1530
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.7999530956848033e-05,
+      "loss": 0.3992,
+      "step": 1535
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.805816135084428e-05,
+      "loss": 0.3733,
+      "step": 1540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8116791744840524e-05,
+      "loss": 0.3744,
+      "step": 1545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8175422138836773e-05,
+      "loss": 0.3748,
+      "step": 1550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8234052532833022e-05,
+      "loss": 0.3561,
+      "step": 1555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.8292682926829268e-05,
+      "loss": 0.3616,
+      "step": 1560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8351313320825517e-05,
+      "loss": 0.399,
+      "step": 1565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8409943714821766e-05,
+      "loss": 0.375,
+      "step": 1570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.846857410881801e-05,
+      "loss": 0.3917,
+      "step": 1575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.852720450281426e-05,
+      "loss": 0.3619,
+      "step": 1580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.858583489681051e-05,
+      "loss": 0.3954,
+      "step": 1585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8644465290806755e-05,
+      "loss": 0.3946,
+      "step": 1590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8703095684803e-05,
+      "loss": 0.3951,
+      "step": 1595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.876172607879925e-05,
+      "loss": 0.3481,
+      "step": 1600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.88203564727955e-05,
+      "loss": 0.3811,
+      "step": 1605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8878986866791744e-05,
+      "loss": 0.3951,
+      "step": 1610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8937617260787993e-05,
+      "loss": 0.3794,
+      "step": 1615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.8996247654784242e-05,
+      "loss": 0.4044,
+      "step": 1620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9054878048780488e-05,
+      "loss": 0.3489,
+      "step": 1625
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9113508442776737e-05,
+      "loss": 0.3395,
+      "step": 1630
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9172138836772986e-05,
+      "loss": 0.4209,
+      "step": 1635
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 0.3852,
+      "step": 1640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.928939962476548e-05,
+      "loss": 0.3539,
+      "step": 1645
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9348030018761726e-05,
+      "loss": 0.4027,
+      "step": 1650
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9406660412757975e-05,
+      "loss": 0.369,
+      "step": 1655
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.946529080675422e-05,
+      "loss": 0.3624,
+      "step": 1660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.952392120075047e-05,
+      "loss": 0.365,
+      "step": 1665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.958255159474672e-05,
+      "loss": 0.3642,
+      "step": 1670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9641181988742964e-05,
+      "loss": 0.3889,
+      "step": 1675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9699812382739213e-05,
+      "loss": 0.3828,
+      "step": 1680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9758442776735462e-05,
+      "loss": 0.3699,
+      "step": 1685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9817073170731708e-05,
+      "loss": 0.3726,
+      "step": 1690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9875703564727957e-05,
+      "loss": 0.3941,
+      "step": 1695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9934333958724203e-05,
+      "loss": 0.3476,
+      "step": 1700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.999296435272045e-05,
+      "loss": 0.3631,
+      "step": 1705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0051594746716697e-05,
+      "loss": 0.3619,
+      "step": 1710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0110225140712946e-05,
+      "loss": 0.3699,
+      "step": 1715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0168855534709195e-05,
+      "loss": 0.3832,
+      "step": 1720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.022748592870544e-05,
+      "loss": 0.3747,
+      "step": 1725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.028611632270169e-05,
+      "loss": 0.3727,
+      "step": 1730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.034474671669794e-05,
+      "loss": 0.3725,
+      "step": 1735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0403377110694185e-05,
+      "loss": 0.3771,
+      "step": 1740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0462007504690434e-05,
+      "loss": 0.3689,
+      "step": 1745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0520637898686683e-05,
+      "loss": 0.3746,
+      "step": 1750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0579268292682928e-05,
+      "loss": 0.3559,
+      "step": 1755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0637898686679174e-05,
+      "loss": 0.3715,
+      "step": 1760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0696529080675423e-05,
+      "loss": 0.3851,
+      "step": 1765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0755159474671672e-05,
+      "loss": 0.3851,
+      "step": 1770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0813789868667917e-05,
+      "loss": 0.3639,
+      "step": 1775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0872420262664166e-05,
+      "loss": 0.3669,
+      "step": 1780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.0931050656660415e-05,
+      "loss": 0.3942,
+      "step": 1785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.098968105065666e-05,
+      "loss": 0.3495,
+      "step": 1790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.104831144465291e-05,
+      "loss": 0.3516,
+      "step": 1795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.110694183864916e-05,
+      "loss": 0.3966,
+      "step": 1800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1165572232645405e-05,
+      "loss": 0.3839,
+      "step": 1805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.122420262664165e-05,
+      "loss": 0.3659,
+      "step": 1810
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.12828330206379e-05,
+      "loss": 0.3885,
+      "step": 1815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.134146341463415e-05,
+      "loss": 0.3648,
+      "step": 1820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1400093808630394e-05,
+      "loss": 0.38,
+      "step": 1825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1458724202626643e-05,
+      "loss": 0.3699,
+      "step": 1830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1517354596622892e-05,
+      "loss": 0.3835,
+      "step": 1835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1575984990619138e-05,
+      "loss": 0.3604,
+      "step": 1840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1634615384615387e-05,
+      "loss": 0.3876,
+      "step": 1845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1693245778611636e-05,
+      "loss": 0.3233,
+      "step": 1850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.175187617260788e-05,
+      "loss": 0.3619,
+      "step": 1855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1810506566604127e-05,
+      "loss": 0.3788,
+      "step": 1860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1869136960600376e-05,
+      "loss": 0.3812,
+      "step": 1865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927767354596625e-05,
+      "loss": 0.3977,
+      "step": 1870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.198639774859287e-05,
+      "loss": 0.3496,
+      "step": 1875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.204502814258912e-05,
+      "loss": 0.3678,
+      "step": 1880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.210365853658537e-05,
+      "loss": 0.3521,
+      "step": 1885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2162288930581614e-05,
+      "loss": 0.3765,
+      "step": 1890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2220919324577863e-05,
+      "loss": 0.3787,
+      "step": 1895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2279549718574112e-05,
+      "loss": 0.3749,
+      "step": 1900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2338180112570358e-05,
+      "loss": 0.3536,
+      "step": 1905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2396810506566603e-05,
+      "loss": 0.367,
+      "step": 1910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2455440900562852e-05,
+      "loss": 0.3746,
+      "step": 1915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.25140712945591e-05,
+      "loss": 0.3574,
+      "step": 1920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2572701688555347e-05,
+      "loss": 0.3853,
+      "step": 1925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2631332082551596e-05,
+      "loss": 0.3839,
+      "step": 1930
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2689962476547845e-05,
+      "loss": 0.3809,
+      "step": 1935
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.274859287054409e-05,
+      "loss": 0.3989,
+      "step": 1940
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.280722326454034e-05,
+      "loss": 0.3575,
+      "step": 1945
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.286585365853659e-05,
+      "loss": 0.3877,
+      "step": 1950
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.2924484052532834e-05,
+      "loss": 0.3573,
+      "step": 1955
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.298311444652908e-05,
+      "loss": 0.3377,
+      "step": 1960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.304174484052533e-05,
+      "loss": 0.3755,
+      "step": 1965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3100375234521575e-05,
+      "loss": 0.373,
+      "step": 1970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3159005628517824e-05,
+      "loss": 0.3544,
+      "step": 1975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3217636022514073e-05,
+      "loss": 0.3509,
+      "step": 1980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.327626641651032e-05,
+      "loss": 0.3458,
+      "step": 1985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3334896810506567e-05,
+      "loss": 0.3339,
+      "step": 1990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3393527204502816e-05,
+      "loss": 0.3466,
+      "step": 1995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3452157598499065e-05,
+      "loss": 0.3545,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 0.3645670711994171,
+      "eval_runtime": 10.5377,
+      "eval_samples_per_second": 9.49,
+      "eval_steps_per_second": 9.49,
+      "step": 2000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.351078799249531e-05,
+      "loss": 0.3561,
+      "step": 2005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.356941838649156e-05,
+      "loss": 0.3915,
+      "step": 2010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3628048780487806e-05,
+      "loss": 0.3427,
+      "step": 2015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.368667917448405e-05,
+      "loss": 0.3641,
+      "step": 2020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.37453095684803e-05,
+      "loss": 0.3611,
+      "step": 2025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.380393996247655e-05,
+      "loss": 0.3538,
+      "step": 2030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3862570356472795e-05,
+      "loss": 0.3584,
+      "step": 2035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3921200750469044e-05,
+      "loss": 0.3742,
+      "step": 2040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.3979831144465293e-05,
+      "loss": 0.3686,
+      "step": 2045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4038461538461542e-05,
+      "loss": 0.4034,
+      "step": 2050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4097091932457787e-05,
+      "loss": 0.3634,
+      "step": 2055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4155722326454036e-05,
+      "loss": 0.3832,
+      "step": 2060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4214352720450282e-05,
+      "loss": 0.364,
+      "step": 2065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4272983114446528e-05,
+      "loss": 0.3563,
+      "step": 2070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4331613508442777e-05,
+      "loss": 0.3542,
+      "step": 2075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.382,
+      "step": 2080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.444887429643527e-05,
+      "loss": 0.3793,
+      "step": 2085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.450750469043152e-05,
+      "loss": 0.3624,
+      "step": 2090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.456613508442777e-05,
+      "loss": 0.3771,
+      "step": 2095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.462476547842402e-05,
+      "loss": 0.3546,
+      "step": 2100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4683395872420264e-05,
+      "loss": 0.3705,
+      "step": 2105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4742026266416513e-05,
+      "loss": 0.3835,
+      "step": 2110
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4800656660412762e-05,
+      "loss": 0.3706,
+      "step": 2115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4859287054409004e-05,
+      "loss": 0.3756,
+      "step": 2120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4917917448405253e-05,
+      "loss": 0.3632,
+      "step": 2125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.4976547842401502e-05,
+      "loss": 0.3488,
+      "step": 2130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5035178236397748e-05,
+      "loss": 0.3553,
+      "step": 2135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5093808630394e-05,
+      "loss": 0.3656,
+      "step": 2140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5152439024390246e-05,
+      "loss": 0.3598,
+      "step": 2145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.521106941838649e-05,
+      "loss": 0.3478,
+      "step": 2150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.526969981238274e-05,
+      "loss": 0.3963,
+      "step": 2155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5328330206378986e-05,
+      "loss": 0.3649,
+      "step": 2160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.538696060037524e-05,
+      "loss": 0.3606,
+      "step": 2165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.544559099437148e-05,
+      "loss": 0.383,
+      "step": 2170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5504221388367733e-05,
+      "loss": 0.3918,
+      "step": 2175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.556285178236398e-05,
+      "loss": 0.3658,
+      "step": 2180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5621482176360228e-05,
+      "loss": 0.3517,
+      "step": 2185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5680112570356473e-05,
+      "loss": 0.3743,
+      "step": 2190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.573874296435272e-05,
+      "loss": 0.3722,
+      "step": 2195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5797373358348968e-05,
+      "loss": 0.4099,
+      "step": 2200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5856003752345214e-05,
+      "loss": 0.3321,
+      "step": 2205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5914634146341466e-05,
+      "loss": 0.3664,
+      "step": 2210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.5973264540337712e-05,
+      "loss": 0.4203,
+      "step": 2215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.603189493433396e-05,
+      "loss": 0.3258,
+      "step": 2220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6090525328330206e-05,
+      "loss": 0.368,
+      "step": 2225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.614915572232646e-05,
+      "loss": 0.3918,
+      "step": 2230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.62077861163227e-05,
+      "loss": 0.348,
+      "step": 2235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6266416510318953e-05,
+      "loss": 0.3674,
+      "step": 2240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.63250469043152e-05,
+      "loss": 0.3862,
+      "step": 2245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6383677298311445e-05,
+      "loss": 0.3807,
+      "step": 2250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6442307692307694e-05,
+      "loss": 0.3581,
+      "step": 2255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.650093808630394e-05,
+      "loss": 0.3731,
+      "step": 2260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6559568480300188e-05,
+      "loss": 0.399,
+      "step": 2265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6618198874296434e-05,
+      "loss": 0.3581,
+      "step": 2270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6676829268292686e-05,
+      "loss": 0.3314,
+      "step": 2275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6735459662288932e-05,
+      "loss": 0.3697,
+      "step": 2280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.679409005628518e-05,
+      "loss": 0.3752,
+      "step": 2285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.6852720450281427e-05,
+      "loss": 0.3518,
+      "step": 2290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.691135084427768e-05,
+      "loss": 0.3563,
+      "step": 2295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.696998123827392e-05,
+      "loss": 0.3996,
+      "step": 2300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7028611632270167e-05,
+      "loss": 0.3122,
+      "step": 2305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.708724202626642e-05,
+      "loss": 0.353,
+      "step": 2310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7145872420262665e-05,
+      "loss": 0.3473,
+      "step": 2315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7204502814258914e-05,
+      "loss": 0.3672,
+      "step": 2320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.726313320825516e-05,
+      "loss": 0.3672,
+      "step": 2325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.732176360225141e-05,
+      "loss": 0.3638,
+      "step": 2330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7380393996247654e-05,
+      "loss": 0.3466,
+      "step": 2335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7439024390243906e-05,
+      "loss": 0.358,
+      "step": 2340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7497654784240152e-05,
+      "loss": 0.3614,
+      "step": 2345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.75562851782364e-05,
+      "loss": 0.3242,
+      "step": 2350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7614915572232647e-05,
+      "loss": 0.3651,
+      "step": 2355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7673545966228892e-05,
+      "loss": 0.3657,
+      "step": 2360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.773217636022514e-05,
+      "loss": 0.3347,
+      "step": 2365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7790806754221387e-05,
+      "loss": 0.3429,
+      "step": 2370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.784943714821764e-05,
+      "loss": 0.3619,
+      "step": 2375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7908067542213885e-05,
+      "loss": 0.3741,
+      "step": 2380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7966697936210134e-05,
+      "loss": 0.321,
+      "step": 2385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.802532833020638e-05,
+      "loss": 0.3701,
+      "step": 2390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.808395872420263e-05,
+      "loss": 0.3606,
+      "step": 2395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8142589118198874e-05,
+      "loss": 0.3868,
+      "step": 2400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.820121951219512e-05,
+      "loss": 0.324,
+      "step": 2405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8259849906191372e-05,
+      "loss": 0.3593,
+      "step": 2410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8318480300187618e-05,
+      "loss": 0.3776,
+      "step": 2415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8377110694183867e-05,
+      "loss": 0.3843,
+      "step": 2420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8435741088180113e-05,
+      "loss": 0.3398,
+      "step": 2425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.849437148217636e-05,
+      "loss": 0.3671,
+      "step": 2430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8553001876172607e-05,
+      "loss": 0.353,
+      "step": 2435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.861163227016886e-05,
+      "loss": 0.3404,
+      "step": 2440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8670262664165105e-05,
+      "loss": 0.3265,
+      "step": 2445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8728893058161354e-05,
+      "loss": 0.3741,
+      "step": 2450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.87875234521576e-05,
+      "loss": 0.3315,
+      "step": 2455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8846153846153845e-05,
+      "loss": 0.3615,
+      "step": 2460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8904784240150094e-05,
+      "loss": 0.3375,
+      "step": 2465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.896341463414634e-05,
+      "loss": 0.3464,
+      "step": 2470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9022045028142592e-05,
+      "loss": 0.405,
+      "step": 2475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9080675422138838e-05,
+      "loss": 0.3392,
+      "step": 2480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9139305816135087e-05,
+      "loss": 0.3461,
+      "step": 2485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9197936210131333e-05,
+      "loss": 0.348,
+      "step": 2490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9256566604127582e-05,
+      "loss": 0.368,
+      "step": 2495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9315196998123827e-05,
+      "loss": 0.367,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 0.3577040433883667,
+      "eval_runtime": 10.5307,
+      "eval_samples_per_second": 9.496,
+      "eval_steps_per_second": 9.496,
+      "step": 2500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.937382739212008e-05,
+      "loss": 0.3809,
+      "step": 2505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9432457786116325e-05,
+      "loss": 0.3426,
+      "step": 2510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.949108818011257e-05,
+      "loss": 0.3887,
+      "step": 2515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.954971857410882e-05,
+      "loss": 0.3503,
+      "step": 2520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9608348968105066e-05,
+      "loss": 0.3561,
+      "step": 2525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9666979362101315e-05,
+      "loss": 0.3581,
+      "step": 2530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.972560975609756e-05,
+      "loss": 0.3724,
+      "step": 2535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9784240150093813e-05,
+      "loss": 0.378,
+      "step": 2540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9842870544090058e-05,
+      "loss": 0.3481,
+      "step": 2545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9901500938086307e-05,
+      "loss": 0.3424,
+      "step": 2550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9960131332082553e-05,
+      "loss": 0.3445,
+      "step": 2555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0018761726078802e-05,
+      "loss": 0.3466,
+      "step": 2560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0077392120075048e-05,
+      "loss": 0.3691,
+      "step": 2565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0136022514071293e-05,
+      "loss": 0.3692,
+      "step": 2570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0194652908067546e-05,
+      "loss": 0.3332,
+      "step": 2575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.025328330206379e-05,
+      "loss": 0.3502,
+      "step": 2580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.031191369606004e-05,
+      "loss": 0.3345,
+      "step": 2585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0370544090056286e-05,
+      "loss": 0.3595,
+      "step": 2590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0429174484052535e-05,
+      "loss": 0.3554,
+      "step": 2595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.048780487804878e-05,
+      "loss": 0.3632,
+      "step": 2600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.054643527204503e-05,
+      "loss": 0.3838,
+      "step": 2605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.060506566604128e-05,
+      "loss": 0.3539,
+      "step": 2610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0663696060037524e-05,
+      "loss": 0.3716,
+      "step": 2615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.072232645403377e-05,
+      "loss": 0.3561,
+      "step": 2620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0780956848030015e-05,
+      "loss": 0.3504,
+      "step": 2625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.083958724202627e-05,
+      "loss": 0.3515,
+      "step": 2630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.089821763602251e-05,
+      "loss": 0.3483,
+      "step": 2635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.0956848030018766e-05,
+      "loss": 0.3907,
+      "step": 2640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.101547842401501e-05,
+      "loss": 0.3511,
+      "step": 2645
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1074108818011264e-05,
+      "loss": 0.3438,
+      "step": 2650
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.11327392120075e-05,
+      "loss": 0.3856,
+      "step": 2655
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1191369606003755e-05,
+      "loss": 0.33,
+      "step": 2660
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.125e-05,
+      "loss": 0.373,
+      "step": 2665
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1308630393996246e-05,
+      "loss": 0.3941,
+      "step": 2670
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.13672607879925e-05,
+      "loss": 0.3685,
+      "step": 2675
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1425891181988744e-05,
+      "loss": 0.3595,
+      "step": 2680
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.148452157598499e-05,
+      "loss": 0.3686,
+      "step": 2685
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1543151969981236e-05,
+      "loss": 0.3551,
+      "step": 2690
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.160178236397749e-05,
+      "loss": 0.357,
+      "step": 2695
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.1660412757973734e-05,
+      "loss": 0.3788,
+      "step": 2700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1719043151969986e-05,
+      "loss": 0.346,
+      "step": 2705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.177767354596623e-05,
+      "loss": 0.3612,
+      "step": 2710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.1836303939962484e-05,
+      "loss": 0.3425,
+      "step": 2715
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.189493433395872e-05,
+      "loss": 0.3502,
+      "step": 2720
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.195356472795497e-05,
+      "loss": 0.3617,
+      "step": 2725
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.201219512195122e-05,
+      "loss": 0.3544,
+      "step": 2730
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2070825515947466e-05,
+      "loss": 0.3392,
+      "step": 2735
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.212945590994372e-05,
+      "loss": 0.3688,
+      "step": 2740
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2188086303939964e-05,
+      "loss": 0.3643,
+      "step": 2745
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.224671669793621e-05,
+      "loss": 0.3766,
+      "step": 2750
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2305347091932456e-05,
+      "loss": 0.3322,
+      "step": 2755
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.236397748592871e-05,
+      "loss": 0.3508,
+      "step": 2760
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2422607879924954e-05,
+      "loss": 0.3747,
+      "step": 2765
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.24812382739212e-05,
+      "loss": 0.3758,
+      "step": 2770
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.253986866791745e-05,
+      "loss": 0.3588,
+      "step": 2775
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.25984990619137e-05,
+      "loss": 0.3316,
+      "step": 2780
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.265712945590994e-05,
+      "loss": 0.3572,
+      "step": 2785
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.271575984990619e-05,
+      "loss": 0.3326,
+      "step": 2790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.277439024390244e-05,
+      "loss": 0.3435,
+      "step": 2795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.283302063789869e-05,
+      "loss": 0.3519,
+      "step": 2800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.289165103189494e-05,
+      "loss": 0.3572,
+      "step": 2805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.2950281425891185e-05,
+      "loss": 0.3398,
+      "step": 2810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.300891181988743e-05,
+      "loss": 0.3444,
+      "step": 2815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3067542213883676e-05,
+      "loss": 0.3467,
+      "step": 2820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.312617260787992e-05,
+      "loss": 0.3668,
+      "step": 2825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3184803001876174e-05,
+      "loss": 0.3473,
+      "step": 2830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.324343339587242e-05,
+      "loss": 0.3297,
+      "step": 2835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.330206378986867e-05,
+      "loss": 0.3245,
+      "step": 2840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.336069418386492e-05,
+      "loss": 0.3357,
+      "step": 2845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.341932457786116e-05,
+      "loss": 0.3226,
+      "step": 2850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.347795497185741e-05,
+      "loss": 0.3611,
+      "step": 2855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.353658536585366e-05,
+      "loss": 0.3491,
+      "step": 2860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.359521575984991e-05,
+      "loss": 0.3388,
+      "step": 2865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.365384615384616e-05,
+      "loss": 0.3568,
+      "step": 2870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3712476547842405e-05,
+      "loss": 0.3603,
+      "step": 2875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.377110694183865e-05,
+      "loss": 0.3639,
+      "step": 2880
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3829737335834896e-05,
+      "loss": 0.3595,
+      "step": 2885
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.388836772983114e-05,
+      "loss": 0.3682,
+      "step": 2890
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.3946998123827394e-05,
+      "loss": 0.3693,
+      "step": 2895
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.400562851782364e-05,
+      "loss": 0.3433,
+      "step": 2900
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.406425891181989e-05,
+      "loss": 0.3549,
+      "step": 2905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.412288930581614e-05,
+      "loss": 0.3357,
+      "step": 2910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.418151969981238e-05,
+      "loss": 0.3879,
+      "step": 2915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.424015009380863e-05,
+      "loss": 0.355,
+      "step": 2920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.429878048780488e-05,
+      "loss": 0.3463,
+      "step": 2925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.435741088180113e-05,
+      "loss": 0.3445,
+      "step": 2930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.441604127579737e-05,
+      "loss": 0.3564,
+      "step": 2935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4474671669793625e-05,
+      "loss": 0.3453,
+      "step": 2940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.453330206378987e-05,
+      "loss": 0.3659,
+      "step": 2945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4591932457786116e-05,
+      "loss": 0.3705,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.465056285178236e-05,
+      "loss": 0.3214,
+      "step": 2955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4709193245778614e-05,
+      "loss": 0.374,
+      "step": 2960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.476782363977486e-05,
+      "loss": 0.3363,
+      "step": 2965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.482645403377111e-05,
+      "loss": 0.3535,
+      "step": 2970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.488508442776736e-05,
+      "loss": 0.3567,
+      "step": 2975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.4943714821763604e-05,
+      "loss": 0.3518,
+      "step": 2980
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.500234521575985e-05,
+      "loss": 0.3385,
+      "step": 2985
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5060975609756095e-05,
+      "loss": 0.3581,
+      "step": 2990
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.511960600375235e-05,
+      "loss": 0.3654,
+      "step": 2995
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.517823639774859e-05,
+      "loss": 0.353,
+      "step": 3000
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.3522051274776459,
+      "eval_runtime": 10.5471,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 3000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5236866791744845e-05,
+      "loss": 0.3403,
+      "step": 3005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.529549718574109e-05,
+      "loss": 0.3401,
+      "step": 3010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5354127579737336e-05,
+      "loss": 0.3447,
+      "step": 3015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.541275797373358e-05,
+      "loss": 0.3644,
+      "step": 3020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5471388367729835e-05,
+      "loss": 0.3646,
+      "step": 3025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.553001876172608e-05,
+      "loss": 0.3469,
+      "step": 3030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5588649155722326e-05,
+      "loss": 0.3766,
+      "step": 3035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.564727954971858e-05,
+      "loss": 0.3477,
+      "step": 3040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5705909943714824e-05,
+      "loss": 0.349,
+      "step": 3045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.576454033771107e-05,
+      "loss": 0.3357,
+      "step": 3050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5823170731707315e-05,
+      "loss": 0.3592,
+      "step": 3055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.588180112570357e-05,
+      "loss": 0.3172,
+      "step": 3060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.594043151969981e-05,
+      "loss": 0.3552,
+      "step": 3065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.5999061913696065e-05,
+      "loss": 0.356,
+      "step": 3070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.605769230769231e-05,
+      "loss": 0.3345,
+      "step": 3075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.611632270168856e-05,
+      "loss": 0.3301,
+      "step": 3080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.61749530956848e-05,
+      "loss": 0.3499,
+      "step": 3085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.623358348968105e-05,
+      "loss": 0.3707,
+      "step": 3090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.62922138836773e-05,
+      "loss": 0.3458,
+      "step": 3095
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6350844277673546e-05,
+      "loss": 0.3501,
+      "step": 3100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.64094746716698e-05,
+      "loss": 0.3591,
+      "step": 3105
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6468105065666044e-05,
+      "loss": 0.3478,
+      "step": 3110
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.652673545966229e-05,
+      "loss": 0.3257,
+      "step": 3115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6585365853658535e-05,
+      "loss": 0.3472,
+      "step": 3120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.664399624765479e-05,
+      "loss": 0.3647,
+      "step": 3125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.670262664165103e-05,
+      "loss": 0.3712,
+      "step": 3130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.6761257035647286e-05,
+      "loss": 0.3607,
+      "step": 3135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.681988742964353e-05,
+      "loss": 0.3935,
+      "step": 3140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.687851782363978e-05,
+      "loss": 0.3593,
+      "step": 3145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.693714821763602e-05,
+      "loss": 0.3689,
+      "step": 3150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.699577861163227e-05,
+      "loss": 0.3612,
+      "step": 3155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.705440900562852e-05,
+      "loss": 0.3765,
+      "step": 3160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7113039399624766e-05,
+      "loss": 0.3804,
+      "step": 3165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.717166979362102e-05,
+      "loss": 0.3414,
+      "step": 3170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7230300187617264e-05,
+      "loss": 0.3391,
+      "step": 3175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.728893058161351e-05,
+      "loss": 0.39,
+      "step": 3180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7347560975609755e-05,
+      "loss": 0.3589,
+      "step": 3185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7406191369606e-05,
+      "loss": 0.3554,
+      "step": 3190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7464821763602253e-05,
+      "loss": 0.3238,
+      "step": 3195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.75234521575985e-05,
+      "loss": 0.3561,
+      "step": 3200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.758208255159475e-05,
+      "loss": 0.3526,
+      "step": 3205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7640712945591e-05,
+      "loss": 0.3791,
+      "step": 3210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.769934333958724e-05,
+      "loss": 0.3545,
+      "step": 3215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.775797373358349e-05,
+      "loss": 0.3523,
+      "step": 3220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.781660412757974e-05,
+      "loss": 0.3347,
+      "step": 3225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7875234521575986e-05,
+      "loss": 0.3463,
+      "step": 3230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.793386491557224e-05,
+      "loss": 0.3371,
+      "step": 3235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7992495309568484e-05,
+      "loss": 0.3451,
+      "step": 3240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.805112570356473e-05,
+      "loss": 0.3648,
+      "step": 3245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8109756097560976e-05,
+      "loss": 0.372,
+      "step": 3250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.816838649155722e-05,
+      "loss": 0.3444,
+      "step": 3255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.8227016885553474e-05,
+      "loss": 0.355,
+      "step": 3260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.828564727954972e-05,
+      "loss": 0.3656,
+      "step": 3265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.834427767354597e-05,
+      "loss": 0.3422,
+      "step": 3270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.840290806754222e-05,
+      "loss": 0.361,
+      "step": 3275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.3587,
+      "step": 3280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.852016885553471e-05,
+      "loss": 0.3556,
+      "step": 3285
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.857879924953096e-05,
+      "loss": 0.3273,
+      "step": 3290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8637429643527207e-05,
+      "loss": 0.339,
+      "step": 3295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.869606003752345e-05,
+      "loss": 0.3469,
+      "step": 3300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8754690431519705e-05,
+      "loss": 0.3426,
+      "step": 3305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.881332082551595e-05,
+      "loss": 0.3597,
+      "step": 3310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8871951219512196e-05,
+      "loss": 0.3435,
+      "step": 3315
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.893058161350844e-05,
+      "loss": 0.3399,
+      "step": 3320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.8989212007504694e-05,
+      "loss": 0.3429,
+      "step": 3325
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.904784240150094e-05,
+      "loss": 0.3387,
+      "step": 3330
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.910647279549719e-05,
+      "loss": 0.3409,
+      "step": 3335
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.916510318949344e-05,
+      "loss": 0.3602,
+      "step": 3340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.922373358348968e-05,
+      "loss": 0.3468,
+      "step": 3345
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.928236397748593e-05,
+      "loss": 0.3489,
+      "step": 3350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9340994371482174e-05,
+      "loss": 0.3501,
+      "step": 3355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.939962476547843e-05,
+      "loss": 0.3368,
+      "step": 3360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.945825515947467e-05,
+      "loss": 0.372,
+      "step": 3365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9516885553470925e-05,
+      "loss": 0.3321,
+      "step": 3370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.957551594746717e-05,
+      "loss": 0.3535,
+      "step": 3375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9634146341463416e-05,
+      "loss": 0.3559,
+      "step": 3380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.969277673545966e-05,
+      "loss": 0.3842,
+      "step": 3385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9751407129455914e-05,
+      "loss": 0.3575,
+      "step": 3390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.981003752345216e-05,
+      "loss": 0.3682,
+      "step": 3395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.9868667917448405e-05,
+      "loss": 0.3297,
+      "step": 3400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.992729831144466e-05,
+      "loss": 0.3697,
+      "step": 3405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.99859287054409e-05,
+      "loss": 0.3504,
+      "step": 3410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.004455909943715e-05,
+      "loss": 0.3731,
+      "step": 3415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0103189493433394e-05,
+      "loss": 0.3451,
+      "step": 3420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.016181988742965e-05,
+      "loss": 0.3504,
+      "step": 3425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.022045028142589e-05,
+      "loss": 0.3507,
+      "step": 3430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0279080675422145e-05,
+      "loss": 0.331,
+      "step": 3435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.033771106941839e-05,
+      "loss": 0.3374,
+      "step": 3440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0396341463414636e-05,
+      "loss": 0.3286,
+      "step": 3445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.045497185741088e-05,
+      "loss": 0.3469,
+      "step": 3450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.051360225140713e-05,
+      "loss": 0.3475,
+      "step": 3455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.057223264540338e-05,
+      "loss": 0.3546,
+      "step": 3460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0630863039399625e-05,
+      "loss": 0.3629,
+      "step": 3465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.068949343339588e-05,
+      "loss": 0.3342,
+      "step": 3470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0748123827392123e-05,
+      "loss": 0.3458,
+      "step": 3475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.080675422138837e-05,
+      "loss": 0.3519,
+      "step": 3480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.0865384615384615e-05,
+      "loss": 0.3356,
+      "step": 3485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.092401500938087e-05,
+      "loss": 0.3324,
+      "step": 3490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.098264540337711e-05,
+      "loss": 0.3544,
+      "step": 3495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1041275797373365e-05,
+      "loss": 0.3438,
+      "step": 3500
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.3493571877479553,
+      "eval_runtime": 10.5406,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 3500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.109990619136961e-05,
+      "loss": 0.3457,
+      "step": 3505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1158536585365856e-05,
+      "loss": 0.332,
+      "step": 3510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.12171669793621e-05,
+      "loss": 0.3388,
+      "step": 3515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.127579737335835e-05,
+      "loss": 0.3509,
+      "step": 3520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.13344277673546e-05,
+      "loss": 0.3456,
+      "step": 3525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1393058161350846e-05,
+      "loss": 0.335,
+      "step": 3530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.14516885553471e-05,
+      "loss": 0.3253,
+      "step": 3535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1510318949343344e-05,
+      "loss": 0.3525,
+      "step": 3540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.156894934333959e-05,
+      "loss": 0.3493,
+      "step": 3545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.1627579737335835e-05,
+      "loss": 0.3619,
+      "step": 3550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.168621013133208e-05,
+      "loss": 0.3241,
+      "step": 3555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.174484052532833e-05,
+      "loss": 0.3642,
+      "step": 3560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.180347091932458e-05,
+      "loss": 0.3621,
+      "step": 3565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.186210131332083e-05,
+      "loss": 0.3247,
+      "step": 3570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.1920731707317077e-05,
+      "loss": 0.3477,
+      "step": 3575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.197936210131332e-05,
+      "loss": 0.3352,
+      "step": 3580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.203799249530957e-05,
+      "loss": 0.3672,
+      "step": 3585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.209662288930582e-05,
+      "loss": 0.3714,
+      "step": 3590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2155253283302066e-05,
+      "loss": 0.3332,
+      "step": 3595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.221388367729832e-05,
+      "loss": 0.3222,
+      "step": 3600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2272514071294564e-05,
+      "loss": 0.3322,
+      "step": 3605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.233114446529081e-05,
+      "loss": 0.345,
+      "step": 3610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2389774859287055e-05,
+      "loss": 0.3589,
+      "step": 3615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.24484052532833e-05,
+      "loss": 0.3631,
+      "step": 3620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.250703564727955e-05,
+      "loss": 0.3355,
+      "step": 3625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.25656660412758e-05,
+      "loss": 0.3479,
+      "step": 3630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.262429643527205e-05,
+      "loss": 0.3446,
+      "step": 3635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.26829268292683e-05,
+      "loss": 0.3448,
+      "step": 3640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.274155722326454e-05,
+      "loss": 0.3515,
+      "step": 3645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.280018761726079e-05,
+      "loss": 0.3433,
+      "step": 3650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.285881801125704e-05,
+      "loss": 0.3223,
+      "step": 3655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.2917448405253286e-05,
+      "loss": 0.3672,
+      "step": 3660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.297607879924953e-05,
+      "loss": 0.3445,
+      "step": 3665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3034709193245784e-05,
+      "loss": 0.3612,
+      "step": 3670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.309333958724203e-05,
+      "loss": 0.3635,
+      "step": 3675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3151969981238275e-05,
+      "loss": 0.3712,
+      "step": 3680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.321060037523452e-05,
+      "loss": 0.3576,
+      "step": 3685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.326923076923077e-05,
+      "loss": 0.3451,
+      "step": 3690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.332786116322702e-05,
+      "loss": 0.337,
+      "step": 3695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.338649155722327e-05,
+      "loss": 0.3955,
+      "step": 3700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.344512195121952e-05,
+      "loss": 0.3417,
+      "step": 3705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.350375234521576e-05,
+      "loss": 0.353,
+      "step": 3710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.356238273921201e-05,
+      "loss": 0.3516,
+      "step": 3715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3621013133208254e-05,
+      "loss": 0.3788,
+      "step": 3720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3679643527204506e-05,
+      "loss": 0.3611,
+      "step": 3725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.373827392120075e-05,
+      "loss": 0.3479,
+      "step": 3730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3796904315197004e-05,
+      "loss": 0.3494,
+      "step": 3735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.385553470919325e-05,
+      "loss": 0.3495,
+      "step": 3740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.3914165103189495e-05,
+      "loss": 0.3221,
+      "step": 3745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.397279549718574e-05,
+      "loss": 0.3679,
+      "step": 3750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4031425891181993e-05,
+      "loss": 0.3789,
+      "step": 3755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.409005628517824e-05,
+      "loss": 0.3436,
+      "step": 3760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4148686679174485e-05,
+      "loss": 0.3464,
+      "step": 3765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.420731707317074e-05,
+      "loss": 0.3243,
+      "step": 3770
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.426594746716698e-05,
+      "loss": 0.3591,
+      "step": 3775
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.432457786116323e-05,
+      "loss": 0.3671,
+      "step": 3780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4383208255159474e-05,
+      "loss": 0.3528,
+      "step": 3785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4441838649155726e-05,
+      "loss": 0.3637,
+      "step": 3790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.450046904315197e-05,
+      "loss": 0.3185,
+      "step": 3795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4559099437148224e-05,
+      "loss": 0.3577,
+      "step": 3800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.461772983114447e-05,
+      "loss": 0.347,
+      "step": 3805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4676360225140716e-05,
+      "loss": 0.3868,
+      "step": 3810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.473499061913696e-05,
+      "loss": 0.3359,
+      "step": 3815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.479362101313321e-05,
+      "loss": 0.3245,
+      "step": 3820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.485225140712946e-05,
+      "loss": 0.3556,
+      "step": 3825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.4910881801125705e-05,
+      "loss": 0.3418,
+      "step": 3830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.496951219512196e-05,
+      "loss": 0.3455,
+      "step": 3835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.50281425891182e-05,
+      "loss": 0.3431,
+      "step": 3840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.508677298311445e-05,
+      "loss": 0.3509,
+      "step": 3845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5145403377110694e-05,
+      "loss": 0.3297,
+      "step": 3850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5204033771106947e-05,
+      "loss": 0.3432,
+      "step": 3855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.526266416510319e-05,
+      "loss": 0.3348,
+      "step": 3860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5321294559099445e-05,
+      "loss": 0.3428,
+      "step": 3865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.537992495309569e-05,
+      "loss": 0.3588,
+      "step": 3870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.543855534709193e-05,
+      "loss": 0.3464,
+      "step": 3875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.549718574108818e-05,
+      "loss": 0.3704,
+      "step": 3880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.555581613508443e-05,
+      "loss": 0.3672,
+      "step": 3885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.561444652908068e-05,
+      "loss": 0.3704,
+      "step": 3890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5673076923076925e-05,
+      "loss": 0.3336,
+      "step": 3895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.573170731707318e-05,
+      "loss": 0.3272,
+      "step": 3900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.579033771106942e-05,
+      "loss": 0.3223,
+      "step": 3905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.584896810506567e-05,
+      "loss": 0.361,
+      "step": 3910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.5907598499061914e-05,
+      "loss": 0.3281,
+      "step": 3915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.596622889305816e-05,
+      "loss": 0.3614,
+      "step": 3920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.602485928705441e-05,
+      "loss": 0.3411,
+      "step": 3925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.608348968105066e-05,
+      "loss": 0.3313,
+      "step": 3930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.614212007504691e-05,
+      "loss": 0.3629,
+      "step": 3935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.620075046904315e-05,
+      "loss": 0.3631,
+      "step": 3940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.62593808630394e-05,
+      "loss": 0.3539,
+      "step": 3945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.631801125703565e-05,
+      "loss": 0.3254,
+      "step": 3950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.63766416510319e-05,
+      "loss": 0.3704,
+      "step": 3955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6435272045028145e-05,
+      "loss": 0.3478,
+      "step": 3960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.64939024390244e-05,
+      "loss": 0.3576,
+      "step": 3965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.655253283302064e-05,
+      "loss": 0.3851,
+      "step": 3970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.661116322701688e-05,
+      "loss": 0.3313,
+      "step": 3975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.6669793621013135e-05,
+      "loss": 0.3325,
+      "step": 3980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.672842401500938e-05,
+      "loss": 0.3399,
+      "step": 3985
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.678705440900563e-05,
+      "loss": 0.3315,
+      "step": 3990
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.684568480300188e-05,
+      "loss": 0.3511,
+      "step": 3995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.690431519699813e-05,
+      "loss": 0.3507,
+      "step": 4000
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.34727856516838074,
+      "eval_runtime": 10.5437,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 4000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.696294559099437e-05,
+      "loss": 0.3252,
+      "step": 4005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.702157598499062e-05,
+      "loss": 0.3467,
+      "step": 4010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.708020637898687e-05,
+      "loss": 0.3564,
+      "step": 4015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.713883677298312e-05,
+      "loss": 0.3565,
+      "step": 4020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7197467166979365e-05,
+      "loss": 0.3467,
+      "step": 4025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.725609756097561e-05,
+      "loss": 0.3248,
+      "step": 4030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7314727954971863e-05,
+      "loss": 0.3628,
+      "step": 4035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.73733583489681e-05,
+      "loss": 0.3153,
+      "step": 4040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7431988742964355e-05,
+      "loss": 0.3289,
+      "step": 4045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.74906191369606e-05,
+      "loss": 0.3446,
+      "step": 4050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.754924953095685e-05,
+      "loss": 0.3631,
+      "step": 4055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76078799249531e-05,
+      "loss": 0.3446,
+      "step": 4060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766651031894935e-05,
+      "loss": 0.3435,
+      "step": 4065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772514071294559e-05,
+      "loss": 0.3314,
+      "step": 4070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778377110694184e-05,
+      "loss": 0.3587,
+      "step": 4075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.784240150093809e-05,
+      "loss": 0.3589,
+      "step": 4080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.790103189493433e-05,
+      "loss": 0.3404,
+      "step": 4085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7959662288930586e-05,
+      "loss": 0.3554,
+      "step": 4090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.801829268292683e-05,
+      "loss": 0.3638,
+      "step": 4095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8076923076923084e-05,
+      "loss": 0.3528,
+      "step": 4100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.813555347091932e-05,
+      "loss": 0.3536,
+      "step": 4105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.8194183864915575e-05,
+      "loss": 0.3846,
+      "step": 4110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.825281425891182e-05,
+      "loss": 0.3407,
+      "step": 4115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.831144465290807e-05,
+      "loss": 0.3505,
+      "step": 4120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.837007504690432e-05,
+      "loss": 0.3477,
+      "step": 4125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8428705440900564e-05,
+      "loss": 0.3686,
+      "step": 4130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8487335834896817e-05,
+      "loss": 0.3406,
+      "step": 4135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8545966228893055e-05,
+      "loss": 0.3638,
+      "step": 4140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.860459662288931e-05,
+      "loss": 0.3467,
+      "step": 4145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8663227016885553e-05,
+      "loss": 0.3367,
+      "step": 4150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8721857410881806e-05,
+      "loss": 0.3366,
+      "step": 4155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.878048780487805e-05,
+      "loss": 0.3646,
+      "step": 4160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8839118198874304e-05,
+      "loss": 0.3653,
+      "step": 4165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.889774859287054e-05,
+      "loss": 0.3176,
+      "step": 4170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.8956378986866795e-05,
+      "loss": 0.3331,
+      "step": 4175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.901500938086304e-05,
+      "loss": 0.3376,
+      "step": 4180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9073639774859286e-05,
+      "loss": 0.3516,
+      "step": 4185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.913227016885554e-05,
+      "loss": 0.3702,
+      "step": 4190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9190900562851784e-05,
+      "loss": 0.3589,
+      "step": 4195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.924953095684804e-05,
+      "loss": 0.3326,
+      "step": 4200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9308161350844276e-05,
+      "loss": 0.3562,
+      "step": 4205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.936679174484053e-05,
+      "loss": 0.3464,
+      "step": 4210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9425422138836774e-05,
+      "loss": 0.3674,
+      "step": 4215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9484052532833026e-05,
+      "loss": 0.3546,
+      "step": 4220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.954268292682927e-05,
+      "loss": 0.3918,
+      "step": 4225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9601313320825524e-05,
+      "loss": 0.3447,
+      "step": 4230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.965994371482176e-05,
+      "loss": 0.3669,
+      "step": 4235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.971857410881801e-05,
+      "loss": 0.3438,
+      "step": 4240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.977720450281426e-05,
+      "loss": 0.3425,
+      "step": 4245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9835834896810507e-05,
+      "loss": 0.3144,
+      "step": 4250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.989446529080676e-05,
+      "loss": 0.3529,
+      "step": 4255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9953095684803005e-05,
+      "loss": 0.3431,
+      "step": 4260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999999350783e-05,
+      "loss": 0.3392,
+      "step": 4265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999976628187e-05,
+      "loss": 0.3406,
+      "step": 4270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999999214447376e-05,
+      "loss": 0.3477,
+      "step": 4275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999998338004374e-05,
+      "loss": 0.3506,
+      "step": 4280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999713695287e-05,
+      "loss": 0.3389,
+      "step": 4285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999561129288e-05,
+      "loss": 0.3461,
+      "step": 4290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999376102442e-05,
+      "loss": 0.3139,
+      "step": 4295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999999158614751e-05,
+      "loss": 0.3506,
+      "step": 4300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999998908666219e-05,
+      "loss": 0.3673,
+      "step": 4305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999986262568495e-05,
+      "loss": 0.3583,
+      "step": 4310
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999983113866445e-05,
+      "loss": 0.341,
+      "step": 4315
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999979640556095e-05,
+      "loss": 0.3345,
+      "step": 4320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999975842637484e-05,
+      "loss": 0.3396,
+      "step": 4325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999997172011066e-05,
+      "loss": 0.3524,
+      "step": 4330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999996727297568e-05,
+      "loss": 0.3158,
+      "step": 4335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.99999625012326e-05,
+      "loss": 0.37,
+      "step": 4340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999995740488148e-05,
+      "loss": 0.3554,
+      "step": 4345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999995198392239e-05,
+      "loss": 0.3597,
+      "step": 4350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999946238355405e-05,
+      "loss": 0.3107,
+      "step": 4355
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999994016818059e-05,
+      "loss": 0.3513,
+      "step": 4360
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999993377339802e-05,
+      "loss": 0.3312,
+      "step": 4365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.99999270540078e-05,
+      "loss": 0.3627,
+      "step": 4370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999920010009995e-05,
+      "loss": 0.3232,
+      "step": 4375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.99999126414047e-05,
+      "loss": 0.3548,
+      "step": 4380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999904948192025e-05,
+      "loss": 0.3475,
+      "step": 4385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999989693037205e-05,
+      "loss": 0.3552,
+      "step": 4390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.99998885879449e-05,
+      "loss": 0.3655,
+      "step": 4395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.9999879920910664e-05,
+      "loss": 0.3671,
+      "step": 4400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.999987092926946e-05,
+      "loss": 0.3337,
+      "step": 4405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999861613021416e-05,
+      "loss": 0.3321,
+      "step": 4410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999985197216663e-05,
+      "loss": 0.3543,
+      "step": 4415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999984200670526e-05,
+      "loss": 0.3368,
+      "step": 4420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.99998317166374e-05,
+      "loss": 0.351,
+      "step": 4425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999982110196322e-05,
+      "loss": 0.3441,
+      "step": 4430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999810162682824e-05,
+      "loss": 0.3481,
+      "step": 4435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999979889879637e-05,
+      "loss": 0.3649,
+      "step": 4440
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999787310304e-05,
+      "loss": 0.3493,
+      "step": 4445
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999775397205875e-05,
+      "loss": 0.3405,
+      "step": 4450
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999763159502146e-05,
+      "loss": 0.3482,
+      "step": 4455
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999975059719296e-05,
+      "loss": 0.3288,
+      "step": 4460
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999737710278485e-05,
+      "loss": 0.3454,
+      "step": 4465
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.99997244987589e-05,
+      "loss": 0.3426,
+      "step": 4470
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999971096263436e-05,
+      "loss": 0.3203,
+      "step": 4475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999969710190505e-05,
+      "loss": 0.3542,
+      "step": 4480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999968291657115e-05,
+      "loss": 0.3207,
+      "step": 4485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999966840663285e-05,
+      "loss": 0.336,
+      "step": 4490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999965357209032e-05,
+      "loss": 0.3624,
+      "step": 4495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999963841294377e-05,
+      "loss": 0.3367,
+      "step": 4500
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 0.3431379199028015,
+      "eval_runtime": 10.5416,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 4500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999962292919339e-05,
+      "loss": 0.3374,
+      "step": 4505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999607120839374e-05,
+      "loss": 0.345,
+      "step": 4510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999590987881944e-05,
+      "loss": 0.3261,
+      "step": 4515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.99995745303213e-05,
+      "loss": 0.303,
+      "step": 4520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999955774815766e-05,
+      "loss": 0.3219,
+      "step": 4525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999954064139122e-05,
+      "loss": 0.3604,
+      "step": 4530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999952321002224e-05,
+      "loss": 0.3528,
+      "step": 4535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999505454050915e-05,
+      "loss": 0.3385,
+      "step": 4540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999487373477486e-05,
+      "loss": 0.344,
+      "step": 4545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.99994689683022e-05,
+      "loss": 0.3397,
+      "step": 4550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999945023852527e-05,
+      "loss": 0.3112,
+      "step": 4555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999943118414696e-05,
+      "loss": 0.361,
+      "step": 4560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999411805167515e-05,
+      "loss": 0.3396,
+      "step": 4565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999939210158718e-05,
+      "loss": 0.3423,
+      "step": 4570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999937207340622e-05,
+      "loss": 0.3222,
+      "step": 4575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999935172062488e-05,
+      "loss": 0.2994,
+      "step": 4580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999331043243436e-05,
+      "loss": 0.3463,
+      "step": 4585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999310041262154e-05,
+      "loss": 0.3634,
+      "step": 4590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999288714681306e-05,
+      "loss": 0.3839,
+      "step": 4595
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999926706350117e-05,
+      "loss": 0.345,
+      "step": 4600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999924508772202e-05,
+      "loss": 0.3378,
+      "step": 4605
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999222787344156e-05,
+      "loss": 0.3724,
+      "step": 4610
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999920016236786e-05,
+      "loss": 0.3219,
+      "step": 4615
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999177212793424e-05,
+      "loss": 0.353,
+      "step": 4620
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999915393862115e-05,
+      "loss": 0.3162,
+      "step": 4625
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999130339851325e-05,
+      "loss": 0.3491,
+      "step": 4630
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999106416484274e-05,
+      "loss": 0.3276,
+      "step": 4635
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.99990821685203e-05,
+      "loss": 0.3685,
+      "step": 4640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999905759595972e-05,
+      "loss": 0.3249,
+      "step": 4645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9999032698802853e-05,
+      "loss": 0.3665,
+      "step": 4650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999900747705002e-05,
+      "loss": 0.3448,
+      "step": 4655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9998981930701556e-05,
+      "loss": 0.3512,
+      "step": 4660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999895605975778e-05,
+      "loss": 0.3621,
+      "step": 4665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9998929864219035e-05,
+      "loss": 0.3259,
+      "step": 4670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.999890334408566e-05,
+      "loss": 0.3641,
+      "step": 4675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9998876499358e-05,
+      "loss": 0.3532,
+      "step": 4680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.9998849330036414e-05,
+      "loss": 0.3338,
+      "step": 4685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999882183612123e-05,
+      "loss": 0.3298,
+      "step": 4690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999879401761283e-05,
+      "loss": 0.3281,
+      "step": 4695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998765874511554e-05,
+      "loss": 0.3606,
+      "step": 4700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999873740681779e-05,
+      "loss": 0.3785,
+      "step": 4705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998708614531885e-05,
+      "loss": 0.3164,
+      "step": 4710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999867949765422e-05,
+      "loss": 0.334,
+      "step": 4715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999865005618519e-05,
+      "loss": 0.3884,
+      "step": 4720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999862029012516e-05,
+      "loss": 0.3537,
+      "step": 4725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998590199474516e-05,
+      "loss": 0.2878,
+      "step": 4730
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999855978423365e-05,
+      "loss": 0.3349,
+      "step": 4735
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999852904440296e-05,
+      "loss": 0.3547,
+      "step": 4740
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999849797998286e-05,
+      "loss": 0.3466,
+      "step": 4745
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999846659097373e-05,
+      "loss": 0.3519,
+      "step": 4750
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999843487737598e-05,
+      "loss": 0.3389,
+      "step": 4755
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999840283919003e-05,
+      "loss": 0.3458,
+      "step": 4760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999837047641629e-05,
+      "loss": 0.3598,
+      "step": 4765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998337789055195e-05,
+      "loss": 0.3349,
+      "step": 4770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998304777107155e-05,
+      "loss": 0.3621,
+      "step": 4775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99982714405726e-05,
+      "loss": 0.3359,
+      "step": 4780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999823777945196e-05,
+      "loss": 0.3329,
+      "step": 4785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999820379374568e-05,
+      "loss": 0.3453,
+      "step": 4790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.99981694834542e-05,
+      "loss": 0.3564,
+      "step": 4795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999813484857796e-05,
+      "loss": 0.3436,
+      "step": 4800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998099889117416e-05,
+      "loss": 0.35,
+      "step": 4805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9998064605073025e-05,
+      "loss": 0.336,
+      "step": 4810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999802899644524e-05,
+      "loss": 0.3788,
+      "step": 4815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997993063234505e-05,
+      "loss": 0.3505,
+      "step": 4820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997956805441316e-05,
+      "loss": 0.36,
+      "step": 4825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999792022306613e-05,
+      "loss": 0.3545,
+      "step": 4830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999788331610943e-05,
+      "loss": 0.3646,
+      "step": 4835
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999784608457169e-05,
+      "loss": 0.3414,
+      "step": 4840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997808528453386e-05,
+      "loss": 0.3378,
+      "step": 4845
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999777064775502e-05,
+      "loss": 0.3516,
+      "step": 4850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999773244247706e-05,
+      "loss": 0.343,
+      "step": 4855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997693912620035e-05,
+      "loss": 0.3429,
+      "step": 4860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999765505818442e-05,
+      "loss": 0.3423,
+      "step": 4865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999761587917073e-05,
+      "loss": 0.321,
+      "step": 4870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999757637557948e-05,
+      "loss": 0.3474,
+      "step": 4875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997536547411165e-05,
+      "loss": 0.3708,
+      "step": 4880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999749639466632e-05,
+      "loss": 0.3307,
+      "step": 4885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999745591734545e-05,
+      "loss": 0.3124,
+      "step": 4890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999741511544909e-05,
+      "loss": 0.3249,
+      "step": 4895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997373988977776e-05,
+      "loss": 0.321,
+      "step": 4900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999733253793203e-05,
+      "loss": 0.3264,
+      "step": 4905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999729076231239e-05,
+      "loss": 0.3436,
+      "step": 4910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999724866211941e-05,
+      "loss": 0.3249,
+      "step": 4915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999720623735363e-05,
+      "loss": 0.3331,
+      "step": 4920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997163488015595e-05,
+      "loss": 0.3406,
+      "step": 4925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999712041410587e-05,
+      "loss": 0.3397,
+      "step": 4930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997077015625007e-05,
+      "loss": 0.3426,
+      "step": 4935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9997033292573575e-05,
+      "loss": 0.3314,
+      "step": 4940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999698924495214e-05,
+      "loss": 0.3547,
+      "step": 4945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999694487276128e-05,
+      "loss": 0.3362,
+      "step": 4950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999690017600155e-05,
+      "loss": 0.3154,
+      "step": 4955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999685515467355e-05,
+      "loss": 0.3549,
+      "step": 4960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999680980877787e-05,
+      "loss": 0.3302,
+      "step": 4965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.999676413831507e-05,
+      "loss": 0.3415,
+      "step": 4970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999671814328577e-05,
+      "loss": 0.3449,
+      "step": 4975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996671823690565e-05,
+      "loss": 0.3202,
+      "step": 4980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996625179530046e-05,
+      "loss": 0.3373,
+      "step": 4985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999657821080481e-05,
+      "loss": 0.3344,
+      "step": 4990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999653091751549e-05,
+      "loss": 0.3466,
+      "step": 4995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996483299662694e-05,
+      "loss": 0.3277,
+      "step": 5000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.33953267335891724,
+      "eval_runtime": 10.5352,
+      "eval_samples_per_second": 9.492,
+      "eval_steps_per_second": 9.492,
+      "step": 5000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999643535724703e-05,
+      "loss": 0.345,
+      "step": 5005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999638709026913e-05,
+      "loss": 0.3589,
+      "step": 5010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999633849872961e-05,
+      "loss": 0.3331,
+      "step": 5015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999628958262911e-05,
+      "loss": 0.3244,
+      "step": 5020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996240341968265e-05,
+      "loss": 0.3285,
+      "step": 5025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999619077674771e-05,
+      "loss": 0.3572,
+      "step": 5030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999614088696809e-05,
+      "loss": 0.3537,
+      "step": 5035
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9996090672630056e-05,
+      "loss": 0.3523,
+      "step": 5040
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999604013373426e-05,
+      "loss": 0.3402,
+      "step": 5045
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999598927028135e-05,
+      "loss": 0.3613,
+      "step": 5050
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995938082271996e-05,
+      "loss": 0.356,
+      "step": 5055
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999588656970686e-05,
+      "loss": 0.3402,
+      "step": 5060
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999583473258661e-05,
+      "loss": 0.3801,
+      "step": 5065
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999578257091192e-05,
+      "loss": 0.341,
+      "step": 5070
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999573008468347e-05,
+      "loss": 0.3114,
+      "step": 5075
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995677273901924e-05,
+      "loss": 0.356,
+      "step": 5080
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999562413856799e-05,
+      "loss": 0.3366,
+      "step": 5085
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999557067868235e-05,
+      "loss": 0.3245,
+      "step": 5090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999551689424569e-05,
+      "loss": 0.3187,
+      "step": 5095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999546278525872e-05,
+      "loss": 0.3288,
+      "step": 5100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995408351722145e-05,
+      "loss": 0.35,
+      "step": 5105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999535359363666e-05,
+      "loss": 0.3453,
+      "step": 5110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999529851100298e-05,
+      "loss": 0.3534,
+      "step": 5115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999524310382182e-05,
+      "loss": 0.3667,
+      "step": 5120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995187372093906e-05,
+      "loss": 0.3592,
+      "step": 5125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999513131581995e-05,
+      "loss": 0.3338,
+      "step": 5130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9995074935000694e-05,
+      "loss": 0.3393,
+      "step": 5135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999501822963686e-05,
+      "loss": 0.3725,
+      "step": 5140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999496119972918e-05,
+      "loss": 0.3396,
+      "step": 5145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999490384527841e-05,
+      "loss": 0.3536,
+      "step": 5150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994846166285284e-05,
+      "loss": 0.3628,
+      "step": 5155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999478816275054e-05,
+      "loss": 0.343,
+      "step": 5160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999472983467496e-05,
+      "loss": 0.3212,
+      "step": 5165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994671182059274e-05,
+      "loss": 0.3293,
+      "step": 5170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999461220490427e-05,
+      "loss": 0.3619,
+      "step": 5175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999455290321068e-05,
+      "loss": 0.3588,
+      "step": 5180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994493276979304e-05,
+      "loss": 0.3385,
+      "step": 5185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.99944333262109e-05,
+      "loss": 0.3556,
+      "step": 5190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999437305090625e-05,
+      "loss": 0.3072,
+      "step": 5195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999431245106615e-05,
+      "loss": 0.3087,
+      "step": 5200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999425152669136e-05,
+      "loss": 0.3562,
+      "step": 5205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994190277782696e-05,
+      "loss": 0.3484,
+      "step": 5210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994128704340936e-05,
+      "loss": 0.3419,
+      "step": 5215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999406680636689e-05,
+      "loss": 0.3368,
+      "step": 5220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9994004583861356e-05,
+      "loss": 0.3485,
+      "step": 5225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999394203682515e-05,
+      "loss": 0.3424,
+      "step": 5230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9993879165259074e-05,
+      "loss": 0.3392,
+      "step": 5235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999381596916395e-05,
+      "loss": 0.3408,
+      "step": 5240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.9993752448540596e-05,
+      "loss": 0.3023,
+      "step": 5245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.999368860338983e-05,
+      "loss": 0.3318,
+      "step": 5250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.99936244337125e-05,
+      "loss": 0.3265,
+      "step": 5255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993559939509426e-05,
+      "loss": 0.3451,
+      "step": 5260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993495120781445e-05,
+      "loss": 0.3405,
+      "step": 5265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993429977529405e-05,
+      "loss": 0.3369,
+      "step": 5270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999336450975414e-05,
+      "loss": 0.3678,
+      "step": 5275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999329871745652e-05,
+      "loss": 0.3559,
+      "step": 5280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993232600637374e-05,
+      "loss": 0.3529,
+      "step": 5285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999316615929759e-05,
+      "loss": 0.3264,
+      "step": 5290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993099393438006e-05,
+      "loss": 0.3242,
+      "step": 5295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9993032303059486e-05,
+      "loss": 0.326,
+      "step": 5300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999296488816293e-05,
+      "loss": 0.3212,
+      "step": 5305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999289714874918e-05,
+      "loss": 0.341,
+      "step": 5310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999282908481915e-05,
+      "loss": 0.3389,
+      "step": 5315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9992760696373695e-05,
+      "loss": 0.3637,
+      "step": 5320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999269198341372e-05,
+      "loss": 0.3309,
+      "step": 5325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99926229459401e-05,
+      "loss": 0.3434,
+      "step": 5330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999255358395374e-05,
+      "loss": 0.3455,
+      "step": 5335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999248389745556e-05,
+      "loss": 0.342,
+      "step": 5340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999241388644643e-05,
+      "loss": 0.337,
+      "step": 5345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999234355092728e-05,
+      "loss": 0.3357,
+      "step": 5350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9992272890899024e-05,
+      "loss": 0.3281,
+      "step": 5355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999220190636257e-05,
+      "loss": 0.3208,
+      "step": 5360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999213059731885e-05,
+      "loss": 0.3598,
+      "step": 5365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999205896376878e-05,
+      "loss": 0.3522,
+      "step": 5370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99919870057133e-05,
+      "loss": 0.3178,
+      "step": 5375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999191472315333e-05,
+      "loss": 0.3168,
+      "step": 5380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999184211608983e-05,
+      "loss": 0.3462,
+      "step": 5385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999176918452372e-05,
+      "loss": 0.351,
+      "step": 5390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999169592845596e-05,
+      "loss": 0.3342,
+      "step": 5395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99916223478875e-05,
+      "loss": 0.3408,
+      "step": 5400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99915484428193e-05,
+      "loss": 0.3531,
+      "step": 5405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999147421325231e-05,
+      "loss": 0.359,
+      "step": 5410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999139965918749e-05,
+      "loss": 0.3325,
+      "step": 5415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999132478062582e-05,
+      "loss": 0.3474,
+      "step": 5420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991249577568264e-05,
+      "loss": 0.3529,
+      "step": 5425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999117405001581e-05,
+      "loss": 0.3142,
+      "step": 5430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9991098197969426e-05,
+      "loss": 0.3436,
+      "step": 5435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99910220214301e-05,
+      "loss": 0.3514,
+      "step": 5440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999094552039883e-05,
+      "loss": 0.331,
+      "step": 5445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999086869487659e-05,
+      "loss": 0.3635,
+      "step": 5450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990791544864404e-05,
+      "loss": 0.3251,
+      "step": 5455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999071407036325e-05,
+      "loss": 0.3262,
+      "step": 5460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990636271374144e-05,
+      "loss": 0.3029,
+      "step": 5465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99905581478981e-05,
+      "loss": 0.3293,
+      "step": 5470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999047969993612e-05,
+      "loss": 0.3535,
+      "step": 5475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990400927489234e-05,
+      "loss": 0.3354,
+      "step": 5480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999032183055846e-05,
+      "loss": 0.3349,
+      "step": 5485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990242409144835e-05,
+      "loss": 0.3415,
+      "step": 5490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9990162663249373e-05,
+      "loss": 0.3273,
+      "step": 5495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999008259287312e-05,
+      "loss": 0.3508,
+      "step": 5500
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.3391436040401459,
+      "eval_runtime": 10.541,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 5500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.999000219801712e-05,
+      "loss": 0.3461,
+      "step": 5505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.99899214786824e-05,
+      "loss": 0.3272,
+      "step": 5510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9989840434870025e-05,
+      "loss": 0.3521,
+      "step": 5515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.998975906658103e-05,
+      "loss": 0.3151,
+      "step": 5520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9989677373816494e-05,
+      "loss": 0.3279,
+      "step": 5525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9989595356577465e-05,
+      "loss": 0.3475,
+      "step": 5530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.998951301486501e-05,
+      "loss": 0.3282,
+      "step": 5535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.9989430348680186e-05,
+      "loss": 0.3457,
+      "step": 5540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998934735802409e-05,
+      "loss": 0.3311,
+      "step": 5545
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998926404289779e-05,
+      "loss": 0.3496,
+      "step": 5550
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998918040330235e-05,
+      "loss": 0.3349,
+      "step": 5555
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9989096439238884e-05,
+      "loss": 0.3327,
+      "step": 5560
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998901215070846e-05,
+      "loss": 0.3649,
+      "step": 5565
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998892753771219e-05,
+      "loss": 0.3336,
+      "step": 5570
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998884260025116e-05,
+      "loss": 0.3531,
+      "step": 5575
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988757338326474e-05,
+      "loss": 0.3581,
+      "step": 5580
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998867175193924e-05,
+      "loss": 0.3178,
+      "step": 5585
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998858584109059e-05,
+      "loss": 0.3431,
+      "step": 5590
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.99884996057816e-05,
+      "loss": 0.3362,
+      "step": 5595
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998841304601342e-05,
+      "loss": 0.359,
+      "step": 5600
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998832616178716e-05,
+      "loss": 0.331,
+      "step": 5605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998823895310396e-05,
+      "loss": 0.3425,
+      "step": 5610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9988151419964935e-05,
+      "loss": 0.3234,
+      "step": 5615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998806356237124e-05,
+      "loss": 0.3417,
+      "step": 5620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9987975380324e-05,
+      "loss": 0.351,
+      "step": 5625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9987886873824374e-05,
+      "loss": 0.352,
+      "step": 5630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.99877980428735e-05,
+      "loss": 0.364,
+      "step": 5635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998770888747254e-05,
+      "loss": 0.3268,
+      "step": 5640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998761940762265e-05,
+      "loss": 0.3856,
+      "step": 5645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998752960332498e-05,
+      "loss": 0.3381,
+      "step": 5650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998743947458072e-05,
+      "loss": 0.3492,
+      "step": 5655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998734902139102e-05,
+      "loss": 0.3318,
+      "step": 5660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9987258243757055e-05,
+      "loss": 0.3393,
+      "step": 5665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998716714168e-05,
+      "loss": 0.3467,
+      "step": 5670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9987075715161064e-05,
+      "loss": 0.3667,
+      "step": 5675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998698396420141e-05,
+      "loss": 0.3488,
+      "step": 5680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998689188880223e-05,
+      "loss": 0.3166,
+      "step": 5685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998679948896474e-05,
+      "loss": 0.3351,
+      "step": 5690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998670676469012e-05,
+      "loss": 0.3503,
+      "step": 5695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998661371597958e-05,
+      "loss": 0.3226,
+      "step": 5700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998652034283432e-05,
+      "loss": 0.3403,
+      "step": 5705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998642664525557e-05,
+      "loss": 0.3472,
+      "step": 5710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998633262324453e-05,
+      "loss": 0.3504,
+      "step": 5715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9986238276802435e-05,
+      "loss": 0.3345,
+      "step": 5720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998614360593049e-05,
+      "loss": 0.3224,
+      "step": 5725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998604861062995e-05,
+      "loss": 0.3267,
+      "step": 5730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9985953290902024e-05,
+      "loss": 0.3438,
+      "step": 5735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998585764674797e-05,
+      "loss": 0.3184,
+      "step": 5740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998576167816902e-05,
+      "loss": 0.3406,
+      "step": 5745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9985665385166416e-05,
+      "loss": 0.3495,
+      "step": 5750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998556876774141e-05,
+      "loss": 0.3361,
+      "step": 5755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998547182589527e-05,
+      "loss": 0.3761,
+      "step": 5760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998537455962924e-05,
+      "loss": 0.3232,
+      "step": 5765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998527696894458e-05,
+      "loss": 0.3331,
+      "step": 5770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998517905384258e-05,
+      "loss": 0.3593,
+      "step": 5775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998508081432448e-05,
+      "loss": 0.3332,
+      "step": 5780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998498225039158e-05,
+      "loss": 0.3465,
+      "step": 5785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998488336204515e-05,
+      "loss": 0.3313,
+      "step": 5790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998478414928647e-05,
+      "loss": 0.3309,
+      "step": 5795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998468461211684e-05,
+      "loss": 0.3517,
+      "step": 5800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998458475053754e-05,
+      "loss": 0.3301,
+      "step": 5805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9984484564549884e-05,
+      "loss": 0.3438,
+      "step": 5810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998438405415515e-05,
+      "loss": 0.334,
+      "step": 5815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998428321935466e-05,
+      "loss": 0.3086,
+      "step": 5820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.998418206014971e-05,
+      "loss": 0.3291,
+      "step": 5825
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9984080576541636e-05,
+      "loss": 0.3316,
+      "step": 5830
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9983978768531724e-05,
+      "loss": 0.3494,
+      "step": 5835
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9983876636121325e-05,
+      "loss": 0.3372,
+      "step": 5840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998377417931175e-05,
+      "loss": 0.3628,
+      "step": 5845
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998367139810433e-05,
+      "loss": 0.3209,
+      "step": 5850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99835682925004e-05,
+      "loss": 0.3382,
+      "step": 5855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998346486250131e-05,
+      "loss": 0.3258,
+      "step": 5860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998336110810838e-05,
+      "loss": 0.3392,
+      "step": 5865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9983257029322975e-05,
+      "loss": 0.3324,
+      "step": 5870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998315262614645e-05,
+      "loss": 0.3524,
+      "step": 5875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9983047898580134e-05,
+      "loss": 0.3644,
+      "step": 5880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998294284662542e-05,
+      "loss": 0.3748,
+      "step": 5885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998283747028366e-05,
+      "loss": 0.326,
+      "step": 5890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9982731769556216e-05,
+      "loss": 0.3364,
+      "step": 5895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9982625744444464e-05,
+      "loss": 0.3307,
+      "step": 5900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9982519394949776e-05,
+      "loss": 0.3832,
+      "step": 5905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998241272107355e-05,
+      "loss": 0.3344,
+      "step": 5910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998230572281715e-05,
+      "loss": 0.3497,
+      "step": 5915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9982198400181975e-05,
+      "loss": 0.3361,
+      "step": 5920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998209075316942e-05,
+      "loss": 0.3129,
+      "step": 5925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998198278178088e-05,
+      "loss": 0.3266,
+      "step": 5930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9981874486017746e-05,
+      "loss": 0.3299,
+      "step": 5935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9981765865881444e-05,
+      "loss": 0.3378,
+      "step": 5940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998165692137338e-05,
+      "loss": 0.3558,
+      "step": 5945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998154765249496e-05,
+      "loss": 0.3249,
+      "step": 5950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998143805924761e-05,
+      "loss": 0.3064,
+      "step": 5955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998132814163275e-05,
+      "loss": 0.3358,
+      "step": 5960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998121789965181e-05,
+      "loss": 0.3356,
+      "step": 5965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998110733330621e-05,
+      "loss": 0.3289,
+      "step": 5970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99809964425974e-05,
+      "loss": 0.3755,
+      "step": 5975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998088522752681e-05,
+      "loss": 0.3381,
+      "step": 5980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9980773688095896e-05,
+      "loss": 0.3545,
+      "step": 5985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.99806618243061e-05,
+      "loss": 0.3409,
+      "step": 5990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998054963615886e-05,
+      "loss": 0.336,
+      "step": 5995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998043712365566e-05,
+      "loss": 0.3559,
+      "step": 6000
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 0.3358333706855774,
+      "eval_runtime": 10.5554,
+      "eval_samples_per_second": 9.474,
+      "eval_steps_per_second": 9.474,
+      "step": 6000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9980324286797935e-05,
+      "loss": 0.3347,
+      "step": 6005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9980211125587166e-05,
+      "loss": 0.3317,
+      "step": 6010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.998009764002482e-05,
+      "loss": 0.3474,
+      "step": 6015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9979983830112366e-05,
+      "loss": 0.3507,
+      "step": 6020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997986969585129e-05,
+      "loss": 0.3373,
+      "step": 6025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997975523724306e-05,
+      "loss": 0.3213,
+      "step": 6030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9979640454289175e-05,
+      "loss": 0.3132,
+      "step": 6035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9979525346991126e-05,
+      "loss": 0.3358,
+      "step": 6040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997940991535039e-05,
+      "loss": 0.3218,
+      "step": 6045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997929415936849e-05,
+      "loss": 0.3567,
+      "step": 6050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.9979178079046914e-05,
+      "loss": 0.3256,
+      "step": 6055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997906167438717e-05,
+      "loss": 0.3522,
+      "step": 6060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997894494539077e-05,
+      "loss": 0.351,
+      "step": 6065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997882789205924e-05,
+      "loss": 0.3127,
+      "step": 6070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997871051439409e-05,
+      "loss": 0.3374,
+      "step": 6075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997859281239684e-05,
+      "loss": 0.3133,
+      "step": 6080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997847478606902e-05,
+      "loss": 0.3224,
+      "step": 6085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997835643541218e-05,
+      "loss": 0.3394,
+      "step": 6090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997823776042782e-05,
+      "loss": 0.3583,
+      "step": 6095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997811876111753e-05,
+      "loss": 0.3395,
+      "step": 6100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997799943748281e-05,
+      "loss": 0.3493,
+      "step": 6105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.997787978952524e-05,
+      "loss": 0.3356,
+      "step": 6110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997775981724636e-05,
+      "loss": 0.3484,
+      "step": 6115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9977639520647735e-05,
+      "loss": 0.3517,
+      "step": 6120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9977518899730915e-05,
+      "loss": 0.3185,
+      "step": 6125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997739795449747e-05,
+      "loss": 0.3593,
+      "step": 6130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997727668494898e-05,
+      "loss": 0.3254,
+      "step": 6135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9977155091087e-05,
+      "loss": 0.3356,
+      "step": 6140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997703317291313e-05,
+      "loss": 0.355,
+      "step": 6145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997691093042896e-05,
+      "loss": 0.3249,
+      "step": 6150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997678836363604e-05,
+      "loss": 0.3273,
+      "step": 6155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997666547253599e-05,
+      "loss": 0.3471,
+      "step": 6160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.99765422571304e-05,
+      "loss": 0.3326,
+      "step": 6165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997641871742087e-05,
+      "loss": 0.3423,
+      "step": 6170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9976294853409e-05,
+      "loss": 0.3468,
+      "step": 6175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9976170665096405e-05,
+      "loss": 0.3496,
+      "step": 6180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997604615248469e-05,
+      "loss": 0.3701,
+      "step": 6185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997592131557548e-05,
+      "loss": 0.34,
+      "step": 6190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997579615437039e-05,
+      "loss": 0.3093,
+      "step": 6195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997567066887105e-05,
+      "loss": 0.3141,
+      "step": 6200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9975544859079074e-05,
+      "loss": 0.3394,
+      "step": 6205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997541872499611e-05,
+      "loss": 0.3649,
+      "step": 6210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997529226662381e-05,
+      "loss": 0.3132,
+      "step": 6215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997516548396378e-05,
+      "loss": 0.3583,
+      "step": 6220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9975038377017705e-05,
+      "loss": 0.3419,
+      "step": 6225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.99749109457872e-05,
+      "loss": 0.3315,
+      "step": 6230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997478319027394e-05,
+      "loss": 0.3451,
+      "step": 6235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9974655110479576e-05,
+      "loss": 0.3336,
+      "step": 6240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997452670640578e-05,
+      "loss": 0.3494,
+      "step": 6245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997439797805421e-05,
+      "loss": 0.3194,
+      "step": 6250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997426892542654e-05,
+      "loss": 0.3357,
+      "step": 6255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997413954852445e-05,
+      "loss": 0.3278,
+      "step": 6260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997400984734961e-05,
+      "loss": 0.3397,
+      "step": 6265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997387982190372e-05,
+      "loss": 0.3387,
+      "step": 6270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997374947218845e-05,
+      "loss": 0.3167,
+      "step": 6275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997361879820551e-05,
+      "loss": 0.2976,
+      "step": 6280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997348779995659e-05,
+      "loss": 0.3262,
+      "step": 6285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997335647744338e-05,
+      "loss": 0.3545,
+      "step": 6290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997322483066759e-05,
+      "loss": 0.3226,
+      "step": 6295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9973092859630945e-05,
+      "loss": 0.35,
+      "step": 6300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997296056433514e-05,
+      "loss": 0.3216,
+      "step": 6305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997282794478191e-05,
+      "loss": 0.3178,
+      "step": 6310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997269500097296e-05,
+      "loss": 0.3279,
+      "step": 6315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997256173291002e-05,
+      "loss": 0.3349,
+      "step": 6320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9972428140594824e-05,
+      "loss": 0.3049,
+      "step": 6325
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.99722942240291e-05,
+      "loss": 0.3463,
+      "step": 6330
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9972159983214605e-05,
+      "loss": 0.3329,
+      "step": 6335
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9972025418153065e-05,
+      "loss": 0.3367,
+      "step": 6340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9971890528846235e-05,
+      "loss": 0.337,
+      "step": 6345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997175531529586e-05,
+      "loss": 0.3383,
+      "step": 6350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9971619777503705e-05,
+      "loss": 0.3563,
+      "step": 6355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.9971483915471516e-05,
+      "loss": 0.3604,
+      "step": 6360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997134772920107e-05,
+      "loss": 0.3429,
+      "step": 6365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997121121869414e-05,
+      "loss": 0.3422,
+      "step": 6370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997107438395247e-05,
+      "loss": 0.3272,
+      "step": 6375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997093722497787e-05,
+      "loss": 0.3286,
+      "step": 6380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.99707997417721e-05,
+      "loss": 0.3103,
+      "step": 6385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997066193433695e-05,
+      "loss": 0.3118,
+      "step": 6390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.997052380267421e-05,
+      "loss": 0.3341,
+      "step": 6395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997038534678568e-05,
+      "loss": 0.35,
+      "step": 6400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997024656667315e-05,
+      "loss": 0.3533,
+      "step": 6405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.997010746233843e-05,
+      "loss": 0.3333,
+      "step": 6410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996996803378331e-05,
+      "loss": 0.336,
+      "step": 6415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996982828100962e-05,
+      "loss": 0.3255,
+      "step": 6420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996968820401916e-05,
+      "loss": 0.3319,
+      "step": 6425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996954780281375e-05,
+      "loss": 0.358,
+      "step": 6430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996940707739523e-05,
+      "loss": 0.3364,
+      "step": 6435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9969266027765404e-05,
+      "loss": 0.3389,
+      "step": 6440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9969124653926127e-05,
+      "loss": 0.3334,
+      "step": 6445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996898295587921e-05,
+      "loss": 0.344,
+      "step": 6450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9968840933626506e-05,
+      "loss": 0.3552,
+      "step": 6455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996869858716986e-05,
+      "loss": 0.3372,
+      "step": 6460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9968555916511114e-05,
+      "loss": 0.3244,
+      "step": 6465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996841292165213e-05,
+      "loss": 0.3458,
+      "step": 6470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996826960259475e-05,
+      "loss": 0.3342,
+      "step": 6475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996812595934085e-05,
+      "loss": 0.3492,
+      "step": 6480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996798199189229e-05,
+      "loss": 0.343,
+      "step": 6485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996783770025094e-05,
+      "loss": 0.3175,
+      "step": 6490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967693084418665e-05,
+      "loss": 0.3195,
+      "step": 6495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967548144397356e-05,
+      "loss": 0.3236,
+      "step": 6500
+    },
+    {
+      "epoch": 0.23,
+      "eval_loss": 0.3339012861251831,
+      "eval_runtime": 10.5429,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 6500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967402880188884e-05,
+      "loss": 0.3493,
+      "step": 6505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9967257291795147e-05,
+      "loss": 0.3457,
+      "step": 6510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996711137921802e-05,
+      "loss": 0.3331,
+      "step": 6515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996696514245941e-05,
+      "loss": 0.3379,
+      "step": 6520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9966818581521214e-05,
+      "loss": 0.3319,
+      "step": 6525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996667169640533e-05,
+      "loss": 0.3487,
+      "step": 6530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996652448711366e-05,
+      "loss": 0.3356,
+      "step": 6535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996637695364813e-05,
+      "loss": 0.3379,
+      "step": 6540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996622909601065e-05,
+      "loss": 0.3411,
+      "step": 6545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996608091420314e-05,
+      "loss": 0.323,
+      "step": 6550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996593240822752e-05,
+      "loss": 0.3342,
+      "step": 6555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996578357808572e-05,
+      "loss": 0.336,
+      "step": 6560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996563442377967e-05,
+      "loss": 0.3418,
+      "step": 6565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996548494531131e-05,
+      "loss": 0.3527,
+      "step": 6570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996533514268259e-05,
+      "loss": 0.3421,
+      "step": 6575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996518501589545e-05,
+      "loss": 0.3535,
+      "step": 6580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996503456495182e-05,
+      "loss": 0.3448,
+      "step": 6585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996488378985367e-05,
+      "loss": 0.3458,
+      "step": 6590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9964732690602964e-05,
+      "loss": 0.3429,
+      "step": 6595
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9964581267201646e-05,
+      "loss": 0.3458,
+      "step": 6600
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.99644295196517e-05,
+      "loss": 0.3492,
+      "step": 6605
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996427744795509e-05,
+      "loss": 0.3352,
+      "step": 6610
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996412505211378e-05,
+      "loss": 0.3193,
+      "step": 6615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996397233212978e-05,
+      "loss": 0.3528,
+      "step": 6620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996381928800503e-05,
+      "loss": 0.3292,
+      "step": 6625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996366591974155e-05,
+      "loss": 0.3336,
+      "step": 6630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9963512227341315e-05,
+      "loss": 0.3314,
+      "step": 6635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996335821080632e-05,
+      "loss": 0.3274,
+      "step": 6640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996320387013858e-05,
+      "loss": 0.3435,
+      "step": 6645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996304920534008e-05,
+      "loss": 0.3747,
+      "step": 6650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9962894216412845e-05,
+      "loss": 0.3473,
+      "step": 6655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996273890335888e-05,
+      "loss": 0.32,
+      "step": 6660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9962583266180194e-05,
+      "loss": 0.3586,
+      "step": 6665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.9962427304878825e-05,
+      "loss": 0.3452,
+      "step": 6670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.996227101945678e-05,
+      "loss": 0.3305,
+      "step": 6675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9962114409916105e-05,
+      "loss": 0.2937,
+      "step": 6680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996195747625881e-05,
+      "loss": 0.3321,
+      "step": 6685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9961800218486965e-05,
+      "loss": 0.3408,
+      "step": 6690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996164263660258e-05,
+      "loss": 0.3322,
+      "step": 6695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996148473060773e-05,
+      "loss": 0.3353,
+      "step": 6700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996132650050444e-05,
+      "loss": 0.3284,
+      "step": 6705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996116794629478e-05,
+      "loss": 0.3554,
+      "step": 6710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99610090679808e-05,
+      "loss": 0.34,
+      "step": 6715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9960849865564576e-05,
+      "loss": 0.3264,
+      "step": 6720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996069033904816e-05,
+      "loss": 0.3286,
+      "step": 6725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9960530488433634e-05,
+      "loss": 0.3279,
+      "step": 6730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996037031372306e-05,
+      "loss": 0.3283,
+      "step": 6735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9960209814918534e-05,
+      "loss": 0.3508,
+      "step": 6740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.996004899202213e-05,
+      "loss": 0.3734,
+      "step": 6745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995988784503595e-05,
+      "loss": 0.3299,
+      "step": 6750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995972637396206e-05,
+      "loss": 0.3035,
+      "step": 6755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995956457880257e-05,
+      "loss": 0.3312,
+      "step": 6760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99594024595596e-05,
+      "loss": 0.319,
+      "step": 6765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9959240016235233e-05,
+      "loss": 0.3293,
+      "step": 6770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995907724883158e-05,
+      "loss": 0.3427,
+      "step": 6775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995891415735075e-05,
+      "loss": 0.3499,
+      "step": 6780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995875074179488e-05,
+      "loss": 0.355,
+      "step": 6785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995858700216607e-05,
+      "loss": 0.3269,
+      "step": 6790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995842293846647e-05,
+      "loss": 0.3251,
+      "step": 6795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995825855069818e-05,
+      "loss": 0.3315,
+      "step": 6800
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995809383886336e-05,
+      "loss": 0.3505,
+      "step": 6805
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9957928802964135e-05,
+      "loss": 0.3231,
+      "step": 6810
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995776344300266e-05,
+      "loss": 0.3082,
+      "step": 6815
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995759775898107e-05,
+      "loss": 0.3222,
+      "step": 6820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995743175090152e-05,
+      "loss": 0.327,
+      "step": 6825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9957265418766165e-05,
+      "loss": 0.3357,
+      "step": 6830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9957098762577174e-05,
+      "loss": 0.3527,
+      "step": 6835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9956931782336694e-05,
+      "loss": 0.3178,
+      "step": 6840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99567644780469e-05,
+      "loss": 0.3072,
+      "step": 6845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995659684970997e-05,
+      "loss": 0.338,
+      "step": 6850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995642889732808e-05,
+      "loss": 0.3291,
+      "step": 6855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9956260620903406e-05,
+      "loss": 0.3177,
+      "step": 6860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995609202043814e-05,
+      "loss": 0.3261,
+      "step": 6865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9955923095934454e-05,
+      "loss": 0.3423,
+      "step": 6870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995575384739456e-05,
+      "loss": 0.3177,
+      "step": 6875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995558427482065e-05,
+      "loss": 0.3263,
+      "step": 6880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995541437821492e-05,
+      "loss": 0.3622,
+      "step": 6885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995524415757958e-05,
+      "loss": 0.3131,
+      "step": 6890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995507361291685e-05,
+      "loss": 0.3471,
+      "step": 6895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995490274422893e-05,
+      "loss": 0.3,
+      "step": 6900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995473155151804e-05,
+      "loss": 0.339,
+      "step": 6905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995456003478641e-05,
+      "loss": 0.3626,
+      "step": 6910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995438819403626e-05,
+      "loss": 0.2978,
+      "step": 6915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995421602926983e-05,
+      "loss": 0.3304,
+      "step": 6920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9954043540489346e-05,
+      "loss": 0.2999,
+      "step": 6925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995387072769706e-05,
+      "loss": 0.3274,
+      "step": 6930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.99536975908952e-05,
+      "loss": 0.3499,
+      "step": 6935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995352413008603e-05,
+      "loss": 0.337,
+      "step": 6940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995335034527178e-05,
+      "loss": 0.3155,
+      "step": 6945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995317623645474e-05,
+      "loss": 0.3544,
+      "step": 6950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.9953001803637136e-05,
+      "loss": 0.3439,
+      "step": 6955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.995282704682125e-05,
+      "loss": 0.3484,
+      "step": 6960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995265196600935e-05,
+      "loss": 0.3429,
+      "step": 6965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9952476561203706e-05,
+      "loss": 0.3532,
+      "step": 6970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9952300832406605e-05,
+      "loss": 0.3107,
+      "step": 6975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9952124779620324e-05,
+      "loss": 0.3309,
+      "step": 6980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995194840284714e-05,
+      "loss": 0.2985,
+      "step": 6985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995177170208935e-05,
+      "loss": 0.3145,
+      "step": 6990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995159467734925e-05,
+      "loss": 0.3284,
+      "step": 6995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9951417328629136e-05,
+      "loss": 0.3456,
+      "step": 7000
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.3314030170440674,
+      "eval_runtime": 10.5578,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 9.472,
+      "step": 7000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9951239655931314e-05,
+      "loss": 0.3508,
+      "step": 7005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995106165925809e-05,
+      "loss": 0.3357,
+      "step": 7010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9950883338611765e-05,
+      "loss": 0.3471,
+      "step": 7015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9950704693994673e-05,
+      "loss": 0.2958,
+      "step": 7020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995052572540912e-05,
+      "loss": 0.3492,
+      "step": 7025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995034643285743e-05,
+      "loss": 0.3428,
+      "step": 7030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.995016681634194e-05,
+      "loss": 0.3175,
+      "step": 7035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9949986875864976e-05,
+      "loss": 0.3196,
+      "step": 7040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994980661142887e-05,
+      "loss": 0.3415,
+      "step": 7045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9949626023035974e-05,
+      "loss": 0.3353,
+      "step": 7050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994944511068862e-05,
+      "loss": 0.3419,
+      "step": 7055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9949263874389176e-05,
+      "loss": 0.3174,
+      "step": 7060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994908231413997e-05,
+      "loss": 0.3193,
+      "step": 7065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994890042994337e-05,
+      "loss": 0.351,
+      "step": 7070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994871822180175e-05,
+      "loss": 0.3311,
+      "step": 7075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994853568971746e-05,
+      "loss": 0.3302,
+      "step": 7080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994835283369288e-05,
+      "loss": 0.372,
+      "step": 7085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994816965373038e-05,
+      "loss": 0.3315,
+      "step": 7090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994798614983234e-05,
+      "loss": 0.3142,
+      "step": 7095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994780232200113e-05,
+      "loss": 0.3445,
+      "step": 7100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9947618170239154e-05,
+      "loss": 0.3185,
+      "step": 7105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99474336945488e-05,
+      "loss": 0.3178,
+      "step": 7110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9947248894932464e-05,
+      "loss": 0.3449,
+      "step": 7115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994706377139253e-05,
+      "loss": 0.3358,
+      "step": 7120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994687832393142e-05,
+      "loss": 0.3166,
+      "step": 7125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994669255255154e-05,
+      "loss": 0.3177,
+      "step": 7130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994650645725529e-05,
+      "loss": 0.3748,
+      "step": 7135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99463200380451e-05,
+      "loss": 0.3148,
+      "step": 7140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9946133294923384e-05,
+      "loss": 0.3303,
+      "step": 7145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9945946227892566e-05,
+      "loss": 0.3862,
+      "step": 7150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9945758836955075e-05,
+      "loss": 0.334,
+      "step": 7155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9945571122113344e-05,
+      "loss": 0.3192,
+      "step": 7160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994538308336981e-05,
+      "loss": 0.3359,
+      "step": 7165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9945194720726926e-05,
+      "loss": 0.3084,
+      "step": 7170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9945006034187124e-05,
+      "loss": 0.3323,
+      "step": 7175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9944817023752853e-05,
+      "loss": 0.3196,
+      "step": 7180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9944627689426574e-05,
+      "loss": 0.3358,
+      "step": 7185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994443803121075e-05,
+      "loss": 0.3596,
+      "step": 7190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994424804910783e-05,
+      "loss": 0.3423,
+      "step": 7195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9944057743120284e-05,
+      "loss": 0.3223,
+      "step": 7200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.99438671132506e-05,
+      "loss": 0.3438,
+      "step": 7205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994367615950123e-05,
+      "loss": 0.2991,
+      "step": 7210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994348488187467e-05,
+      "loss": 0.3617,
+      "step": 7215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9943293280373395e-05,
+      "loss": 0.3359,
+      "step": 7220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994310135499989e-05,
+      "loss": 0.3333,
+      "step": 7225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994290910575666e-05,
+      "loss": 0.3275,
+      "step": 7230
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994271653264619e-05,
+      "loss": 0.3242,
+      "step": 7235
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.994252363567098e-05,
+      "loss": 0.3308,
+      "step": 7240
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.9942330414833546e-05,
+      "loss": 0.3246,
+      "step": 7245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9942136870136383e-05,
+      "loss": 0.3392,
+      "step": 7250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9941943001582015e-05,
+      "loss": 0.331,
+      "step": 7255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994174880917295e-05,
+      "loss": 0.3131,
+      "step": 7260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9941554292911715e-05,
+      "loss": 0.3576,
+      "step": 7265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9941359452800826e-05,
+      "loss": 0.3267,
+      "step": 7270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994116428884283e-05,
+      "loss": 0.3251,
+      "step": 7275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994096880104026e-05,
+      "loss": 0.3205,
+      "step": 7280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994077298939564e-05,
+      "loss": 0.3455,
+      "step": 7285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994057685391152e-05,
+      "loss": 0.3158,
+      "step": 7290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994038039459045e-05,
+      "loss": 0.3288,
+      "step": 7295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.994018361143497e-05,
+      "loss": 0.3069,
+      "step": 7300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993998650444764e-05,
+      "loss": 0.3741,
+      "step": 7305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9939789073631025e-05,
+      "loss": 0.3298,
+      "step": 7310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993959131898769e-05,
+      "loss": 0.3459,
+      "step": 7315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993939324052019e-05,
+      "loss": 0.3345,
+      "step": 7320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9939194838231104e-05,
+      "loss": 0.3655,
+      "step": 7325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993899611212302e-05,
+      "loss": 0.3267,
+      "step": 7330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993879706219849e-05,
+      "loss": 0.3244,
+      "step": 7335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993859768846013e-05,
+      "loss": 0.3159,
+      "step": 7340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.99383979909105e-05,
+      "loss": 0.3231,
+      "step": 7345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993819796955221e-05,
+      "loss": 0.3526,
+      "step": 7350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9937997624387864e-05,
+      "loss": 0.329,
+      "step": 7355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993779695542005e-05,
+      "loss": 0.3234,
+      "step": 7360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993759596265138e-05,
+      "loss": 0.3473,
+      "step": 7365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9937394646084454e-05,
+      "loss": 0.3387,
+      "step": 7370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993719300572189e-05,
+      "loss": 0.3333,
+      "step": 7375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993699104156632e-05,
+      "loss": 0.3357,
+      "step": 7380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993678875362036e-05,
+      "loss": 0.338,
+      "step": 7385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993658614188662e-05,
+      "loss": 0.3469,
+      "step": 7390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9936383206367745e-05,
+      "loss": 0.3189,
+      "step": 7395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993617994706637e-05,
+      "loss": 0.3178,
+      "step": 7400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9935976363985126e-05,
+      "loss": 0.3372,
+      "step": 7405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993577245712666e-05,
+      "loss": 0.3349,
+      "step": 7410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993556822649363e-05,
+      "loss": 0.3583,
+      "step": 7415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993536367208868e-05,
+      "loss": 0.318,
+      "step": 7420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993515879391447e-05,
+      "loss": 0.3181,
+      "step": 7425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993495359197364e-05,
+      "loss": 0.3264,
+      "step": 7430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993474806626889e-05,
+      "loss": 0.2904,
+      "step": 7435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9934542216802856e-05,
+      "loss": 0.3285,
+      "step": 7440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9934336043578225e-05,
+      "loss": 0.3649,
+      "step": 7445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993412954659767e-05,
+      "loss": 0.3389,
+      "step": 7450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993392272586388e-05,
+      "loss": 0.3359,
+      "step": 7455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9933715581379536e-05,
+      "loss": 0.35,
+      "step": 7460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993350811314733e-05,
+      "loss": 0.3076,
+      "step": 7465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993330032116995e-05,
+      "loss": 0.353,
+      "step": 7470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9933092205450094e-05,
+      "loss": 0.3349,
+      "step": 7475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9932883765990466e-05,
+      "loss": 0.3446,
+      "step": 7480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993267500279378e-05,
+      "loss": 0.3393,
+      "step": 7485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993246591586274e-05,
+      "loss": 0.3423,
+      "step": 7490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993225650520006e-05,
+      "loss": 0.3387,
+      "step": 7495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9932046770808454e-05,
+      "loss": 0.3369,
+      "step": 7500
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 0.3308403789997101,
+      "eval_runtime": 10.5623,
+      "eval_samples_per_second": 9.468,
+      "eval_steps_per_second": 9.468,
+      "step": 7500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993183671269066e-05,
+      "loss": 0.3551,
+      "step": 7505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993162633084939e-05,
+      "loss": 0.3412,
+      "step": 7510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993141562528739e-05,
+      "loss": 0.3525,
+      "step": 7515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9931204596007386e-05,
+      "loss": 0.314,
+      "step": 7520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.9930993243012114e-05,
+      "loss": 0.3427,
+      "step": 7525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.993078156630433e-05,
+      "loss": 0.3562,
+      "step": 7530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9930569565886775e-05,
+      "loss": 0.3455,
+      "step": 7535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.993035724176221e-05,
+      "loss": 0.3324,
+      "step": 7540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.993014459393338e-05,
+      "loss": 0.3426,
+      "step": 7545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992993162240306e-05,
+      "loss": 0.328,
+      "step": 7550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9929718327174004e-05,
+      "loss": 0.3342,
+      "step": 7555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992950470824898e-05,
+      "loss": 0.3309,
+      "step": 7560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9929290765630766e-05,
+      "loss": 0.3365,
+      "step": 7565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9929076499322145e-05,
+      "loss": 0.342,
+      "step": 7570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9928861909325894e-05,
+      "loss": 0.3079,
+      "step": 7575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9928646995644804e-05,
+      "loss": 0.3414,
+      "step": 7580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992843175828165e-05,
+      "loss": 0.3426,
+      "step": 7585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992821619723924e-05,
+      "loss": 0.3287,
+      "step": 7590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992800031252038e-05,
+      "loss": 0.3117,
+      "step": 7595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9927784104127864e-05,
+      "loss": 0.3147,
+      "step": 7600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992756757206449e-05,
+      "loss": 0.3535,
+      "step": 7605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992735071633308e-05,
+      "loss": 0.3252,
+      "step": 7610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992713353693646e-05,
+      "loss": 0.3412,
+      "step": 7615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992691603387742e-05,
+      "loss": 0.3453,
+      "step": 7620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992669820715882e-05,
+      "loss": 0.3312,
+      "step": 7625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9926480056783465e-05,
+      "loss": 0.3548,
+      "step": 7630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99262615827542e-05,
+      "loss": 0.3218,
+      "step": 7635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992604278507385e-05,
+      "loss": 0.3271,
+      "step": 7640
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9925823663745255e-05,
+      "loss": 0.3429,
+      "step": 7645
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992560421877127e-05,
+      "loss": 0.3242,
+      "step": 7650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992538445015474e-05,
+      "loss": 0.3192,
+      "step": 7655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992516435789852e-05,
+      "loss": 0.3249,
+      "step": 7660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992494394200548e-05,
+      "loss": 0.3245,
+      "step": 7665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992472320247845e-05,
+      "loss": 0.3199,
+      "step": 7670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992450213932033e-05,
+      "loss": 0.3396,
+      "step": 7675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9924280752533965e-05,
+      "loss": 0.3112,
+      "step": 7680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992405904212224e-05,
+      "loss": 0.3247,
+      "step": 7685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9923837008088046e-05,
+      "loss": 0.3239,
+      "step": 7690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9923614650434245e-05,
+      "loss": 0.3494,
+      "step": 7695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992339196916373e-05,
+      "loss": 0.3242,
+      "step": 7700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99231689642794e-05,
+      "loss": 0.3402,
+      "step": 7705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992294563578415e-05,
+      "loss": 0.3341,
+      "step": 7710
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992272198368086e-05,
+      "loss": 0.3508,
+      "step": 7715
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992249800797247e-05,
+      "loss": 0.3563,
+      "step": 7720
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992227370866186e-05,
+      "loss": 0.3281,
+      "step": 7725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992204908575194e-05,
+      "loss": 0.3573,
+      "step": 7730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9921824139245656e-05,
+      "loss": 0.3089,
+      "step": 7735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992159886914589e-05,
+      "loss": 0.3237,
+      "step": 7740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9921373275455594e-05,
+      "loss": 0.3421,
+      "step": 7745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99211473581777e-05,
+      "loss": 0.3322,
+      "step": 7750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992092111731512e-05,
+      "loss": 0.3562,
+      "step": 7755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992069455287081e-05,
+      "loss": 0.3329,
+      "step": 7760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.99204676648477e-05,
+      "loss": 0.3719,
+      "step": 7765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992024045324874e-05,
+      "loss": 0.3369,
+      "step": 7770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.992001291807688e-05,
+      "loss": 0.3362,
+      "step": 7775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991978505933508e-05,
+      "loss": 0.3308,
+      "step": 7780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9919556877026284e-05,
+      "loss": 0.3392,
+      "step": 7785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991932837115347e-05,
+      "loss": 0.338,
+      "step": 7790
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991909954171961e-05,
+      "loss": 0.3218,
+      "step": 7795
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991887038872766e-05,
+      "loss": 0.3144,
+      "step": 7800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991864091218059e-05,
+      "loss": 0.3482,
+      "step": 7805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.9918411112081395e-05,
+      "loss": 0.3249,
+      "step": 7810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.991818098843305e-05,
+      "loss": 0.3433,
+      "step": 7815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991795054123856e-05,
+      "loss": 0.3122,
+      "step": 7820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991771977050089e-05,
+      "loss": 0.3412,
+      "step": 7825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991748867622306e-05,
+      "loss": 0.3673,
+      "step": 7830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991725725840805e-05,
+      "loss": 0.3109,
+      "step": 7835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991702551705888e-05,
+      "loss": 0.3558,
+      "step": 7840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9916793452178554e-05,
+      "loss": 0.3088,
+      "step": 7845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9916561063770084e-05,
+      "loss": 0.3375,
+      "step": 7850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991632835183649e-05,
+      "loss": 0.3374,
+      "step": 7855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.99160953163808e-05,
+      "loss": 0.3386,
+      "step": 7860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991586195740603e-05,
+      "loss": 0.3391,
+      "step": 7865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991562827491521e-05,
+      "loss": 0.2986,
+      "step": 7870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9915394268911375e-05,
+      "loss": 0.3754,
+      "step": 7875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9915159939397574e-05,
+      "loss": 0.315,
+      "step": 7880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991492528637683e-05,
+      "loss": 0.3335,
+      "step": 7885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9914690309852205e-05,
+      "loss": 0.3211,
+      "step": 7890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991445500982675e-05,
+      "loss": 0.3403,
+      "step": 7895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9914219386303513e-05,
+      "loss": 0.3454,
+      "step": 7900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991398343928555e-05,
+      "loss": 0.3164,
+      "step": 7905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9913747168775934e-05,
+      "loss": 0.3341,
+      "step": 7910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991351057477774e-05,
+      "loss": 0.3248,
+      "step": 7915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9913273657294025e-05,
+      "loss": 0.3214,
+      "step": 7920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991303641632786e-05,
+      "loss": 0.3329,
+      "step": 7925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991279885188235e-05,
+      "loss": 0.3253,
+      "step": 7930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991256096396055e-05,
+      "loss": 0.3295,
+      "step": 7935
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991232275256558e-05,
+      "loss": 0.3242,
+      "step": 7940
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991208421770051e-05,
+      "loss": 0.3328,
+      "step": 7945
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9911845359368434e-05,
+      "loss": 0.3206,
+      "step": 7950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9911606177572476e-05,
+      "loss": 0.3497,
+      "step": 7955
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991136667231572e-05,
+      "loss": 0.3293,
+      "step": 7960
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991112684360129e-05,
+      "loss": 0.3348,
+      "step": 7965
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.99108866914323e-05,
+      "loss": 0.3222,
+      "step": 7970
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9910646215811854e-05,
+      "loss": 0.3057,
+      "step": 7975
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9910405416743086e-05,
+      "loss": 0.3665,
+      "step": 7980
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.991016429422912e-05,
+      "loss": 0.329,
+      "step": 7985
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990992284827309e-05,
+      "loss": 0.3296,
+      "step": 7990
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9909681078878136e-05,
+      "loss": 0.3494,
+      "step": 7995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9909438986047375e-05,
+      "loss": 0.3159,
+      "step": 8000
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 0.3288015127182007,
+      "eval_runtime": 10.5609,
+      "eval_samples_per_second": 9.469,
+      "eval_steps_per_second": 9.469,
+      "step": 8000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9909196569783975e-05,
+      "loss": 0.3127,
+      "step": 8005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9908953830091066e-05,
+      "loss": 0.3201,
+      "step": 8010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990871076697181e-05,
+      "loss": 0.3433,
+      "step": 8015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990846738042936e-05,
+      "loss": 0.3307,
+      "step": 8020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990822367046688e-05,
+      "loss": 0.3218,
+      "step": 8025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990797963708752e-05,
+      "loss": 0.3419,
+      "step": 8030
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990773528029446e-05,
+      "loss": 0.338,
+      "step": 8035
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990749060009089e-05,
+      "loss": 0.3297,
+      "step": 8040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9907245596479956e-05,
+      "loss": 0.3106,
+      "step": 8045
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990700026946485e-05,
+      "loss": 0.3051,
+      "step": 8050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9906754619048767e-05,
+      "loss": 0.3461,
+      "step": 8055
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9906508645234885e-05,
+      "loss": 0.323,
+      "step": 8060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990626234802641e-05,
+      "loss": 0.3196,
+      "step": 8065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990601572742652e-05,
+      "loss": 0.3257,
+      "step": 8070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990576878343844e-05,
+      "loss": 0.3501,
+      "step": 8075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.9905521516065356e-05,
+      "loss": 0.3417,
+      "step": 8080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.99052739253105e-05,
+      "loss": 0.3415,
+      "step": 8085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990502601117707e-05,
+      "loss": 0.3378,
+      "step": 8090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990477777366829e-05,
+      "loss": 0.3518,
+      "step": 8095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.990452921278738e-05,
+      "loss": 0.35,
+      "step": 8100
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990428032853757e-05,
+      "loss": 0.3323,
+      "step": 8105
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.99040311209221e-05,
+      "loss": 0.3318,
+      "step": 8110
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9903781589944193e-05,
+      "loss": 0.3153,
+      "step": 8115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.99035317356071e-05,
+      "loss": 0.312,
+      "step": 8120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990328155791405e-05,
+      "loss": 0.3254,
+      "step": 8125
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9903031056868306e-05,
+      "loss": 0.3438,
+      "step": 8130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990278023247311e-05,
+      "loss": 0.331,
+      "step": 8135
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990252908473174e-05,
+      "loss": 0.3616,
+      "step": 8140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990227761364743e-05,
+      "loss": 0.3103,
+      "step": 8145
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990202581922345e-05,
+      "loss": 0.3174,
+      "step": 8150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990177370146308e-05,
+      "loss": 0.3259,
+      "step": 8155
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.99015212603696e-05,
+      "loss": 0.3287,
+      "step": 8160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990126849594627e-05,
+      "loss": 0.3591,
+      "step": 8165
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9901015408196375e-05,
+      "loss": 0.3171,
+      "step": 8170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9900761997123216e-05,
+      "loss": 0.3008,
+      "step": 8175
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990050826273006e-05,
+      "loss": 0.3335,
+      "step": 8180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.990025420502022e-05,
+      "loss": 0.3139,
+      "step": 8185
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989999982399699e-05,
+      "loss": 0.3203,
+      "step": 8190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9899745119663674e-05,
+      "loss": 0.3375,
+      "step": 8195
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989949009202357e-05,
+      "loss": 0.362,
+      "step": 8200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9899234741080006e-05,
+      "loss": 0.313,
+      "step": 8205
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989897906683628e-05,
+      "loss": 0.3655,
+      "step": 8210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9898723069295725e-05,
+      "loss": 0.349,
+      "step": 8215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9898466748461653e-05,
+      "loss": 0.3174,
+      "step": 8220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989821010433741e-05,
+      "loss": 0.3315,
+      "step": 8225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.98979531369263e-05,
+      "loss": 0.3352,
+      "step": 8230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9897695846231696e-05,
+      "loss": 0.2966,
+      "step": 8235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989743823225691e-05,
+      "loss": 0.3223,
+      "step": 8240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.98971802950053e-05,
+      "loss": 0.3506,
+      "step": 8245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989692203448021e-05,
+      "loss": 0.334,
+      "step": 8250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9896663450684986e-05,
+      "loss": 0.3311,
+      "step": 8255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989640454362301e-05,
+      "loss": 0.3512,
+      "step": 8260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9896145313297614e-05,
+      "loss": 0.3332,
+      "step": 8265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989588575971219e-05,
+      "loss": 0.3331,
+      "step": 8270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989562588287009e-05,
+      "loss": 0.3218,
+      "step": 8275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.98953656827747e-05,
+      "loss": 0.3283,
+      "step": 8280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989510515942939e-05,
+      "loss": 0.3352,
+      "step": 8285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989484431283755e-05,
+      "loss": 0.3431,
+      "step": 8290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989458314300256e-05,
+      "loss": 0.3274,
+      "step": 8295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989432164992781e-05,
+      "loss": 0.291,
+      "step": 8300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.98940598336167e-05,
+      "loss": 0.3318,
+      "step": 8305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989379769407263e-05,
+      "loss": 0.3251,
+      "step": 8310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9893535231299004e-05,
+      "loss": 0.334,
+      "step": 8315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9893272445299236e-05,
+      "loss": 0.3066,
+      "step": 8320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9893009336076716e-05,
+      "loss": 0.3556,
+      "step": 8325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989274590363489e-05,
+      "loss": 0.3614,
+      "step": 8330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989248214797715e-05,
+      "loss": 0.3364,
+      "step": 8335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9892218069106944e-05,
+      "loss": 0.3103,
+      "step": 8340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989195366702768e-05,
+      "loss": 0.3487,
+      "step": 8345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989168894174281e-05,
+      "loss": 0.3232,
+      "step": 8350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989142389325576e-05,
+      "loss": 0.3302,
+      "step": 8355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989115852156998e-05,
+      "loss": 0.3242,
+      "step": 8360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989089282668891e-05,
+      "loss": 0.3235,
+      "step": 8365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.989062680861599e-05,
+      "loss": 0.3434,
+      "step": 8370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9890360467354694e-05,
+      "loss": 0.3097,
+      "step": 8375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.9890093802908466e-05,
+      "loss": 0.2919,
+      "step": 8380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988982681528078e-05,
+      "loss": 0.3215,
+      "step": 8385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988955950447509e-05,
+      "loss": 0.3283,
+      "step": 8390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988929187049488e-05,
+      "loss": 0.3286,
+      "step": 8395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9889023913343605e-05,
+      "loss": 0.3343,
+      "step": 8400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988875563302477e-05,
+      "loss": 0.3434,
+      "step": 8405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988848702954184e-05,
+      "loss": 0.3349,
+      "step": 8410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9888218102898305e-05,
+      "loss": 0.3262,
+      "step": 8415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988794885309766e-05,
+      "loss": 0.3224,
+      "step": 8420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9887679280143404e-05,
+      "loss": 0.3409,
+      "step": 8425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988740938403903e-05,
+      "loss": 0.3289,
+      "step": 8430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9887139164788056e-05,
+      "loss": 0.331,
+      "step": 8435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988686862239397e-05,
+      "loss": 0.3153,
+      "step": 8440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.98865977568603e-05,
+      "loss": 0.3435,
+      "step": 8445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988632656819057e-05,
+      "loss": 0.3379,
+      "step": 8450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988605505638827e-05,
+      "loss": 0.3532,
+      "step": 8455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9885783221456963e-05,
+      "loss": 0.3551,
+      "step": 8460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988551106340015e-05,
+      "loss": 0.3478,
+      "step": 8465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988523858222138e-05,
+      "loss": 0.3282,
+      "step": 8470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9884965777924185e-05,
+      "loss": 0.3283,
+      "step": 8475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988469265051211e-05,
+      "loss": 0.3209,
+      "step": 8480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9884419199988705e-05,
+      "loss": 0.3086,
+      "step": 8485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988414542635751e-05,
+      "loss": 0.2962,
+      "step": 8490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988387132962209e-05,
+      "loss": 0.314,
+      "step": 8495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9883596909786e-05,
+      "loss": 0.351,
+      "step": 8500
+    },
+    {
+      "epoch": 0.3,
+      "eval_loss": 0.32749566435813904,
+      "eval_runtime": 10.5451,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 8500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.98833221668528e-05,
+      "loss": 0.3367,
+      "step": 8505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9883047100826064e-05,
+      "loss": 0.3108,
+      "step": 8510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988277171170936e-05,
+      "loss": 0.3493,
+      "step": 8515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988249599950626e-05,
+      "loss": 0.3309,
+      "step": 8520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988221996422035e-05,
+      "loss": 0.3633,
+      "step": 8525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9881943605855215e-05,
+      "loss": 0.3119,
+      "step": 8530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9881666924414436e-05,
+      "loss": 0.3409,
+      "step": 8535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988138991990161e-05,
+      "loss": 0.3129,
+      "step": 8540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9881112592320333e-05,
+      "loss": 0.3077,
+      "step": 8545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988083494167421e-05,
+      "loss": 0.3065,
+      "step": 8550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988055696796683e-05,
+      "loss": 0.3331,
+      "step": 8555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.988027867120183e-05,
+      "loss": 0.3345,
+      "step": 8560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9880000051382804e-05,
+      "loss": 0.315,
+      "step": 8565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9879721108513364e-05,
+      "loss": 0.331,
+      "step": 8570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987944184259715e-05,
+      "loss": 0.3345,
+      "step": 8575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987916225363778e-05,
+      "loss": 0.3136,
+      "step": 8580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9878882341638886e-05,
+      "loss": 0.3176,
+      "step": 8585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9878602106604097e-05,
+      "loss": 0.3392,
+      "step": 8590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987832154853706e-05,
+      "loss": 0.3551,
+      "step": 8595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9878040667441405e-05,
+      "loss": 0.3136,
+      "step": 8600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9877759463320795e-05,
+      "loss": 0.3382,
+      "step": 8605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9877477936178864e-05,
+      "loss": 0.3103,
+      "step": 8610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9877196086019286e-05,
+      "loss": 0.321,
+      "step": 8615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9876913912845705e-05,
+      "loss": 0.3064,
+      "step": 8620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9876631416661796e-05,
+      "loss": 0.3302,
+      "step": 8625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9876348597471224e-05,
+      "loss": 0.3565,
+      "step": 8630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9876065455277654e-05,
+      "loss": 0.3379,
+      "step": 8635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9875781990084766e-05,
+      "loss": 0.3173,
+      "step": 8640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9875498201896254e-05,
+      "loss": 0.3347,
+      "step": 8645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987521409071578e-05,
+      "loss": 0.3462,
+      "step": 8650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.9874929656547054e-05,
+      "loss": 0.3155,
+      "step": 8655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987464489939375e-05,
+      "loss": 0.3135,
+      "step": 8660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.987435981925958e-05,
+      "loss": 0.3295,
+      "step": 8665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987407441614824e-05,
+      "loss": 0.3148,
+      "step": 8670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987378869006344e-05,
+      "loss": 0.3619,
+      "step": 8675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987350264100889e-05,
+      "loss": 0.3541,
+      "step": 8680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987321626898829e-05,
+      "loss": 0.3266,
+      "step": 8685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987292957400537e-05,
+      "loss": 0.3196,
+      "step": 8690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987264255606385e-05,
+      "loss": 0.3261,
+      "step": 8695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987235521516747e-05,
+      "loss": 0.3328,
+      "step": 8700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987206755131993e-05,
+      "loss": 0.3345,
+      "step": 8705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9871779564525e-05,
+      "loss": 0.3279,
+      "step": 8710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987149125478639e-05,
+      "loss": 0.3352,
+      "step": 8715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987120262210787e-05,
+      "loss": 0.3469,
+      "step": 8720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9870913666493166e-05,
+      "loss": 0.3329,
+      "step": 8725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9870624387946046e-05,
+      "loss": 0.3516,
+      "step": 8730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9870334786470256e-05,
+      "loss": 0.3184,
+      "step": 8735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.987004486206956e-05,
+      "loss": 0.3499,
+      "step": 8740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986975461474772e-05,
+      "loss": 0.2988,
+      "step": 8745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986946404450851e-05,
+      "loss": 0.349,
+      "step": 8750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986917315135569e-05,
+      "loss": 0.3331,
+      "step": 8755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986888193529305e-05,
+      "loss": 0.3303,
+      "step": 8760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986859039632436e-05,
+      "loss": 0.3464,
+      "step": 8765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986829853445343e-05,
+      "loss": 0.3315,
+      "step": 8770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986800634968402e-05,
+      "loss": 0.3099,
+      "step": 8775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9867713842019934e-05,
+      "loss": 0.3392,
+      "step": 8780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9867421011464984e-05,
+      "loss": 0.3358,
+      "step": 8785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986712785802295e-05,
+      "loss": 0.3171,
+      "step": 8790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986683438169766e-05,
+      "loss": 0.3225,
+      "step": 8795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9866540582492907e-05,
+      "loss": 0.2957,
+      "step": 8800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98662464604125e-05,
+      "loss": 0.3151,
+      "step": 8805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9865952015460285e-05,
+      "loss": 0.3267,
+      "step": 8810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9865657247640065e-05,
+      "loss": 0.327,
+      "step": 8815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9865362156955674e-05,
+      "loss": 0.3355,
+      "step": 8820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9865066743410946e-05,
+      "loss": 0.3326,
+      "step": 8825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986477100700972e-05,
+      "loss": 0.3034,
+      "step": 8830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9864474947755816e-05,
+      "loss": 0.3149,
+      "step": 8835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9864178565653084e-05,
+      "loss": 0.3197,
+      "step": 8840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98638818607054e-05,
+      "loss": 0.3333,
+      "step": 8845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986358483291658e-05,
+      "loss": 0.3281,
+      "step": 8850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.98632874822905e-05,
+      "loss": 0.3363,
+      "step": 8855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986298980883102e-05,
+      "loss": 0.3327,
+      "step": 8860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9862691812542e-05,
+      "loss": 0.3381,
+      "step": 8865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986239349342732e-05,
+      "loss": 0.3479,
+      "step": 8870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9862094851490836e-05,
+      "loss": 0.3281,
+      "step": 8875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986179588673644e-05,
+      "loss": 0.3317,
+      "step": 8880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9861496599168006e-05,
+      "loss": 0.3445,
+      "step": 8885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9861196988789425e-05,
+      "loss": 0.3412,
+      "step": 8890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986089705560458e-05,
+      "loss": 0.3177,
+      "step": 8895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9860596799617375e-05,
+      "loss": 0.3395,
+      "step": 8900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.986029622083171e-05,
+      "loss": 0.3146,
+      "step": 8905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9859995319251475e-05,
+      "loss": 0.3429,
+      "step": 8910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985969409488059e-05,
+      "loss": 0.3353,
+      "step": 8915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985939254772295e-05,
+      "loss": 0.325,
+      "step": 8920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9859090677782494e-05,
+      "loss": 0.3591,
+      "step": 8925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985878848506312e-05,
+      "loss": 0.3047,
+      "step": 8930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9858485969568755e-05,
+      "loss": 0.3302,
+      "step": 8935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9858183131303344e-05,
+      "loss": 0.2988,
+      "step": 8940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.9857879970270805e-05,
+      "loss": 0.3423,
+      "step": 8945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.985757648647507e-05,
+      "loss": 0.3119,
+      "step": 8950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985727267992009e-05,
+      "loss": 0.3219,
+      "step": 8955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98569685506098e-05,
+      "loss": 0.3093,
+      "step": 8960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9856664098548165e-05,
+      "loss": 0.3414,
+      "step": 8965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9856359323739124e-05,
+      "loss": 0.3321,
+      "step": 8970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985605422618663e-05,
+      "loss": 0.3158,
+      "step": 8975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985574880589467e-05,
+      "loss": 0.3373,
+      "step": 8980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985544306286717e-05,
+      "loss": 0.341,
+      "step": 8985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9855136997108146e-05,
+      "loss": 0.334,
+      "step": 8990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985483060862153e-05,
+      "loss": 0.3127,
+      "step": 8995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985452389741132e-05,
+      "loss": 0.3007,
+      "step": 9000
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 0.32597243785858154,
+      "eval_runtime": 10.5417,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 9000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9854216863481505e-05,
+      "loss": 0.357,
+      "step": 9005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985390950683606e-05,
+      "loss": 0.3185,
+      "step": 9010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985360182747898e-05,
+      "loss": 0.3264,
+      "step": 9015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985329382541426e-05,
+      "loss": 0.333,
+      "step": 9020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.98529855006459e-05,
+      "loss": 0.3089,
+      "step": 9025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985267685317789e-05,
+      "loss": 0.3267,
+      "step": 9030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985236788301427e-05,
+      "loss": 0.3334,
+      "step": 9035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985205859015902e-05,
+      "loss": 0.3135,
+      "step": 9040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985174897461616e-05,
+      "loss": 0.338,
+      "step": 9045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985143903638973e-05,
+      "loss": 0.3489,
+      "step": 9050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985112877548373e-05,
+      "loss": 0.3145,
+      "step": 9055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9850818191902216e-05,
+      "loss": 0.3318,
+      "step": 9060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9850507285649196e-05,
+      "loss": 0.316,
+      "step": 9065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.985019605672871e-05,
+      "loss": 0.3234,
+      "step": 9070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984988450514482e-05,
+      "loss": 0.3117,
+      "step": 9075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984957263090154e-05,
+      "loss": 0.3166,
+      "step": 9080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984926043400295e-05,
+      "loss": 0.3375,
+      "step": 9085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984894791445308e-05,
+      "loss": 0.32,
+      "step": 9090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984863507225601e-05,
+      "loss": 0.3319,
+      "step": 9095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984832190741577e-05,
+      "loss": 0.3137,
+      "step": 9100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984800841993647e-05,
+      "loss": 0.3157,
+      "step": 9105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9847694609822133e-05,
+      "loss": 0.3358,
+      "step": 9110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9847380477076866e-05,
+      "loss": 0.3316,
+      "step": 9115
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984706602170474e-05,
+      "loss": 0.3314,
+      "step": 9120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9846751243709835e-05,
+      "loss": 0.3301,
+      "step": 9125
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9846436143096245e-05,
+      "loss": 0.3047,
+      "step": 9130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9846120719868045e-05,
+      "loss": 0.324,
+      "step": 9135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984580497402935e-05,
+      "loss": 0.3417,
+      "step": 9140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9845488905584245e-05,
+      "loss": 0.3142,
+      "step": 9145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9845172514536844e-05,
+      "loss": 0.3444,
+      "step": 9150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984485580089125e-05,
+      "loss": 0.3562,
+      "step": 9155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9844538764651584e-05,
+      "loss": 0.3122,
+      "step": 9160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9844221405821946e-05,
+      "loss": 0.3444,
+      "step": 9165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984390372440647e-05,
+      "loss": 0.2972,
+      "step": 9170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984358572040928e-05,
+      "loss": 0.3356,
+      "step": 9175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9843267393834495e-05,
+      "loss": 0.3507,
+      "step": 9180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9842948744686256e-05,
+      "loss": 0.3003,
+      "step": 9185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984262977296871e-05,
+      "loss": 0.3545,
+      "step": 9190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9842310478685975e-05,
+      "loss": 0.3167,
+      "step": 9195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984199086184222e-05,
+      "loss": 0.3063,
+      "step": 9200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984167092244158e-05,
+      "loss": 0.3186,
+      "step": 9205
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9841350660488215e-05,
+      "loss": 0.3231,
+      "step": 9210
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9841030075986284e-05,
+      "loss": 0.3199,
+      "step": 9215
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9840709168939945e-05,
+      "loss": 0.3169,
+      "step": 9220
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9840387939353375e-05,
+      "loss": 0.3304,
+      "step": 9225
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.984006638723073e-05,
+      "loss": 0.3147,
+      "step": 9230
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.9839744512576205e-05,
+      "loss": 0.329,
+      "step": 9235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9839422315393954e-05,
+      "loss": 0.3072,
+      "step": 9240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983909979568818e-05,
+      "loss": 0.3553,
+      "step": 9245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983877695346306e-05,
+      "loss": 0.332,
+      "step": 9250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983845378872279e-05,
+      "loss": 0.3178,
+      "step": 9255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983813030147158e-05,
+      "loss": 0.328,
+      "step": 9260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9837806491713606e-05,
+      "loss": 0.3063,
+      "step": 9265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983748235945309e-05,
+      "loss": 0.3263,
+      "step": 9270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9837157904694234e-05,
+      "loss": 0.3139,
+      "step": 9275
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983683312744124e-05,
+      "loss": 0.3351,
+      "step": 9280
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983650802769834e-05,
+      "loss": 0.3058,
+      "step": 9285
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983618260546976e-05,
+      "loss": 0.3224,
+      "step": 9290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9835856860759714e-05,
+      "loss": 0.3122,
+      "step": 9295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9835530793572425e-05,
+      "loss": 0.2956,
+      "step": 9300
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983520440391214e-05,
+      "loss": 0.3209,
+      "step": 9305
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9834877691783094e-05,
+      "loss": 0.3261,
+      "step": 9310
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983455065718953e-05,
+      "loss": 0.3342,
+      "step": 9315
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983422330013569e-05,
+      "loss": 0.3251,
+      "step": 9320
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983389562062583e-05,
+      "loss": 0.3228,
+      "step": 9325
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.98335676186642e-05,
+      "loss": 0.3524,
+      "step": 9330
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9833239294255054e-05,
+      "loss": 0.342,
+      "step": 9335
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9832910647402666e-05,
+      "loss": 0.3058,
+      "step": 9340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9832581678111304e-05,
+      "loss": 0.3527,
+      "step": 9345
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983225238638523e-05,
+      "loss": 0.3186,
+      "step": 9350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983192277222872e-05,
+      "loss": 0.3057,
+      "step": 9355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9831592835646064e-05,
+      "loss": 0.3024,
+      "step": 9360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983126257664153e-05,
+      "loss": 0.3142,
+      "step": 9365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9830931995219426e-05,
+      "loss": 0.3291,
+      "step": 9370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.983060109138403e-05,
+      "loss": 0.3053,
+      "step": 9375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9830269865139643e-05,
+      "loss": 0.34,
+      "step": 9380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9829938316490574e-05,
+      "loss": 0.3537,
+      "step": 9385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9829606445441106e-05,
+      "loss": 0.3217,
+      "step": 9390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982927425199557e-05,
+      "loss": 0.2932,
+      "step": 9395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982894173615827e-05,
+      "loss": 0.3026,
+      "step": 9400
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982860889793352e-05,
+      "loss": 0.3515,
+      "step": 9405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982827573732565e-05,
+      "loss": 0.3247,
+      "step": 9410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9827942254338975e-05,
+      "loss": 0.3391,
+      "step": 9415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982760844897785e-05,
+      "loss": 0.3342,
+      "step": 9420
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9827274321246575e-05,
+      "loss": 0.3341,
+      "step": 9425
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.98269398711495e-05,
+      "loss": 0.3457,
+      "step": 9430
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982660509869099e-05,
+      "loss": 0.3443,
+      "step": 9435
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9826270003875364e-05,
+      "loss": 0.3341,
+      "step": 9440
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9825934586706986e-05,
+      "loss": 0.309,
+      "step": 9445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9825598847190205e-05,
+      "loss": 0.3391,
+      "step": 9450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9825262785329384e-05,
+      "loss": 0.3325,
+      "step": 9455
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9824926401128894e-05,
+      "loss": 0.2993,
+      "step": 9460
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9824589694593086e-05,
+      "loss": 0.3034,
+      "step": 9465
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982425266572634e-05,
+      "loss": 0.3118,
+      "step": 9470
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982391531453304e-05,
+      "loss": 0.3312,
+      "step": 9475
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982357764101755e-05,
+      "loss": 0.3397,
+      "step": 9480
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982323964518427e-05,
+      "loss": 0.3235,
+      "step": 9485
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982290132703759e-05,
+      "loss": 0.3259,
+      "step": 9490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.9822562686581884e-05,
+      "loss": 0.336,
+      "step": 9495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982222372382156e-05,
+      "loss": 0.3457,
+      "step": 9500
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 0.32349956035614014,
+      "eval_runtime": 10.5421,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 9500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982188443876102e-05,
+      "loss": 0.3081,
+      "step": 9505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982154483140468e-05,
+      "loss": 0.3298,
+      "step": 9510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982120490175692e-05,
+      "loss": 0.3431,
+      "step": 9515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.982086464982219e-05,
+      "loss": 0.303,
+      "step": 9520
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.982052407560488e-05,
+      "loss": 0.3386,
+      "step": 9525
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.982018317910943e-05,
+      "loss": 0.3359,
+      "step": 9530
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9819841960340246e-05,
+      "loss": 0.3161,
+      "step": 9535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981950041930178e-05,
+      "loss": 0.3158,
+      "step": 9540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9819158555998455e-05,
+      "loss": 0.3248,
+      "step": 9545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981881637043471e-05,
+      "loss": 0.3429,
+      "step": 9550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9818473862614996e-05,
+      "loss": 0.3222,
+      "step": 9555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9818131032543746e-05,
+      "loss": 0.3426,
+      "step": 9560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981778788022543e-05,
+      "loss": 0.3337,
+      "step": 9565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981744440566449e-05,
+      "loss": 0.2884,
+      "step": 9570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981710060886539e-05,
+      "loss": 0.319,
+      "step": 9575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9816756489832594e-05,
+      "loss": 0.3413,
+      "step": 9580
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981641204857057e-05,
+      "loss": 0.3181,
+      "step": 9585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981606728508379e-05,
+      "loss": 0.3521,
+      "step": 9590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9815722199376734e-05,
+      "loss": 0.3322,
+      "step": 9595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9815376791453876e-05,
+      "loss": 0.3417,
+      "step": 9600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9815031061319705e-05,
+      "loss": 0.3309,
+      "step": 9605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9814685008978713e-05,
+      "loss": 0.3283,
+      "step": 9610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981433863443538e-05,
+      "loss": 0.3128,
+      "step": 9615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981399193769422e-05,
+      "loss": 0.3348,
+      "step": 9620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9813644918759725e-05,
+      "loss": 0.3077,
+      "step": 9625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981329757763641e-05,
+      "loss": 0.3369,
+      "step": 9630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9812949914328766e-05,
+      "loss": 0.3688,
+      "step": 9635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981260192884133e-05,
+      "loss": 0.3196,
+      "step": 9640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981225362117861e-05,
+      "loss": 0.3187,
+      "step": 9645
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981190499134513e-05,
+      "loss": 0.3324,
+      "step": 9650
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981155603934541e-05,
+      "loss": 0.3113,
+      "step": 9655
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981120676518398e-05,
+      "loss": 0.3361,
+      "step": 9660
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9810857168865386e-05,
+      "loss": 0.3341,
+      "step": 9665
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981050725039417e-05,
+      "loss": 0.3221,
+      "step": 9670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.981015700977486e-05,
+      "loss": 0.3095,
+      "step": 9675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980980644701202e-05,
+      "loss": 0.3289,
+      "step": 9680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980945556211019e-05,
+      "loss": 0.2993,
+      "step": 9685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980910435507393e-05,
+      "loss": 0.3396,
+      "step": 9690
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9808752825907795e-05,
+      "loss": 0.3295,
+      "step": 9695
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980840097461635e-05,
+      "loss": 0.3072,
+      "step": 9700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980804880120418e-05,
+      "loss": 0.3135,
+      "step": 9705
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9807696305675844e-05,
+      "loss": 0.335,
+      "step": 9710
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980734348803592e-05,
+      "loss": 0.3319,
+      "step": 9715
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980699034828898e-05,
+      "loss": 0.3338,
+      "step": 9720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980663688643963e-05,
+      "loss": 0.3719,
+      "step": 9725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980628310249244e-05,
+      "loss": 0.3326,
+      "step": 9730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980592899645201e-05,
+      "loss": 0.3386,
+      "step": 9735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980557456832295e-05,
+      "loss": 0.3241,
+      "step": 9740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980521981810985e-05,
+      "loss": 0.3372,
+      "step": 9745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980486474581732e-05,
+      "loss": 0.3224,
+      "step": 9750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980450935144996e-05,
+      "loss": 0.2923,
+      "step": 9755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980415363501239e-05,
+      "loss": 0.3383,
+      "step": 9760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9803797596509235e-05,
+      "loss": 0.3293,
+      "step": 9765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980344123594512e-05,
+      "loss": 0.3373,
+      "step": 9770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.9803084553324666e-05,
+      "loss": 0.3358,
+      "step": 9775
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.98027275486525e-05,
+      "loss": 0.3051,
+      "step": 9780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980237022193326e-05,
+      "loss": 0.3373,
+      "step": 9785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.98020125731716e-05,
+      "loss": 0.3485,
+      "step": 9790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980165460237214e-05,
+      "loss": 0.3667,
+      "step": 9795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980129630953955e-05,
+      "loss": 0.3064,
+      "step": 9800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.980093769467846e-05,
+      "loss": 0.326,
+      "step": 9805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9800578757793546e-05,
+      "loss": 0.304,
+      "step": 9810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.980021949888946e-05,
+      "loss": 0.3163,
+      "step": 9815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979985991797087e-05,
+      "loss": 0.3347,
+      "step": 9820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979950001504244e-05,
+      "loss": 0.3157,
+      "step": 9825
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9799139790108846e-05,
+      "loss": 0.2742,
+      "step": 9830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979877924317477e-05,
+      "loss": 0.3241,
+      "step": 9835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979841837424488e-05,
+      "loss": 0.3275,
+      "step": 9840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979805718332388e-05,
+      "loss": 0.331,
+      "step": 9845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979769567041643e-05,
+      "loss": 0.327,
+      "step": 9850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979733383552727e-05,
+      "loss": 0.3175,
+      "step": 9855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979697167866105e-05,
+      "loss": 0.318,
+      "step": 9860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979660919982251e-05,
+      "loss": 0.3552,
+      "step": 9865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979624639901633e-05,
+      "loss": 0.316,
+      "step": 9870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979588327624725e-05,
+      "loss": 0.3402,
+      "step": 9875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979551983151995e-05,
+      "loss": 0.3271,
+      "step": 9880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9795156064839166e-05,
+      "loss": 0.3719,
+      "step": 9885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9794791976209634e-05,
+      "loss": 0.3154,
+      "step": 9890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979442756563605e-05,
+      "loss": 0.3374,
+      "step": 9895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9794062833123176e-05,
+      "loss": 0.328,
+      "step": 9900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979369777867574e-05,
+      "loss": 0.3276,
+      "step": 9905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979333240229847e-05,
+      "loss": 0.3205,
+      "step": 9910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9792966703996116e-05,
+      "loss": 0.309,
+      "step": 9915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979260068377344e-05,
+      "loss": 0.3277,
+      "step": 9920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979223434163517e-05,
+      "loss": 0.3036,
+      "step": 9925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979186767758608e-05,
+      "loss": 0.3299,
+      "step": 9930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9791500691630936e-05,
+      "loss": 0.335,
+      "step": 9935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979113338377448e-05,
+      "loss": 0.3341,
+      "step": 9940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979076575402151e-05,
+      "loss": 0.3399,
+      "step": 9945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979039780237677e-05,
+      "loss": 0.3285,
+      "step": 9950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.979002952884506e-05,
+      "loss": 0.3308,
+      "step": 9955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978966093343115e-05,
+      "loss": 0.3288,
+      "step": 9960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978929201613983e-05,
+      "loss": 0.341,
+      "step": 9965
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978892277697589e-05,
+      "loss": 0.3037,
+      "step": 9970
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978855321594413e-05,
+      "loss": 0.315,
+      "step": 9975
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978818333304933e-05,
+      "loss": 0.3041,
+      "step": 9980
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978781312829633e-05,
+      "loss": 0.32,
+      "step": 9985
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978744260168988e-05,
+      "loss": 0.3322,
+      "step": 9990
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978707175323485e-05,
+      "loss": 0.336,
+      "step": 9995
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978670058293601e-05,
+      "loss": 0.2962,
+      "step": 10000
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 0.3222596049308777,
+      "eval_runtime": 10.5408,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 10000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978632909079821e-05,
+      "loss": 0.3262,
+      "step": 10005
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9785957276826254e-05,
+      "loss": 0.3352,
+      "step": 10010
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978558514102498e-05,
+      "loss": 0.3371,
+      "step": 10015
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978521268339922e-05,
+      "loss": 0.3208,
+      "step": 10020
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.97848399039538e-05,
+      "loss": 0.3262,
+      "step": 10025
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978446680269358e-05,
+      "loss": 0.3099,
+      "step": 10030
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9784093379623386e-05,
+      "loss": 0.3112,
+      "step": 10035
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.9783719634748074e-05,
+      "loss": 0.3207,
+      "step": 10040
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978334556807249e-05,
+      "loss": 0.3204,
+      "step": 10045
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.97829711796015e-05,
+      "loss": 0.327,
+      "step": 10050
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978259646933996e-05,
+      "loss": 0.34,
+      "step": 10055
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978222143729274e-05,
+      "loss": 0.338,
+      "step": 10060
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978184608346471e-05,
+      "loss": 0.3158,
+      "step": 10065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978147040786073e-05,
+      "loss": 0.3068,
+      "step": 10070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.97810944104857e-05,
+      "loss": 0.3244,
+      "step": 10075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978071809134448e-05,
+      "loss": 0.3069,
+      "step": 10080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.978034145044198e-05,
+      "loss": 0.3123,
+      "step": 10085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.977996448778307e-05,
+      "loss": 0.3258,
+      "step": 10090
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977958720337265e-05,
+      "loss": 0.3306,
+      "step": 10095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977920959721561e-05,
+      "loss": 0.3117,
+      "step": 10100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977883166931688e-05,
+      "loss": 0.2943,
+      "step": 10105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977845341968135e-05,
+      "loss": 0.3441,
+      "step": 10110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9778074848313925e-05,
+      "loss": 0.3405,
+      "step": 10115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977769595521953e-05,
+      "loss": 0.3054,
+      "step": 10120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977731674040308e-05,
+      "loss": 0.3339,
+      "step": 10125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.97769372038695e-05,
+      "loss": 0.3445,
+      "step": 10130
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977655734562373e-05,
+      "loss": 0.3327,
+      "step": 10135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9776177165670676e-05,
+      "loss": 0.3147,
+      "step": 10140
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.97757966640153e-05,
+      "loss": 0.3163,
+      "step": 10145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977541584066253e-05,
+      "loss": 0.3185,
+      "step": 10150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9775034695617314e-05,
+      "loss": 0.3235,
+      "step": 10155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9774653228884594e-05,
+      "loss": 0.3276,
+      "step": 10160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977427144046933e-05,
+      "loss": 0.3118,
+      "step": 10165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977388933037649e-05,
+      "loss": 0.3274,
+      "step": 10170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977350689861101e-05,
+      "loss": 0.3286,
+      "step": 10175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9773124145177875e-05,
+      "loss": 0.308,
+      "step": 10180
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977274107008205e-05,
+      "loss": 0.3173,
+      "step": 10185
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9772357673328506e-05,
+      "loss": 0.3216,
+      "step": 10190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977197395492223e-05,
+      "loss": 0.3111,
+      "step": 10195
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977158991486818e-05,
+      "loss": 0.315,
+      "step": 10200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9771205553171376e-05,
+      "loss": 0.283,
+      "step": 10205
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9770820869836784e-05,
+      "loss": 0.3386,
+      "step": 10210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977043586486941e-05,
+      "loss": 0.3396,
+      "step": 10215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.977005053827425e-05,
+      "loss": 0.3357,
+      "step": 10220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976966489005631e-05,
+      "loss": 0.3087,
+      "step": 10225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9769278920220596e-05,
+      "loss": 0.3442,
+      "step": 10230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9768892628772115e-05,
+      "loss": 0.343,
+      "step": 10235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9768506015715896e-05,
+      "loss": 0.3256,
+      "step": 10240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976811908105694e-05,
+      "loss": 0.3094,
+      "step": 10245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9767731824800277e-05,
+      "loss": 0.3192,
+      "step": 10250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976734424695095e-05,
+      "loss": 0.3161,
+      "step": 10255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976695634751397e-05,
+      "loss": 0.3199,
+      "step": 10260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9766568126494394e-05,
+      "loss": 0.2714,
+      "step": 10265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976617958389724e-05,
+      "loss": 0.2952,
+      "step": 10270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976579071972758e-05,
+      "loss": 0.3468,
+      "step": 10275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976540153399044e-05,
+      "loss": 0.3307,
+      "step": 10280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976501202669089e-05,
+      "loss": 0.3136,
+      "step": 10285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976462219783397e-05,
+      "loss": 0.3317,
+      "step": 10290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976423204742475e-05,
+      "loss": 0.2882,
+      "step": 10295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976384157546831e-05,
+      "loss": 0.3635,
+      "step": 10300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.97634507819697e-05,
+      "loss": 0.307,
+      "step": 10305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9763059666933996e-05,
+      "loss": 0.2976,
+      "step": 10310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9762668230366286e-05,
+      "loss": 0.3068,
+      "step": 10315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9762276472271653e-05,
+      "loss": 0.3155,
+      "step": 10320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976188439265517e-05,
+      "loss": 0.3258,
+      "step": 10325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9761491991521945e-05,
+      "loss": 0.3216,
+      "step": 10330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.976109926887706e-05,
+      "loss": 0.3007,
+      "step": 10335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9760706224725616e-05,
+      "loss": 0.3354,
+      "step": 10340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.9760312859072726e-05,
+      "loss": 0.3335,
+      "step": 10345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.975991917192349e-05,
+      "loss": 0.3095,
+      "step": 10350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.975952516328302e-05,
+      "loss": 0.3143,
+      "step": 10355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.975913083315643e-05,
+      "loss": 0.3287,
+      "step": 10360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.975873618154885e-05,
+      "loss": 0.3102,
+      "step": 10365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.975834120846539e-05,
+      "loss": 0.3033,
+      "step": 10370
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975794591391119e-05,
+      "loss": 0.3203,
+      "step": 10375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975755029789138e-05,
+      "loss": 0.342,
+      "step": 10380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9757154360411095e-05,
+      "loss": 0.2972,
+      "step": 10385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975675810147547e-05,
+      "loss": 0.3124,
+      "step": 10390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975636152108966e-05,
+      "loss": 0.3097,
+      "step": 10395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975596461925881e-05,
+      "loss": 0.3229,
+      "step": 10400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975556739598808e-05,
+      "loss": 0.3471,
+      "step": 10405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975516985128261e-05,
+      "loss": 0.3405,
+      "step": 10410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9754771985147586e-05,
+      "loss": 0.316,
+      "step": 10415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975437379758816e-05,
+      "loss": 0.3192,
+      "step": 10420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.97539752886095e-05,
+      "loss": 0.305,
+      "step": 10425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975357645821678e-05,
+      "loss": 0.3185,
+      "step": 10430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9753177306415185e-05,
+      "loss": 0.3341,
+      "step": 10435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9752777833209904e-05,
+      "loss": 0.3518,
+      "step": 10440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975237803860611e-05,
+      "loss": 0.3262,
+      "step": 10445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9751977922609e-05,
+      "loss": 0.3294,
+      "step": 10450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975157748522376e-05,
+      "loss": 0.3283,
+      "step": 10455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975117672645562e-05,
+      "loss": 0.369,
+      "step": 10460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9750775646309744e-05,
+      "loss": 0.3316,
+      "step": 10465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.975037424479136e-05,
+      "loss": 0.3258,
+      "step": 10470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974997252190568e-05,
+      "loss": 0.3497,
+      "step": 10475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974957047765792e-05,
+      "loss": 0.2926,
+      "step": 10480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974916811205329e-05,
+      "loss": 0.3198,
+      "step": 10485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9748765425097035e-05,
+      "loss": 0.3363,
+      "step": 10490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974836241679436e-05,
+      "loss": 0.3114,
+      "step": 10495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974795908715051e-05,
+      "loss": 0.3215,
+      "step": 10500
+    },
+    {
+      "epoch": 0.37,
+      "eval_loss": 0.31911852955818176,
+      "eval_runtime": 10.5355,
+      "eval_samples_per_second": 9.492,
+      "eval_steps_per_second": 9.492,
+      "step": 10500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974755543617072e-05,
+      "loss": 0.312,
+      "step": 10505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974715146386024e-05,
+      "loss": 0.328,
+      "step": 10510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9746747170224306e-05,
+      "loss": 0.3036,
+      "step": 10515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974634255526817e-05,
+      "loss": 0.3429,
+      "step": 10520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974593761899708e-05,
+      "loss": 0.31,
+      "step": 10525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9745532361416295e-05,
+      "loss": 0.3408,
+      "step": 10530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974512678253108e-05,
+      "loss": 0.3312,
+      "step": 10535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.97447208823467e-05,
+      "loss": 0.3257,
+      "step": 10540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974431466086843e-05,
+      "loss": 0.3016,
+      "step": 10545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9743908118101545e-05,
+      "loss": 0.3356,
+      "step": 10550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9743501254051315e-05,
+      "loss": 0.3142,
+      "step": 10555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974309406872303e-05,
+      "loss": 0.3145,
+      "step": 10560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974268656212198e-05,
+      "loss": 0.3287,
+      "step": 10565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9742278734253444e-05,
+      "loss": 0.3307,
+      "step": 10570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974187058512273e-05,
+      "loss": 0.3203,
+      "step": 10575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9741462114735126e-05,
+      "loss": 0.3111,
+      "step": 10580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974105332309594e-05,
+      "loss": 0.3086,
+      "step": 10585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974064421021049e-05,
+      "loss": 0.2876,
+      "step": 10590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.974023477608407e-05,
+      "loss": 0.3169,
+      "step": 10595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973982502072201e-05,
+      "loss": 0.3043,
+      "step": 10600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9739414944129633e-05,
+      "loss": 0.3481,
+      "step": 10605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973900454631225e-05,
+      "loss": 0.3462,
+      "step": 10610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973859382727519e-05,
+      "loss": 0.3047,
+      "step": 10615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.97381827870238e-05,
+      "loss": 0.3369,
+      "step": 10620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973777142556342e-05,
+      "loss": 0.3028,
+      "step": 10625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973735974289937e-05,
+      "loss": 0.325,
+      "step": 10630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.9736947739037e-05,
+      "loss": 0.3325,
+      "step": 10635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973653541398168e-05,
+      "loss": 0.3084,
+      "step": 10640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973612276773874e-05,
+      "loss": 0.3269,
+      "step": 10645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973570980031355e-05,
+      "loss": 0.3139,
+      "step": 10650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.973529651171147e-05,
+      "loss": 0.3211,
+      "step": 10655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973488290193786e-05,
+      "loss": 0.3487,
+      "step": 10660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9734468970998105e-05,
+      "loss": 0.3103,
+      "step": 10665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9734054718897574e-05,
+      "loss": 0.3237,
+      "step": 10670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973364014564163e-05,
+      "loss": 0.2995,
+      "step": 10675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973322525123568e-05,
+      "loss": 0.3166,
+      "step": 10680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973281003568509e-05,
+      "loss": 0.2974,
+      "step": 10685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973239449899527e-05,
+      "loss": 0.3526,
+      "step": 10690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.97319786411716e-05,
+      "loss": 0.3314,
+      "step": 10695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9731562462219495e-05,
+      "loss": 0.3269,
+      "step": 10700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9731145962144344e-05,
+      "loss": 0.3073,
+      "step": 10705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973072914095156e-05,
+      "loss": 0.3118,
+      "step": 10710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.973031199864656e-05,
+      "loss": 0.3227,
+      "step": 10715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9729894535234756e-05,
+      "loss": 0.3423,
+      "step": 10720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972947675072157e-05,
+      "loss": 0.3177,
+      "step": 10725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972905864511241e-05,
+      "loss": 0.3132,
+      "step": 10730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9728640218412744e-05,
+      "loss": 0.3445,
+      "step": 10735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9728221470627965e-05,
+      "loss": 0.3371,
+      "step": 10740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9727802401763537e-05,
+      "loss": 0.3068,
+      "step": 10745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972738301182489e-05,
+      "loss": 0.308,
+      "step": 10750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972696330081746e-05,
+      "loss": 0.3264,
+      "step": 10755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9726543268746725e-05,
+      "loss": 0.3176,
+      "step": 10760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972612291561811e-05,
+      "loss": 0.3149,
+      "step": 10765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972570224143708e-05,
+      "loss": 0.3294,
+      "step": 10770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972528124620911e-05,
+      "loss": 0.341,
+      "step": 10775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9724859929939655e-05,
+      "loss": 0.331,
+      "step": 10780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972443829263419e-05,
+      "loss": 0.316,
+      "step": 10785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9724016334298186e-05,
+      "loss": 0.3089,
+      "step": 10790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972359405493712e-05,
+      "loss": 0.3033,
+      "step": 10795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972317145455648e-05,
+      "loss": 0.3289,
+      "step": 10800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9722748533161753e-05,
+      "loss": 0.3353,
+      "step": 10805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972232529075843e-05,
+      "loss": 0.3247,
+      "step": 10810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972190172735201e-05,
+      "loss": 0.3195,
+      "step": 10815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9721477842947984e-05,
+      "loss": 0.3657,
+      "step": 10820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972105363755186e-05,
+      "loss": 0.3068,
+      "step": 10825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.972062911116914e-05,
+      "loss": 0.3295,
+      "step": 10830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9720204263805345e-05,
+      "loss": 0.321,
+      "step": 10835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9719779095466e-05,
+      "loss": 0.3057,
+      "step": 10840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9719353606156606e-05,
+      "loss": 0.3415,
+      "step": 10845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.97189277958827e-05,
+      "loss": 0.3129,
+      "step": 10850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.97185016646498e-05,
+      "loss": 0.3154,
+      "step": 10855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971807521246345e-05,
+      "loss": 0.3251,
+      "step": 10860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971764843932919e-05,
+      "loss": 0.3323,
+      "step": 10865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971722134525254e-05,
+      "loss": 0.3168,
+      "step": 10870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9716793930239066e-05,
+      "loss": 0.3006,
+      "step": 10875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971636619429432e-05,
+      "loss": 0.3111,
+      "step": 10880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971593813742383e-05,
+      "loss": 0.3157,
+      "step": 10885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9715509759633194e-05,
+      "loss": 0.3366,
+      "step": 10890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9715081060927936e-05,
+      "loss": 0.3153,
+      "step": 10895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971465204131365e-05,
+      "loss": 0.3141,
+      "step": 10900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971422270079588e-05,
+      "loss": 0.342,
+      "step": 10905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971379303938023e-05,
+      "loss": 0.3119,
+      "step": 10910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9713363057072256e-05,
+      "loss": 0.2966,
+      "step": 10915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971293275387755e-05,
+      "loss": 0.3179,
+      "step": 10920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.9712502129801695e-05,
+      "loss": 0.315,
+      "step": 10925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971207118485029e-05,
+      "loss": 0.3437,
+      "step": 10930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971163991902893e-05,
+      "loss": 0.3267,
+      "step": 10935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.971120833234321e-05,
+      "loss": 0.3061,
+      "step": 10940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.971077642479873e-05,
+      "loss": 0.3452,
+      "step": 10945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9710344196401114e-05,
+      "loss": 0.2993,
+      "step": 10950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9709911647155953e-05,
+      "loss": 0.3241,
+      "step": 10955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9709478777068875e-05,
+      "loss": 0.3116,
+      "step": 10960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.97090455861455e-05,
+      "loss": 0.3161,
+      "step": 10965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970861207439146e-05,
+      "loss": 0.3234,
+      "step": 10970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970817824181236e-05,
+      "loss": 0.2984,
+      "step": 10975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970774408841387e-05,
+      "loss": 0.3408,
+      "step": 10980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970730961420159e-05,
+      "loss": 0.29,
+      "step": 10985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970687481918118e-05,
+      "loss": 0.3175,
+      "step": 10990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970643970335828e-05,
+      "loss": 0.2969,
+      "step": 10995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970600426673855e-05,
+      "loss": 0.3276,
+      "step": 11000
+    },
+    {
+      "epoch": 0.39,
+      "eval_loss": 0.31932923197746277,
+      "eval_runtime": 10.5374,
+      "eval_samples_per_second": 9.49,
+      "eval_steps_per_second": 9.49,
+      "step": 11000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970556850932763e-05,
+      "loss": 0.3115,
+      "step": 11005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970513243113119e-05,
+      "loss": 0.3083,
+      "step": 11010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970469603215488e-05,
+      "loss": 0.3105,
+      "step": 11015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9704259312404385e-05,
+      "loss": 0.3308,
+      "step": 11020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9703822271885355e-05,
+      "loss": 0.332,
+      "step": 11025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9703384910603473e-05,
+      "loss": 0.3089,
+      "step": 11030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970294722856442e-05,
+      "loss": 0.3084,
+      "step": 11035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970250922577387e-05,
+      "loss": 0.3292,
+      "step": 11040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970207090223753e-05,
+      "loss": 0.3208,
+      "step": 11045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.970163225796107e-05,
+      "loss": 0.3034,
+      "step": 11050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.97011932929502e-05,
+      "loss": 0.3415,
+      "step": 11055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9700754007210606e-05,
+      "loss": 0.3013,
+      "step": 11060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9700314400748005e-05,
+      "loss": 0.339,
+      "step": 11065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9699874473568095e-05,
+      "loss": 0.3431,
+      "step": 11070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.96994342256766e-05,
+      "loss": 0.2995,
+      "step": 11075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969899365707922e-05,
+      "loss": 0.3354,
+      "step": 11080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969855276778169e-05,
+      "loss": 0.3413,
+      "step": 11085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969811155778974e-05,
+      "loss": 0.3404,
+      "step": 11090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969767002710907e-05,
+      "loss": 0.3439,
+      "step": 11095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969722817574544e-05,
+      "loss": 0.3299,
+      "step": 11100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969678600370457e-05,
+      "loss": 0.3264,
+      "step": 11105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969634351099222e-05,
+      "loss": 0.3363,
+      "step": 11110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9695900697614126e-05,
+      "loss": 0.3401,
+      "step": 11115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969545756357603e-05,
+      "loss": 0.3148,
+      "step": 11120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969501410888369e-05,
+      "loss": 0.2961,
+      "step": 11125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9694570333542874e-05,
+      "loss": 0.2927,
+      "step": 11130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9694126237559335e-05,
+      "loss": 0.3554,
+      "step": 11135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969368182093883e-05,
+      "loss": 0.3212,
+      "step": 11140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9693237083687155e-05,
+      "loss": 0.3346,
+      "step": 11145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969279202581007e-05,
+      "loss": 0.341,
+      "step": 11150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9692346647313343e-05,
+      "loss": 0.3208,
+      "step": 11155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969190094820278e-05,
+      "loss": 0.3564,
+      "step": 11160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969145492848414e-05,
+      "loss": 0.3175,
+      "step": 11165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969100858816324e-05,
+      "loss": 0.3286,
+      "step": 11170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969056192724586e-05,
+      "loss": 0.3217,
+      "step": 11175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.969011494573781e-05,
+      "loss": 0.3129,
+      "step": 11180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968966764364489e-05,
+      "loss": 0.3072,
+      "step": 11185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.96892200209729e-05,
+      "loss": 0.3381,
+      "step": 11190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968877207772766e-05,
+      "loss": 0.297,
+      "step": 11195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9688323813914995e-05,
+      "loss": 0.3213,
+      "step": 11200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968787522954071e-05,
+      "loss": 0.3401,
+      "step": 11205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968742632461063e-05,
+      "loss": 0.3256,
+      "step": 11210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.9686977099130594e-05,
+      "loss": 0.315,
+      "step": 11215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968652755310643e-05,
+      "loss": 0.3294,
+      "step": 11220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.968607768654398e-05,
+      "loss": 0.3661,
+      "step": 11225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968562749944906e-05,
+      "loss": 0.3357,
+      "step": 11230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9685176991827555e-05,
+      "loss": 0.3292,
+      "step": 11235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968472616368529e-05,
+      "loss": 0.3061,
+      "step": 11240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968427501502812e-05,
+      "loss": 0.3282,
+      "step": 11245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968382354586191e-05,
+      "loss": 0.324,
+      "step": 11250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968337175619251e-05,
+      "loss": 0.3114,
+      "step": 11255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9682919646025805e-05,
+      "loss": 0.3184,
+      "step": 11260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9682467215367655e-05,
+      "loss": 0.3407,
+      "step": 11265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968201446422393e-05,
+      "loss": 0.2974,
+      "step": 11270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968156139260052e-05,
+      "loss": 0.3225,
+      "step": 11275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.968110800050329e-05,
+      "loss": 0.328,
+      "step": 11280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9680654287938145e-05,
+      "loss": 0.3285,
+      "step": 11285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9680200254910966e-05,
+      "loss": 0.3313,
+      "step": 11290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9679745901427656e-05,
+      "loss": 0.3458,
+      "step": 11295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967929122749411e-05,
+      "loss": 0.3087,
+      "step": 11300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967883623311622e-05,
+      "loss": 0.3174,
+      "step": 11305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9678380918299915e-05,
+      "loss": 0.3439,
+      "step": 11310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9677925283051105e-05,
+      "loss": 0.3138,
+      "step": 11315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967746932737569e-05,
+      "loss": 0.3158,
+      "step": 11320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96770130512796e-05,
+      "loss": 0.3368,
+      "step": 11325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9676556454768755e-05,
+      "loss": 0.3406,
+      "step": 11330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9676099537849083e-05,
+      "loss": 0.3389,
+      "step": 11335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967564230052653e-05,
+      "loss": 0.31,
+      "step": 11340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9675184742807025e-05,
+      "loss": 0.3074,
+      "step": 11345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96747268646965e-05,
+      "loss": 0.3408,
+      "step": 11350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967426866620091e-05,
+      "loss": 0.3086,
+      "step": 11355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967381014732621e-05,
+      "loss": 0.329,
+      "step": 11360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9673351308078335e-05,
+      "loss": 0.3225,
+      "step": 11365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9672892148463254e-05,
+      "loss": 0.3171,
+      "step": 11370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967243266848694e-05,
+      "loss": 0.2968,
+      "step": 11375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967197286815534e-05,
+      "loss": 0.3255,
+      "step": 11380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9671512747474436e-05,
+      "loss": 0.313,
+      "step": 11385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.967105230645019e-05,
+      "loss": 0.3233,
+      "step": 11390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.96705915450886e-05,
+      "loss": 0.3151,
+      "step": 11395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9670130463395624e-05,
+      "loss": 0.3142,
+      "step": 11400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9669669061377266e-05,
+      "loss": 0.3363,
+      "step": 11405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9669207339039514e-05,
+      "loss": 0.3107,
+      "step": 11410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9668745296388366e-05,
+      "loss": 0.3451,
+      "step": 11415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966828293342982e-05,
+      "loss": 0.3532,
+      "step": 11420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9667820250169874e-05,
+      "loss": 0.33,
+      "step": 11425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966735724661453e-05,
+      "loss": 0.3177,
+      "step": 11430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966689392276982e-05,
+      "loss": 0.3225,
+      "step": 11435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966643027864174e-05,
+      "loss": 0.3229,
+      "step": 11440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9665966314236324e-05,
+      "loss": 0.3148,
+      "step": 11445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966550202955959e-05,
+      "loss": 0.3068,
+      "step": 11450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966503742461756e-05,
+      "loss": 0.3421,
+      "step": 11455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966457249941629e-05,
+      "loss": 0.3407,
+      "step": 11460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9664107253961786e-05,
+      "loss": 0.3071,
+      "step": 11465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966364168826011e-05,
+      "loss": 0.3285,
+      "step": 11470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9663175802317306e-05,
+      "loss": 0.3165,
+      "step": 11475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966270959613941e-05,
+      "loss": 0.3275,
+      "step": 11480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966224306973249e-05,
+      "loss": 0.3074,
+      "step": 11485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.9661776223102583e-05,
+      "loss": 0.3402,
+      "step": 11490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966130905625578e-05,
+      "loss": 0.3486,
+      "step": 11495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966084156919813e-05,
+      "loss": 0.3687,
+      "step": 11500
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 0.31692856550216675,
+      "eval_runtime": 10.5329,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 11500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.966037376193569e-05,
+      "loss": 0.2997,
+      "step": 11505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.965990563447456e-05,
+      "loss": 0.3563,
+      "step": 11510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.96594371868208e-05,
+      "loss": 0.2938,
+      "step": 11515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.96589684189805e-05,
+      "loss": 0.309,
+      "step": 11520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965849933095976e-05,
+      "loss": 0.3217,
+      "step": 11525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965802992276464e-05,
+      "loss": 0.3492,
+      "step": 11530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9657560194401266e-05,
+      "loss": 0.3112,
+      "step": 11535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9657090145875704e-05,
+      "loss": 0.3295,
+      "step": 11540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965661977719409e-05,
+      "loss": 0.3108,
+      "step": 11545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965614908836251e-05,
+      "loss": 0.3004,
+      "step": 11550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965567807938709e-05,
+      "loss": 0.3019,
+      "step": 11555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965520675027394e-05,
+      "loss": 0.319,
+      "step": 11560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965473510102918e-05,
+      "loss": 0.3118,
+      "step": 11565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965426313165892e-05,
+      "loss": 0.3351,
+      "step": 11570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965379084216931e-05,
+      "loss": 0.3064,
+      "step": 11575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965331823256648e-05,
+      "loss": 0.3145,
+      "step": 11580
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9652845302856545e-05,
+      "loss": 0.3336,
+      "step": 11585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965237205304567e-05,
+      "loss": 0.298,
+      "step": 11590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965189848313999e-05,
+      "loss": 0.2922,
+      "step": 11595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965142459314565e-05,
+      "loss": 0.2873,
+      "step": 11600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9650950383068813e-05,
+      "loss": 0.3168,
+      "step": 11605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9650475852915625e-05,
+      "loss": 0.3136,
+      "step": 11610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.965000100269226e-05,
+      "loss": 0.3312,
+      "step": 11615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964952583240487e-05,
+      "loss": 0.3107,
+      "step": 11620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9649050342059635e-05,
+      "loss": 0.324,
+      "step": 11625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9648574531662726e-05,
+      "loss": 0.3329,
+      "step": 11630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964809840122032e-05,
+      "loss": 0.3211,
+      "step": 11635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.96476219507386e-05,
+      "loss": 0.3055,
+      "step": 11640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964714518022375e-05,
+      "loss": 0.3302,
+      "step": 11645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964666808968197e-05,
+      "loss": 0.3407,
+      "step": 11650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9646190679119445e-05,
+      "loss": 0.3278,
+      "step": 11655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964571294854237e-05,
+      "loss": 0.3406,
+      "step": 11660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964523489795696e-05,
+      "loss": 0.3247,
+      "step": 11665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964475652736942e-05,
+      "loss": 0.3085,
+      "step": 11670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964427783678595e-05,
+      "loss": 0.3015,
+      "step": 11675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964379882621278e-05,
+      "loss": 0.3264,
+      "step": 11680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964331949565611e-05,
+      "loss": 0.3442,
+      "step": 11685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9642839845122195e-05,
+      "loss": 0.3321,
+      "step": 11690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964235987461724e-05,
+      "loss": 0.3283,
+      "step": 11695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9641879584147476e-05,
+      "loss": 0.2993,
+      "step": 11700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964139897371915e-05,
+      "loss": 0.3119,
+      "step": 11705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964091804333849e-05,
+      "loss": 0.3216,
+      "step": 11710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.964043679301176e-05,
+      "loss": 0.313,
+      "step": 11715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963995522274519e-05,
+      "loss": 0.333,
+      "step": 11720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963947333254504e-05,
+      "loss": 0.3288,
+      "step": 11725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9638991122417564e-05,
+      "loss": 0.3167,
+      "step": 11730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963850859236904e-05,
+      "loss": 0.2891,
+      "step": 11735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.96380257424057e-05,
+      "loss": 0.3347,
+      "step": 11740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9637542572533844e-05,
+      "loss": 0.2952,
+      "step": 11745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963705908275973e-05,
+      "loss": 0.2992,
+      "step": 11750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9636575273089646e-05,
+      "loss": 0.3329,
+      "step": 11755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963609114352986e-05,
+      "loss": 0.2935,
+      "step": 11760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963560669408668e-05,
+      "loss": 0.3331,
+      "step": 11765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963512192476637e-05,
+      "loss": 0.3127,
+      "step": 11770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963463683557524e-05,
+      "loss": 0.3097,
+      "step": 11775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9634151426519584e-05,
+      "loss": 0.3086,
+      "step": 11780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9633665697605707e-05,
+      "loss": 0.3225,
+      "step": 11785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.9633179648839916e-05,
+      "loss": 0.3424,
+      "step": 11790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.963269328022853e-05,
+      "loss": 0.3198,
+      "step": 11795
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.963220659177784e-05,
+      "loss": 0.3151,
+      "step": 11800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9631719583494184e-05,
+      "loss": 0.3088,
+      "step": 11805
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9631232255383884e-05,
+      "loss": 0.3407,
+      "step": 11810
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.963074460745327e-05,
+      "loss": 0.3167,
+      "step": 11815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9630256639708675e-05,
+      "loss": 0.3313,
+      "step": 11820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962976835215641e-05,
+      "loss": 0.3076,
+      "step": 11825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9629279744802845e-05,
+      "loss": 0.3118,
+      "step": 11830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962879081765432e-05,
+      "loss": 0.3026,
+      "step": 11835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9628301570717176e-05,
+      "loss": 0.3041,
+      "step": 11840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962781200399776e-05,
+      "loss": 0.3037,
+      "step": 11845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962732211750244e-05,
+      "loss": 0.3158,
+      "step": 11850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962683191123757e-05,
+      "loss": 0.307,
+      "step": 11855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962634138520952e-05,
+      "loss": 0.3302,
+      "step": 11860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9625850539424644e-05,
+      "loss": 0.3337,
+      "step": 11865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9625359373889344e-05,
+      "loss": 0.3347,
+      "step": 11870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962486788860997e-05,
+      "loss": 0.3301,
+      "step": 11875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962437608359293e-05,
+      "loss": 0.3024,
+      "step": 11880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9623883958844574e-05,
+      "loss": 0.3066,
+      "step": 11885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9623391514371334e-05,
+      "loss": 0.3476,
+      "step": 11890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962289875017957e-05,
+      "loss": 0.2938,
+      "step": 11895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962240566627569e-05,
+      "loss": 0.3341,
+      "step": 11900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96219122626661e-05,
+      "loss": 0.3252,
+      "step": 11905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962141853935721e-05,
+      "loss": 0.3255,
+      "step": 11910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.962092449635543e-05,
+      "loss": 0.333,
+      "step": 11915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9620430133667165e-05,
+      "loss": 0.3133,
+      "step": 11920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961993545129884e-05,
+      "loss": 0.3345,
+      "step": 11925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961944044925688e-05,
+      "loss": 0.307,
+      "step": 11930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9618945127547715e-05,
+      "loss": 0.3097,
+      "step": 11935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961844948617777e-05,
+      "loss": 0.3506,
+      "step": 11940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9617953525153485e-05,
+      "loss": 0.3049,
+      "step": 11945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9617457244481294e-05,
+      "loss": 0.3365,
+      "step": 11950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961696064416764e-05,
+      "loss": 0.3425,
+      "step": 11955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961646372421898e-05,
+      "loss": 0.3385,
+      "step": 11960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961596648464176e-05,
+      "loss": 0.3063,
+      "step": 11965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9615468925442433e-05,
+      "loss": 0.306,
+      "step": 11970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961497104662747e-05,
+      "loss": 0.3161,
+      "step": 11975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961447284820333e-05,
+      "loss": 0.3016,
+      "step": 11980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9613974330176484e-05,
+      "loss": 0.3038,
+      "step": 11985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96134754925534e-05,
+      "loss": 0.3315,
+      "step": 11990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9612976335340556e-05,
+      "loss": 0.3475,
+      "step": 11995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961247685854443e-05,
+      "loss": 0.3028,
+      "step": 12000
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 0.31634584069252014,
+      "eval_runtime": 10.5394,
+      "eval_samples_per_second": 9.488,
+      "eval_steps_per_second": 9.488,
+      "step": 12000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961197706217152e-05,
+      "loss": 0.3324,
+      "step": 12005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961147694622831e-05,
+      "loss": 0.3144,
+      "step": 12010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961097651072128e-05,
+      "loss": 0.3181,
+      "step": 12015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.961047575565695e-05,
+      "loss": 0.3045,
+      "step": 12020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96099746810418e-05,
+      "loss": 0.3209,
+      "step": 12025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960947328688236e-05,
+      "loss": 0.3149,
+      "step": 12030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960897157318512e-05,
+      "loss": 0.3207,
+      "step": 12035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96084695399566e-05,
+      "loss": 0.3323,
+      "step": 12040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960796718720333e-05,
+      "loss": 0.3058,
+      "step": 12045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9607464514931824e-05,
+      "loss": 0.3216,
+      "step": 12050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9606961523148597e-05,
+      "loss": 0.3235,
+      "step": 12055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.96064582118602e-05,
+      "loss": 0.3093,
+      "step": 12060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960595458107315e-05,
+      "loss": 0.3025,
+      "step": 12065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.9605450630794e-05,
+      "loss": 0.3243,
+      "step": 12070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.960494636102929e-05,
+      "loss": 0.334,
+      "step": 12075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960444177178557e-05,
+      "loss": 0.336,
+      "step": 12080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960393686306939e-05,
+      "loss": 0.3214,
+      "step": 12085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9603431634887296e-05,
+      "loss": 0.3246,
+      "step": 12090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960292608724587e-05,
+      "loss": 0.3268,
+      "step": 12095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960242022015165e-05,
+      "loss": 0.3246,
+      "step": 12100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960191403361122e-05,
+      "loss": 0.3349,
+      "step": 12105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.960140752763115e-05,
+      "loss": 0.3367,
+      "step": 12110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9600900702218016e-05,
+      "loss": 0.3581,
+      "step": 12115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.96003935573784e-05,
+      "loss": 0.3233,
+      "step": 12120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9599886093118887e-05,
+      "loss": 0.3331,
+      "step": 12125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9599378309446063e-05,
+      "loss": 0.3189,
+      "step": 12130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959887020636652e-05,
+      "loss": 0.3251,
+      "step": 12135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959836178388687e-05,
+      "loss": 0.3252,
+      "step": 12140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9597853042013694e-05,
+      "loss": 0.3224,
+      "step": 12145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959734398075361e-05,
+      "loss": 0.3192,
+      "step": 12150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959683460011322e-05,
+      "loss": 0.3215,
+      "step": 12155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9596324900099145e-05,
+      "loss": 0.3254,
+      "step": 12160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9595814880718e-05,
+      "loss": 0.3287,
+      "step": 12165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959530454197641e-05,
+      "loss": 0.3454,
+      "step": 12170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9594793883881e-05,
+      "loss": 0.3208,
+      "step": 12175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.95942829064384e-05,
+      "loss": 0.3358,
+      "step": 12180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9593771609655244e-05,
+      "loss": 0.3224,
+      "step": 12185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9593259993538164e-05,
+      "loss": 0.3418,
+      "step": 12190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9592748058093824e-05,
+      "loss": 0.329,
+      "step": 12195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959223580332885e-05,
+      "loss": 0.3275,
+      "step": 12200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9591723229249905e-05,
+      "loss": 0.3173,
+      "step": 12205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9591210335863635e-05,
+      "loss": 0.3148,
+      "step": 12210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.959069712317671e-05,
+      "loss": 0.3237,
+      "step": 12215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9590183591195785e-05,
+      "loss": 0.3236,
+      "step": 12220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9589669739927534e-05,
+      "loss": 0.3257,
+      "step": 12225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958915556937862e-05,
+      "loss": 0.314,
+      "step": 12230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9588641079555734e-05,
+      "loss": 0.3264,
+      "step": 12235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958812627046555e-05,
+      "loss": 0.341,
+      "step": 12240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9587611142114744e-05,
+      "loss": 0.3338,
+      "step": 12245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958709569451001e-05,
+      "loss": 0.3163,
+      "step": 12250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958657992765805e-05,
+      "loss": 0.317,
+      "step": 12255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958606384156554e-05,
+      "loss": 0.3226,
+      "step": 12260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958554743623921e-05,
+      "loss": 0.33,
+      "step": 12265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9585030711685745e-05,
+      "loss": 0.318,
+      "step": 12270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958451366791185e-05,
+      "loss": 0.3325,
+      "step": 12275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958399630492425e-05,
+      "loss": 0.3203,
+      "step": 12280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958347862272966e-05,
+      "loss": 0.3243,
+      "step": 12285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9582960621334805e-05,
+      "loss": 0.355,
+      "step": 12290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.95824423007464e-05,
+      "loss": 0.3308,
+      "step": 12295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9581923660971186e-05,
+      "loss": 0.3271,
+      "step": 12300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9581404702015896e-05,
+      "loss": 0.3112,
+      "step": 12305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958088542388726e-05,
+      "loss": 0.3161,
+      "step": 12310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.958036582659204e-05,
+      "loss": 0.3227,
+      "step": 12315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9579845910136955e-05,
+      "loss": 0.3271,
+      "step": 12320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957932567452876e-05,
+      "loss": 0.3227,
+      "step": 12325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957880511977424e-05,
+      "loss": 0.3203,
+      "step": 12330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957828424588012e-05,
+      "loss": 0.3467,
+      "step": 12335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9577763052853186e-05,
+      "loss": 0.3006,
+      "step": 12340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957724154070019e-05,
+      "loss": 0.3177,
+      "step": 12345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957671970942792e-05,
+      "loss": 0.3384,
+      "step": 12350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.9576197559043134e-05,
+      "loss": 0.3048,
+      "step": 12355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.957567508955262e-05,
+      "loss": 0.3106,
+      "step": 12360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9575152300963154e-05,
+      "loss": 0.2756,
+      "step": 12365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957462919328154e-05,
+      "loss": 0.3165,
+      "step": 12370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9574105766514555e-05,
+      "loss": 0.3283,
+      "step": 12375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9573582020669007e-05,
+      "loss": 0.3237,
+      "step": 12380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957305795575169e-05,
+      "loss": 0.3246,
+      "step": 12385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957253357176941e-05,
+      "loss": 0.3304,
+      "step": 12390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957200886872897e-05,
+      "loss": 0.3393,
+      "step": 12395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957148384663719e-05,
+      "loss": 0.307,
+      "step": 12400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.957095850550089e-05,
+      "loss": 0.3274,
+      "step": 12405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9570432845326884e-05,
+      "loss": 0.3076,
+      "step": 12410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9569906866122e-05,
+      "loss": 0.3041,
+      "step": 12415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956938056789306e-05,
+      "loss": 0.309,
+      "step": 12420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956885395064692e-05,
+      "loss": 0.3154,
+      "step": 12425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956832701439039e-05,
+      "loss": 0.3421,
+      "step": 12430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9567799759130335e-05,
+      "loss": 0.3171,
+      "step": 12435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9567272184873586e-05,
+      "loss": 0.3307,
+      "step": 12440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9566744291627e-05,
+      "loss": 0.3093,
+      "step": 12445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956621607939742e-05,
+      "loss": 0.2988,
+      "step": 12450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956568754819173e-05,
+      "loss": 0.2914,
+      "step": 12455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956515869801677e-05,
+      "loss": 0.3039,
+      "step": 12460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9564629528879406e-05,
+      "loss": 0.3196,
+      "step": 12465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956410004078652e-05,
+      "loss": 0.2953,
+      "step": 12470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9563570233744995e-05,
+      "loss": 0.2914,
+      "step": 12475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956304010776168e-05,
+      "loss": 0.302,
+      "step": 12480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9562509662843495e-05,
+      "loss": 0.3218,
+      "step": 12485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.95619788989973e-05,
+      "loss": 0.3176,
+      "step": 12490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.956144781622999e-05,
+      "loss": 0.3164,
+      "step": 12495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9560916414548474e-05,
+      "loss": 0.2955,
+      "step": 12500
+    },
+    {
+      "epoch": 0.44,
+      "eval_loss": 0.31484097242355347,
+      "eval_runtime": 10.5384,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 12500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9560384693959644e-05,
+      "loss": 0.3188,
+      "step": 12505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955985265447041e-05,
+      "loss": 0.3233,
+      "step": 12510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9559320296087664e-05,
+      "loss": 0.3287,
+      "step": 12515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955878761881834e-05,
+      "loss": 0.3187,
+      "step": 12520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955825462266933e-05,
+      "loss": 0.3355,
+      "step": 12525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9557721307647576e-05,
+      "loss": 0.3231,
+      "step": 12530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9557187673759996e-05,
+      "loss": 0.3486,
+      "step": 12535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955665372101351e-05,
+      "loss": 0.3138,
+      "step": 12540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955611944941507e-05,
+      "loss": 0.3292,
+      "step": 12545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.95555848589716e-05,
+      "loss": 0.3262,
+      "step": 12550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9555049949690044e-05,
+      "loss": 0.3101,
+      "step": 12555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955451472157734e-05,
+      "loss": 0.3335,
+      "step": 12560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9554086309533434e-05,
+      "loss": 0.3185,
+      "step": 12565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955355050754219e-05,
+      "loss": 0.3242,
+      "step": 12570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955301438673928e-05,
+      "loss": 0.3294,
+      "step": 12575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9552477947131656e-05,
+      "loss": 0.3064,
+      "step": 12580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9551941188726294e-05,
+      "loss": 0.3415,
+      "step": 12585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955140411153015e-05,
+      "loss": 0.3077,
+      "step": 12590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9550866715550205e-05,
+      "loss": 0.3585,
+      "step": 12595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.955032900079343e-05,
+      "loss": 0.3242,
+      "step": 12600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954979096726682e-05,
+      "loss": 0.3211,
+      "step": 12605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954925261497735e-05,
+      "loss": 0.2988,
+      "step": 12610
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954871394393201e-05,
+      "loss": 0.3153,
+      "step": 12615
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954817495413781e-05,
+      "loss": 0.3298,
+      "step": 12620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954763564560172e-05,
+      "loss": 0.3225,
+      "step": 12625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954709601833078e-05,
+      "loss": 0.3155,
+      "step": 12630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954655607233195e-05,
+      "loss": 0.3131,
+      "step": 12635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9546015807612275e-05,
+      "loss": 0.3121,
+      "step": 12640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.954547522417877e-05,
+      "loss": 0.3246,
+      "step": 12645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954493432203844e-05,
+      "loss": 0.3338,
+      "step": 12650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954439310119831e-05,
+      "loss": 0.3075,
+      "step": 12655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954385156166541e-05,
+      "loss": 0.2985,
+      "step": 12660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9543309703446783e-05,
+      "loss": 0.3016,
+      "step": 12665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9542767526549445e-05,
+      "loss": 0.3416,
+      "step": 12670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954222503098045e-05,
+      "loss": 0.3014,
+      "step": 12675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954168221674683e-05,
+      "loss": 0.313,
+      "step": 12680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954113908385565e-05,
+      "loss": 0.3281,
+      "step": 12685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954059563231394e-05,
+      "loss": 0.3349,
+      "step": 12690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.954005186212878e-05,
+      "loss": 0.3061,
+      "step": 12695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9539507773307215e-05,
+      "loss": 0.3097,
+      "step": 12700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9538963365856305e-05,
+      "loss": 0.3377,
+      "step": 12705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953841863978314e-05,
+      "loss": 0.3343,
+      "step": 12710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953787359509477e-05,
+      "loss": 0.3359,
+      "step": 12715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953732823179829e-05,
+      "loss": 0.3121,
+      "step": 12720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9536782549900776e-05,
+      "loss": 0.3339,
+      "step": 12725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953623654940931e-05,
+      "loss": 0.3041,
+      "step": 12730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953569023033098e-05,
+      "loss": 0.3408,
+      "step": 12735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9535143592672874e-05,
+      "loss": 0.315,
+      "step": 12740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9534596636442106e-05,
+      "loss": 0.3189,
+      "step": 12745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953404936164577e-05,
+      "loss": 0.3107,
+      "step": 12750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953350176829097e-05,
+      "loss": 0.3081,
+      "step": 12755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9532953856384814e-05,
+      "loss": 0.3367,
+      "step": 12760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9532405625934425e-05,
+      "loss": 0.3211,
+      "step": 12765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9531857076946914e-05,
+      "loss": 0.3377,
+      "step": 12770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9531308209429404e-05,
+      "loss": 0.3257,
+      "step": 12775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953075902338903e-05,
+      "loss": 0.3615,
+      "step": 12780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9530209518832915e-05,
+      "loss": 0.307,
+      "step": 12785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952965969576819e-05,
+      "loss": 0.3345,
+      "step": 12790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9529109554202e-05,
+      "loss": 0.3315,
+      "step": 12795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.95285590941415e-05,
+      "loss": 0.3312,
+      "step": 12800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952800831559381e-05,
+      "loss": 0.3103,
+      "step": 12805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9527457218566104e-05,
+      "loss": 0.3058,
+      "step": 12810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952690580306553e-05,
+      "loss": 0.295,
+      "step": 12815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952635406909925e-05,
+      "loss": 0.3236,
+      "step": 12820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952580201667442e-05,
+      "loss": 0.3461,
+      "step": 12825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952524964579821e-05,
+      "loss": 0.3296,
+      "step": 12830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952469695647781e-05,
+      "loss": 0.3343,
+      "step": 12835
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952414394872037e-05,
+      "loss": 0.3293,
+      "step": 12840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952359062253309e-05,
+      "loss": 0.3196,
+      "step": 12845
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9523036977923145e-05,
+      "loss": 0.2872,
+      "step": 12850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952248301489772e-05,
+      "loss": 0.3446,
+      "step": 12855
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952192873346403e-05,
+      "loss": 0.3047,
+      "step": 12860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952137413362924e-05,
+      "loss": 0.3364,
+      "step": 12865
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9520819215400574e-05,
+      "loss": 0.3326,
+      "step": 12870
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.952026397878523e-05,
+      "loss": 0.3088,
+      "step": 12875
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951970842379042e-05,
+      "loss": 0.3121,
+      "step": 12880
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9519152550423355e-05,
+      "loss": 0.3348,
+      "step": 12885
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9518596358691246e-05,
+      "loss": 0.2744,
+      "step": 12890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951803984860133e-05,
+      "loss": 0.3146,
+      "step": 12895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9517483020160816e-05,
+      "loss": 0.3183,
+      "step": 12900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9516925873376955e-05,
+      "loss": 0.3227,
+      "step": 12905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951636840825695e-05,
+      "loss": 0.3254,
+      "step": 12910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951581062480807e-05,
+      "loss": 0.353,
+      "step": 12915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9515252523037545e-05,
+      "loss": 0.3276,
+      "step": 12920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9514694102952625e-05,
+      "loss": 0.3128,
+      "step": 12925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.951413536456055e-05,
+      "loss": 0.3235,
+      "step": 12930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.951357630786859e-05,
+      "loss": 0.3151,
+      "step": 12935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9513016932883994e-05,
+      "loss": 0.319,
+      "step": 12940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9512457239614026e-05,
+      "loss": 0.3616,
+      "step": 12945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.951189722806596e-05,
+      "loss": 0.3197,
+      "step": 12950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.951133689824706e-05,
+      "loss": 0.304,
+      "step": 12955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.95107762501646e-05,
+      "loss": 0.3244,
+      "step": 12960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.951021528382587e-05,
+      "loss": 0.2945,
+      "step": 12965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9509653999238143e-05,
+      "loss": 0.3401,
+      "step": 12970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950909239640871e-05,
+      "loss": 0.3175,
+      "step": 12975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950853047534487e-05,
+      "loss": 0.3314,
+      "step": 12980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950796823605391e-05,
+      "loss": 0.3377,
+      "step": 12985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950740567854313e-05,
+      "loss": 0.3175,
+      "step": 12990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950684280281985e-05,
+      "loss": 0.3056,
+      "step": 12995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950627960889136e-05,
+      "loss": 0.306,
+      "step": 13000
+    },
+    {
+      "epoch": 0.46,
+      "eval_loss": 0.31242769956588745,
+      "eval_runtime": 10.5492,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 13000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9505716096764976e-05,
+      "loss": 0.3369,
+      "step": 13005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950515226644802e-05,
+      "loss": 0.3456,
+      "step": 13010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9504588117947814e-05,
+      "loss": 0.3514,
+      "step": 13015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950402365127168e-05,
+      "loss": 0.3113,
+      "step": 13020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950345886642694e-05,
+      "loss": 0.3034,
+      "step": 13025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950289376342094e-05,
+      "loss": 0.3103,
+      "step": 13030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9502328342261015e-05,
+      "loss": 0.3041,
+      "step": 13035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9501762602954505e-05,
+      "loss": 0.3191,
+      "step": 13040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9501196545508756e-05,
+      "loss": 0.3235,
+      "step": 13045
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.950063016993112e-05,
+      "loss": 0.324,
+      "step": 13050
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9500063476228936e-05,
+      "loss": 0.3255,
+      "step": 13055
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9499496464409576e-05,
+      "loss": 0.3304,
+      "step": 13060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9498929134480406e-05,
+      "loss": 0.3112,
+      "step": 13065
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9498361486448786e-05,
+      "loss": 0.3035,
+      "step": 13070
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949779352032208e-05,
+      "loss": 0.3178,
+      "step": 13075
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949722523610768e-05,
+      "loss": 0.309,
+      "step": 13080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949665663381295e-05,
+      "loss": 0.3165,
+      "step": 13085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949608771344528e-05,
+      "loss": 0.3333,
+      "step": 13090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949551847501205e-05,
+      "loss": 0.3362,
+      "step": 13095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9494948918520655e-05,
+      "loss": 0.2987,
+      "step": 13100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9494379043978494e-05,
+      "loss": 0.2885,
+      "step": 13105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9493808851392964e-05,
+      "loss": 0.3404,
+      "step": 13110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9493238340771465e-05,
+      "loss": 0.3182,
+      "step": 13115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949266751212142e-05,
+      "loss": 0.3212,
+      "step": 13120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949209636545021e-05,
+      "loss": 0.3207,
+      "step": 13125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949152490076528e-05,
+      "loss": 0.3215,
+      "step": 13130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.949095311807404e-05,
+      "loss": 0.3127,
+      "step": 13135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9490381017383906e-05,
+      "loss": 0.3429,
+      "step": 13140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9489808598702325e-05,
+      "loss": 0.3377,
+      "step": 13145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948923586203671e-05,
+      "loss": 0.3528,
+      "step": 13150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948866280739452e-05,
+      "loss": 0.3036,
+      "step": 13155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9488089434783165e-05,
+      "loss": 0.3039,
+      "step": 13160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948751574421011e-05,
+      "loss": 0.3316,
+      "step": 13165
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948694173568281e-05,
+      "loss": 0.3309,
+      "step": 13170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.94863674092087e-05,
+      "loss": 0.3295,
+      "step": 13175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9485792764795255e-05,
+      "loss": 0.3174,
+      "step": 13180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948521780244992e-05,
+      "loss": 0.3137,
+      "step": 13185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948464252218017e-05,
+      "loss": 0.3127,
+      "step": 13190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9484066923993476e-05,
+      "loss": 0.3373,
+      "step": 13195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.948349100789731e-05,
+      "loss": 0.3312,
+      "step": 13200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.9483030046130576e-05,
+      "loss": 0.3225,
+      "step": 13205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.94824535578162e-05,
+      "loss": 0.3187,
+      "step": 13210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.94818767516133e-05,
+      "loss": 0.3131,
+      "step": 13215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.948129962752937e-05,
+      "loss": 0.3162,
+      "step": 13220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.948072218557189e-05,
+      "loss": 0.3216,
+      "step": 13225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.948014442574837e-05,
+      "loss": 0.3128,
+      "step": 13230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947956634806631e-05,
+      "loss": 0.2955,
+      "step": 13235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947898795253321e-05,
+      "loss": 0.3057,
+      "step": 13240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947840923915659e-05,
+      "loss": 0.315,
+      "step": 13245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9477830207943944e-05,
+      "loss": 0.3163,
+      "step": 13250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947725085890281e-05,
+      "loss": 0.3074,
+      "step": 13255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947667119204071e-05,
+      "loss": 0.3027,
+      "step": 13260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947609120736516e-05,
+      "loss": 0.3092,
+      "step": 13265
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94755109048837e-05,
+      "loss": 0.3223,
+      "step": 13270
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947493028460386e-05,
+      "loss": 0.317,
+      "step": 13275
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9474349346533175e-05,
+      "loss": 0.3389,
+      "step": 13280
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947376809067921e-05,
+      "loss": 0.3291,
+      "step": 13285
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947318651704948e-05,
+      "loss": 0.278,
+      "step": 13290
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947260462565155e-05,
+      "loss": 0.3282,
+      "step": 13295
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947202241649298e-05,
+      "loss": 0.3176,
+      "step": 13300
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9471439889581325e-05,
+      "loss": 0.3368,
+      "step": 13305
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947085704492416e-05,
+      "loss": 0.3488,
+      "step": 13310
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.947027388252903e-05,
+      "loss": 0.3198,
+      "step": 13315
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9469690402403525e-05,
+      "loss": 0.3155,
+      "step": 13320
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946910660455522e-05,
+      "loss": 0.3389,
+      "step": 13325
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946852248899169e-05,
+      "loss": 0.3143,
+      "step": 13330
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946793805572052e-05,
+      "loss": 0.3292,
+      "step": 13335
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94673533047493e-05,
+      "loss": 0.3163,
+      "step": 13340
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946676823608562e-05,
+      "loss": 0.325,
+      "step": 13345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946618284973708e-05,
+      "loss": 0.3158,
+      "step": 13350
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946559714571127e-05,
+      "loss": 0.3372,
+      "step": 13355
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946501112401582e-05,
+      "loss": 0.3333,
+      "step": 13360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9464424784658314e-05,
+      "loss": 0.3234,
+      "step": 13365
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946383812764638e-05,
+      "loss": 0.2952,
+      "step": 13370
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946325115298763e-05,
+      "loss": 0.3085,
+      "step": 13375
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946266386068968e-05,
+      "loss": 0.2821,
+      "step": 13380
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946207625076017e-05,
+      "loss": 0.2997,
+      "step": 13385
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.946148832320672e-05,
+      "loss": 0.3096,
+      "step": 13390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9460900078036955e-05,
+      "loss": 0.3095,
+      "step": 13395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9460311515258526e-05,
+      "loss": 0.3151,
+      "step": 13400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945972263487908e-05,
+      "loss": 0.341,
+      "step": 13405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945913343690625e-05,
+      "loss": 0.3015,
+      "step": 13410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9458543921347686e-05,
+      "loss": 0.339,
+      "step": 13415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9457954088211055e-05,
+      "loss": 0.3195,
+      "step": 13420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945736393750401e-05,
+      "loss": 0.3124,
+      "step": 13425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9456773469234206e-05,
+      "loss": 0.3075,
+      "step": 13430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9456182683409316e-05,
+      "loss": 0.3203,
+      "step": 13435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945559158003702e-05,
+      "loss": 0.3248,
+      "step": 13440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945500015912498e-05,
+      "loss": 0.3142,
+      "step": 13445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945440842068087e-05,
+      "loss": 0.3309,
+      "step": 13450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94538163647124e-05,
+      "loss": 0.3139,
+      "step": 13455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945322399122723e-05,
+      "loss": 0.2987,
+      "step": 13460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945263130023306e-05,
+      "loss": 0.3205,
+      "step": 13465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945203829173759e-05,
+      "loss": 0.3159,
+      "step": 13470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945144496574852e-05,
+      "loss": 0.3238,
+      "step": 13475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945085132227354e-05,
+      "loss": 0.3118,
+      "step": 13480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.945025736132038e-05,
+      "loss": 0.3195,
+      "step": 13485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.9449663082896744e-05,
+      "loss": 0.2916,
+      "step": 13490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.944906848701034e-05,
+      "loss": 0.3271,
+      "step": 13495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.94484735736689e-05,
+      "loss": 0.3491,
+      "step": 13500
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 0.31191739439964294,
+      "eval_runtime": 10.5398,
+      "eval_samples_per_second": 9.488,
+      "eval_steps_per_second": 9.488,
+      "step": 13500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9447878342880134e-05,
+      "loss": 0.3164,
+      "step": 13505
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944728279465179e-05,
+      "loss": 0.31,
+      "step": 13510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9446686928991574e-05,
+      "loss": 0.2992,
+      "step": 13515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944609074590725e-05,
+      "loss": 0.3281,
+      "step": 13520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944549424540655e-05,
+      "loss": 0.3215,
+      "step": 13525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944489742749721e-05,
+      "loss": 0.3348,
+      "step": 13530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9444300292186994e-05,
+      "loss": 0.323,
+      "step": 13535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944370283948365e-05,
+      "loss": 0.3194,
+      "step": 13540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9443105069394924e-05,
+      "loss": 0.34,
+      "step": 13545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944250698192859e-05,
+      "loss": 0.3085,
+      "step": 13550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944190857709242e-05,
+      "loss": 0.314,
+      "step": 13555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944130985489417e-05,
+      "loss": 0.3261,
+      "step": 13560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944071081534161e-05,
+      "loss": 0.3415,
+      "step": 13565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.944011145844254e-05,
+      "loss": 0.3244,
+      "step": 13570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9439511784204715e-05,
+      "loss": 0.3109,
+      "step": 13575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943891179263595e-05,
+      "loss": 0.3116,
+      "step": 13580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9438311483744013e-05,
+      "loss": 0.3322,
+      "step": 13585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9437710857536715e-05,
+      "loss": 0.3271,
+      "step": 13590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943710991402184e-05,
+      "loss": 0.3172,
+      "step": 13595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.94365086532072e-05,
+      "loss": 0.2977,
+      "step": 13600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9435907075100595e-05,
+      "loss": 0.2983,
+      "step": 13605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943530517970985e-05,
+      "loss": 0.3503,
+      "step": 13610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943470296704276e-05,
+      "loss": 0.326,
+      "step": 13615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943410043710715e-05,
+      "loss": 0.3147,
+      "step": 13620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9433497589910864e-05,
+      "loss": 0.3018,
+      "step": 13625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9432894425461704e-05,
+      "loss": 0.3277,
+      "step": 13630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.943229094376752e-05,
+      "loss": 0.3264,
+      "step": 13635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9431687144836136e-05,
+      "loss": 0.3253,
+      "step": 13640
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9431083028675394e-05,
+      "loss": 0.3422,
+      "step": 13645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9430478595293136e-05,
+      "loss": 0.3188,
+      "step": 13650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942987384469723e-05,
+      "loss": 0.3083,
+      "step": 13655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942926877689549e-05,
+      "loss": 0.3319,
+      "step": 13660
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942866339189581e-05,
+      "loss": 0.2984,
+      "step": 13665
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942805768970603e-05,
+      "loss": 0.2981,
+      "step": 13670
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9427451670334014e-05,
+      "loss": 0.3058,
+      "step": 13675
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942684533378764e-05,
+      "loss": 0.3059,
+      "step": 13680
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942623868007478e-05,
+      "loss": 0.3228,
+      "step": 13685
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9425631709203314e-05,
+      "loss": 0.3068,
+      "step": 13690
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.94250244211811e-05,
+      "loss": 0.3145,
+      "step": 13695
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9424416816016055e-05,
+      "loss": 0.3242,
+      "step": 13700
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942380889371605e-05,
+      "loss": 0.3058,
+      "step": 13705
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942320065428898e-05,
+      "loss": 0.3208,
+      "step": 13710
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9422592097742745e-05,
+      "loss": 0.332,
+      "step": 13715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942198322408526e-05,
+      "loss": 0.3246,
+      "step": 13720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.94213740333244e-05,
+      "loss": 0.324,
+      "step": 13725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.94207645254681e-05,
+      "loss": 0.3244,
+      "step": 13730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.942015470052426e-05,
+      "loss": 0.2947,
+      "step": 13735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9419544558500813e-05,
+      "loss": 0.331,
+      "step": 13740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941893409940567e-05,
+      "loss": 0.3408,
+      "step": 13745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9418323323246764e-05,
+      "loss": 0.3019,
+      "step": 13750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9417712230032016e-05,
+      "loss": 0.3216,
+      "step": 13755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941710081976937e-05,
+      "loss": 0.3146,
+      "step": 13760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941648909246676e-05,
+      "loss": 0.2873,
+      "step": 13765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941587704813213e-05,
+      "loss": 0.3238,
+      "step": 13770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.941526468677343e-05,
+      "loss": 0.3233,
+      "step": 13775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9414652008398605e-05,
+      "loss": 0.3045,
+      "step": 13780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.9414039013015615e-05,
+      "loss": 0.3186,
+      "step": 13785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9413425700632415e-05,
+      "loss": 0.3047,
+      "step": 13790
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9412812071256974e-05,
+      "loss": 0.3117,
+      "step": 13795
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.941219812489726e-05,
+      "loss": 0.3269,
+      "step": 13800
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9411583861561236e-05,
+      "loss": 0.303,
+      "step": 13805
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.941096928125688e-05,
+      "loss": 0.3261,
+      "step": 13810
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9410354383992186e-05,
+      "loss": 0.2925,
+      "step": 13815
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940973916977512e-05,
+      "loss": 0.3165,
+      "step": 13820
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940912363861368e-05,
+      "loss": 0.3106,
+      "step": 13825
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940850779051586e-05,
+      "loss": 0.3322,
+      "step": 13830
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9407891625489644e-05,
+      "loss": 0.3183,
+      "step": 13835
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9407275143543044e-05,
+      "loss": 0.3155,
+      "step": 13840
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940665834468407e-05,
+      "loss": 0.3208,
+      "step": 13845
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940604122892071e-05,
+      "loss": 0.306,
+      "step": 13850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940542379626099e-05,
+      "loss": 0.3462,
+      "step": 13855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9404806046712927e-05,
+      "loss": 0.3338,
+      "step": 13860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940418798028455e-05,
+      "loss": 0.3077,
+      "step": 13865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940356959698386e-05,
+      "loss": 0.3194,
+      "step": 13870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9402950896818906e-05,
+      "loss": 0.2823,
+      "step": 13875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940233187979772e-05,
+      "loss": 0.3186,
+      "step": 13880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940171254592834e-05,
+      "loss": 0.3073,
+      "step": 13885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.940109289521879e-05,
+      "loss": 0.3308,
+      "step": 13890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9400472927677144e-05,
+      "loss": 0.284,
+      "step": 13895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939985264331143e-05,
+      "loss": 0.3212,
+      "step": 13900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939923204212972e-05,
+      "loss": 0.316,
+      "step": 13905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9398611124140046e-05,
+      "loss": 0.2989,
+      "step": 13910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939798988935049e-05,
+      "loss": 0.3339,
+      "step": 13915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939736833776912e-05,
+      "loss": 0.3322,
+      "step": 13920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9396746469403996e-05,
+      "loss": 0.3308,
+      "step": 13925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.93961242842632e-05,
+      "loss": 0.3017,
+      "step": 13930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9395501782354813e-05,
+      "loss": 0.3068,
+      "step": 13935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939487896368691e-05,
+      "loss": 0.3179,
+      "step": 13940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939425582826757e-05,
+      "loss": 0.3163,
+      "step": 13945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9393632376104904e-05,
+      "loss": 0.3401,
+      "step": 13950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9393008607206995e-05,
+      "loss": 0.32,
+      "step": 13955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9392384521581945e-05,
+      "loss": 0.3022,
+      "step": 13960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939176011923786e-05,
+      "loss": 0.3167,
+      "step": 13965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.939113540018284e-05,
+      "loss": 0.3418,
+      "step": 13970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9390510364425004e-05,
+      "loss": 0.3078,
+      "step": 13975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9389885011972464e-05,
+      "loss": 0.3185,
+      "step": 13980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938925934283334e-05,
+      "loss": 0.3481,
+      "step": 13985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9388633357015755e-05,
+      "loss": 0.317,
+      "step": 13990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938800705452784e-05,
+      "loss": 0.3323,
+      "step": 13995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9387380435377726e-05,
+      "loss": 0.316,
+      "step": 14000
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 0.3117991089820862,
+      "eval_runtime": 10.538,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 14000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938675349957356e-05,
+      "loss": 0.3099,
+      "step": 14005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938612624712346e-05,
+      "loss": 0.3418,
+      "step": 14010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938549867803558e-05,
+      "loss": 0.3279,
+      "step": 14015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9384870792318075e-05,
+      "loss": 0.3224,
+      "step": 14020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938424258997909e-05,
+      "loss": 0.3266,
+      "step": 14025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9383614071026787e-05,
+      "loss": 0.2839,
+      "step": 14030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938298523546932e-05,
+      "loss": 0.3114,
+      "step": 14035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938235608331486e-05,
+      "loss": 0.3205,
+      "step": 14040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938172661457158e-05,
+      "loss": 0.3337,
+      "step": 14045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.9381096829247644e-05,
+      "loss": 0.318,
+      "step": 14050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.938046672735123e-05,
+      "loss": 0.316,
+      "step": 14055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.937983630889053e-05,
+      "loss": 0.3182,
+      "step": 14060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.937920557387372e-05,
+      "loss": 0.3189,
+      "step": 14065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937857452230899e-05,
+      "loss": 0.3122,
+      "step": 14070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937794315420454e-05,
+      "loss": 0.2821,
+      "step": 14075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9377311469568566e-05,
+      "loss": 0.3035,
+      "step": 14080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937667946840926e-05,
+      "loss": 0.3213,
+      "step": 14085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937604715073484e-05,
+      "loss": 0.3141,
+      "step": 14090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937541451655351e-05,
+      "loss": 0.3298,
+      "step": 14095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937478156587349e-05,
+      "loss": 0.3311,
+      "step": 14100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9374148298702994e-05,
+      "loss": 0.3197,
+      "step": 14105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937351471505024e-05,
+      "loss": 0.3159,
+      "step": 14110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937288081492347e-05,
+      "loss": 0.3294,
+      "step": 14115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9372246598330895e-05,
+      "loss": 0.3011,
+      "step": 14120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.937161206528077e-05,
+      "loss": 0.3129,
+      "step": 14125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9370977215781314e-05,
+      "loss": 0.3069,
+      "step": 14130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9370342049840786e-05,
+      "loss": 0.3171,
+      "step": 14135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9369706567467425e-05,
+      "loss": 0.3133,
+      "step": 14140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9369070768669477e-05,
+      "loss": 0.3231,
+      "step": 14145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936843465345522e-05,
+      "loss": 0.3313,
+      "step": 14150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936779822183289e-05,
+      "loss": 0.3145,
+      "step": 14155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9367161473810766e-05,
+      "loss": 0.3093,
+      "step": 14160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93665244093971e-05,
+      "loss": 0.3168,
+      "step": 14165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9365887028600175e-05,
+      "loss": 0.3277,
+      "step": 14170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9365249331428265e-05,
+      "loss": 0.3193,
+      "step": 14175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9364611317889644e-05,
+      "loss": 0.315,
+      "step": 14180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936397298799261e-05,
+      "loss": 0.3124,
+      "step": 14185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9363334341745446e-05,
+      "loss": 0.3186,
+      "step": 14190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936269537915643e-05,
+      "loss": 0.3363,
+      "step": 14195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936205610023388e-05,
+      "loss": 0.3318,
+      "step": 14200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.936141650498608e-05,
+      "loss": 0.2952,
+      "step": 14205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9360776593421347e-05,
+      "loss": 0.3331,
+      "step": 14210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9360136365547985e-05,
+      "loss": 0.3029,
+      "step": 14215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93594958213743e-05,
+      "loss": 0.2965,
+      "step": 14220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935885496090862e-05,
+      "loss": 0.3002,
+      "step": 14225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9358213784159264e-05,
+      "loss": 0.3356,
+      "step": 14230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935757229113455e-05,
+      "loss": 0.3206,
+      "step": 14235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9356930481842814e-05,
+      "loss": 0.3251,
+      "step": 14240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935628835629238e-05,
+      "loss": 0.3013,
+      "step": 14245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93556459144916e-05,
+      "loss": 0.2957,
+      "step": 14250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935500315644881e-05,
+      "loss": 0.2915,
+      "step": 14255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935436008217236e-05,
+      "loss": 0.3131,
+      "step": 14260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935371669167058e-05,
+      "loss": 0.3186,
+      "step": 14265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935307298495185e-05,
+      "loss": 0.2998,
+      "step": 14270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935242896202451e-05,
+      "loss": 0.3304,
+      "step": 14275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935178462289694e-05,
+      "loss": 0.3435,
+      "step": 14280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9351139967577475e-05,
+      "loss": 0.2953,
+      "step": 14285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.935049499607452e-05,
+      "loss": 0.3056,
+      "step": 14290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9349849708396436e-05,
+      "loss": 0.302,
+      "step": 14295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93492041045516e-05,
+      "loss": 0.3219,
+      "step": 14300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934855818454839e-05,
+      "loss": 0.3243,
+      "step": 14305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93479119483952e-05,
+      "loss": 0.3482,
+      "step": 14310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.9347265396100414e-05,
+      "loss": 0.3026,
+      "step": 14315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934661852767244e-05,
+      "loss": 0.32,
+      "step": 14320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934597134311967e-05,
+      "loss": 0.3109,
+      "step": 14325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.93453238424505e-05,
+      "loss": 0.3353,
+      "step": 14330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934467602567335e-05,
+      "loss": 0.2842,
+      "step": 14335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934402789279662e-05,
+      "loss": 0.304,
+      "step": 14340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934337944382874e-05,
+      "loss": 0.3147,
+      "step": 14345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.934273067877811e-05,
+      "loss": 0.2964,
+      "step": 14350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.934208159765317e-05,
+      "loss": 0.3063,
+      "step": 14355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9341432200462343e-05,
+      "loss": 0.3435,
+      "step": 14360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.934078248721407e-05,
+      "loss": 0.306,
+      "step": 14365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.934013245791676e-05,
+      "loss": 0.3341,
+      "step": 14370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933948211257888e-05,
+      "loss": 0.3356,
+      "step": 14375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933883145120886e-05,
+      "loss": 0.3242,
+      "step": 14380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933818047381516e-05,
+      "loss": 0.3052,
+      "step": 14385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933752918040622e-05,
+      "loss": 0.3118,
+      "step": 14390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933687757099051e-05,
+      "loss": 0.3243,
+      "step": 14395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933622564557648e-05,
+      "loss": 0.2979,
+      "step": 14400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9335573404172594e-05,
+      "loss": 0.3052,
+      "step": 14405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9334920846787324e-05,
+      "loss": 0.3168,
+      "step": 14410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933426797342915e-05,
+      "loss": 0.3032,
+      "step": 14415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933361478410654e-05,
+      "loss": 0.319,
+      "step": 14420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933296127882798e-05,
+      "loss": 0.3071,
+      "step": 14425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9332307457601955e-05,
+      "loss": 0.2847,
+      "step": 14430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9331653320436954e-05,
+      "loss": 0.2941,
+      "step": 14435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.933099886734147e-05,
+      "loss": 0.2967,
+      "step": 14440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9330344098324e-05,
+      "loss": 0.304,
+      "step": 14445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932968901339304e-05,
+      "loss": 0.3298,
+      "step": 14450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932903361255711e-05,
+      "loss": 0.2917,
+      "step": 14455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9328377895824705e-05,
+      "loss": 0.2729,
+      "step": 14460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9327721863204354e-05,
+      "loss": 0.316,
+      "step": 14465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932706551470456e-05,
+      "loss": 0.2947,
+      "step": 14470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932640885033386e-05,
+      "loss": 0.3385,
+      "step": 14475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9325751870100765e-05,
+      "loss": 0.2871,
+      "step": 14480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932509457401382e-05,
+      "loss": 0.3188,
+      "step": 14485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9324436962081546e-05,
+      "loss": 0.302,
+      "step": 14490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932377903431249e-05,
+      "loss": 0.3015,
+      "step": 14495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9323120790715194e-05,
+      "loss": 0.3215,
+      "step": 14500
+    },
+    {
+      "epoch": 0.51,
+      "eval_loss": 0.3104207515716553,
+      "eval_runtime": 10.5492,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 14500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9322462231298206e-05,
+      "loss": 0.3215,
+      "step": 14505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932180335607007e-05,
+      "loss": 0.3366,
+      "step": 14510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932114416503936e-05,
+      "loss": 0.3001,
+      "step": 14515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9320484658214605e-05,
+      "loss": 0.2973,
+      "step": 14520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9319824835604386e-05,
+      "loss": 0.3296,
+      "step": 14525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9319164697217276e-05,
+      "loss": 0.2947,
+      "step": 14530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931850424306183e-05,
+      "loss": 0.327,
+      "step": 14535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931784347314665e-05,
+      "loss": 0.3126,
+      "step": 14540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931718238748029e-05,
+      "loss": 0.311,
+      "step": 14545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9316520986071335e-05,
+      "loss": 0.3362,
+      "step": 14550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931585926892839e-05,
+      "loss": 0.3107,
+      "step": 14555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931519723606003e-05,
+      "loss": 0.3039,
+      "step": 14560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9314534887474864e-05,
+      "loss": 0.3238,
+      "step": 14565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9313872223181485e-05,
+      "loss": 0.3258,
+      "step": 14570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93132092431885e-05,
+      "loss": 0.3171,
+      "step": 14575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931254594750451e-05,
+      "loss": 0.2988,
+      "step": 14580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9311882336138136e-05,
+      "loss": 0.3502,
+      "step": 14585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.931121840909799e-05,
+      "loss": 0.3244,
+      "step": 14590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.93105541663927e-05,
+      "loss": 0.3312,
+      "step": 14595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930988960803088e-05,
+      "loss": 0.2911,
+      "step": 14600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930922473402117e-05,
+      "loss": 0.3002,
+      "step": 14605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930855954437219e-05,
+      "loss": 0.2905,
+      "step": 14610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9307894039092596e-05,
+      "loss": 0.3078,
+      "step": 14615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9307228218191e-05,
+      "loss": 0.3168,
+      "step": 14620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9306562081676076e-05,
+      "loss": 0.3214,
+      "step": 14625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9305895629556464e-05,
+      "loss": 0.3257,
+      "step": 14630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.930522886184081e-05,
+      "loss": 0.2989,
+      "step": 14635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9304561778537786e-05,
+      "loss": 0.3381,
+      "step": 14640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9303894379656035e-05,
+      "loss": 0.3171,
+      "step": 14645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930322666520423e-05,
+      "loss": 0.3156,
+      "step": 14650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9302558635191054e-05,
+      "loss": 0.3314,
+      "step": 14655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9301890289625165e-05,
+      "loss": 0.316,
+      "step": 14660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930122162851524e-05,
+      "loss": 0.3382,
+      "step": 14665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930055265186997e-05,
+      "loss": 0.2921,
+      "step": 14670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929988335969805e-05,
+      "loss": 0.3044,
+      "step": 14675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929921375200814e-05,
+      "loss": 0.2977,
+      "step": 14680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9298543828808966e-05,
+      "loss": 0.3022,
+      "step": 14685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9297873590109213e-05,
+      "loss": 0.3132,
+      "step": 14690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9297203035917586e-05,
+      "loss": 0.3095,
+      "step": 14695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9296532166242785e-05,
+      "loss": 0.3058,
+      "step": 14700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929586098109352e-05,
+      "loss": 0.2842,
+      "step": 14705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929518948047852e-05,
+      "loss": 0.2967,
+      "step": 14710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92945176644065e-05,
+      "loss": 0.3238,
+      "step": 14715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9293845532886176e-05,
+      "loss": 0.3106,
+      "step": 14720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929317308592627e-05,
+      "loss": 0.3143,
+      "step": 14725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9292500323535526e-05,
+      "loss": 0.3155,
+      "step": 14730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9291827245722674e-05,
+      "loss": 0.2988,
+      "step": 14735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929115385249645e-05,
+      "loss": 0.3148,
+      "step": 14740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9290480143865616e-05,
+      "loss": 0.3172,
+      "step": 14745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928980611983889e-05,
+      "loss": 0.3113,
+      "step": 14750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928913178042505e-05,
+      "loss": 0.3493,
+      "step": 14755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928845712563284e-05,
+      "loss": 0.3253,
+      "step": 14760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928778215547101e-05,
+      "loss": 0.3245,
+      "step": 14765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928710686994835e-05,
+      "loss": 0.3032,
+      "step": 14770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9286431269073596e-05,
+      "loss": 0.2691,
+      "step": 14775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928575535285555e-05,
+      "loss": 0.2994,
+      "step": 14780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9285079121302974e-05,
+      "loss": 0.3101,
+      "step": 14785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928440257442465e-05,
+      "loss": 0.3046,
+      "step": 14790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928372571222935e-05,
+      "loss": 0.3185,
+      "step": 14795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9283048534725896e-05,
+      "loss": 0.3414,
+      "step": 14800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928237104192305e-05,
+      "loss": 0.3148,
+      "step": 14805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928169323382962e-05,
+      "loss": 0.3139,
+      "step": 14810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92810151104544e-05,
+      "loss": 0.3235,
+      "step": 14815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9280336671806206e-05,
+      "loss": 0.3351,
+      "step": 14820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927965791789384e-05,
+      "loss": 0.3248,
+      "step": 14825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927897884872612e-05,
+      "loss": 0.3035,
+      "step": 14830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927829946431186e-05,
+      "loss": 0.2935,
+      "step": 14835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927761976465988e-05,
+      "loss": 0.3099,
+      "step": 14840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9276939749779005e-05,
+      "loss": 0.3211,
+      "step": 14845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927625941967807e-05,
+      "loss": 0.3079,
+      "step": 14850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92755787743659e-05,
+      "loss": 0.3197,
+      "step": 14855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927489781385134e-05,
+      "loss": 0.3212,
+      "step": 14860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927421653814323e-05,
+      "loss": 0.3044,
+      "step": 14865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927353494725041e-05,
+      "loss": 0.3069,
+      "step": 14870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9272853041181744e-05,
+      "loss": 0.3359,
+      "step": 14875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927217081994607e-05,
+      "loss": 0.3104,
+      "step": 14880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9271488283552266e-05,
+      "loss": 0.3353,
+      "step": 14885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927080543200917e-05,
+      "loss": 0.3059,
+      "step": 14890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9270122265325666e-05,
+      "loss": 0.3228,
+      "step": 14895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926943878351062e-05,
+      "loss": 0.3114,
+      "step": 14900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926875498657291e-05,
+      "loss": 0.3156,
+      "step": 14905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92680708745214e-05,
+      "loss": 0.324,
+      "step": 14910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926738644736498e-05,
+      "loss": 0.3226,
+      "step": 14915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.926670170511255e-05,
+      "loss": 0.3128,
+      "step": 14920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9266016647772996e-05,
+      "loss": 0.3139,
+      "step": 14925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92653312753552e-05,
+      "loss": 0.2981,
+      "step": 14930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9264645587868064e-05,
+      "loss": 0.3413,
+      "step": 14935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92639595853205e-05,
+      "loss": 0.3121,
+      "step": 14940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926327326772141e-05,
+      "loss": 0.3223,
+      "step": 14945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926258663507971e-05,
+      "loss": 0.3568,
+      "step": 14950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926189968740431e-05,
+      "loss": 0.3296,
+      "step": 14955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926121242470413e-05,
+      "loss": 0.3021,
+      "step": 14960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92605248469881e-05,
+      "loss": 0.3203,
+      "step": 14965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925983695426514e-05,
+      "loss": 0.3302,
+      "step": 14970
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925914874654418e-05,
+      "loss": 0.3069,
+      "step": 14975
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925846022383417e-05,
+      "loss": 0.2889,
+      "step": 14980
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9257771386144033e-05,
+      "loss": 0.3106,
+      "step": 14985
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925708223348273e-05,
+      "loss": 0.2993,
+      "step": 14990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9256392765859194e-05,
+      "loss": 0.3323,
+      "step": 14995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9255702983282394e-05,
+      "loss": 0.3239,
+      "step": 15000
+    },
+    {
+      "epoch": 0.53,
+      "eval_loss": 0.3090079426765442,
+      "eval_runtime": 10.5523,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 15000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925501288576127e-05,
+      "loss": 0.3297,
+      "step": 15005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925432247330477e-05,
+      "loss": 0.3281,
+      "step": 15010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92536317459219e-05,
+      "loss": 0.3328,
+      "step": 15015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9252940703621596e-05,
+      "loss": 0.3205,
+      "step": 15020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925224934641284e-05,
+      "loss": 0.3209,
+      "step": 15025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925155767430462e-05,
+      "loss": 0.3308,
+      "step": 15030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925086568730589e-05,
+      "loss": 0.3384,
+      "step": 15035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925017338542566e-05,
+      "loss": 0.318,
+      "step": 15040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924948076867291e-05,
+      "loss": 0.3091,
+      "step": 15045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9248787837056633e-05,
+      "loss": 0.2807,
+      "step": 15050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9248094590585825e-05,
+      "loss": 0.3351,
+      "step": 15055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924740102926948e-05,
+      "loss": 0.3397,
+      "step": 15060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9246707153116624e-05,
+      "loss": 0.286,
+      "step": 15065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9246012962136256e-05,
+      "loss": 0.3172,
+      "step": 15070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9245318456337384e-05,
+      "loss": 0.3121,
+      "step": 15075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924462363572903e-05,
+      "loss": 0.3049,
+      "step": 15080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924392850032021e-05,
+      "loss": 0.3186,
+      "step": 15085
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924323305011996e-05,
+      "loss": 0.2936,
+      "step": 15090
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9242537285137314e-05,
+      "loss": 0.3289,
+      "step": 15095
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924184120538129e-05,
+      "loss": 0.3353,
+      "step": 15100
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924114481086094e-05,
+      "loss": 0.3201,
+      "step": 15105
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924044810158529e-05,
+      "loss": 0.3327,
+      "step": 15110
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9239751077563405e-05,
+      "loss": 0.3331,
+      "step": 15115
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923905373880433e-05,
+      "loss": 0.3006,
+      "step": 15120
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923835608531711e-05,
+      "loss": 0.3307,
+      "step": 15125
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9237658117110805e-05,
+      "loss": 0.3225,
+      "step": 15130
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923695983419449e-05,
+      "loss": 0.3058,
+      "step": 15135
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923626123657722e-05,
+      "loss": 0.307,
+      "step": 15140
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923556232426807e-05,
+      "loss": 0.3158,
+      "step": 15145
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923486309727612e-05,
+      "loss": 0.3194,
+      "step": 15150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923416355561044e-05,
+      "loss": 0.2916,
+      "step": 15155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923346369928012e-05,
+      "loss": 0.3148,
+      "step": 15160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923276352829424e-05,
+      "loss": 0.3263,
+      "step": 15165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92320630426619e-05,
+      "loss": 0.3075,
+      "step": 15170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9231362242392187e-05,
+      "loss": 0.3038,
+      "step": 15175
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92306611274942e-05,
+      "loss": 0.293,
+      "step": 15180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9229959697977065e-05,
+      "loss": 0.3103,
+      "step": 15185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9229257953849855e-05,
+      "loss": 0.324,
+      "step": 15190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9228555895121706e-05,
+      "loss": 0.2869,
+      "step": 15195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9227853521801726e-05,
+      "loss": 0.3081,
+      "step": 15200
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.922715083389903e-05,
+      "loss": 0.3291,
+      "step": 15205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922644783142274e-05,
+      "loss": 0.3184,
+      "step": 15210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9225744514382007e-05,
+      "loss": 0.3145,
+      "step": 15215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922504088278593e-05,
+      "loss": 0.3139,
+      "step": 15220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922433693664368e-05,
+      "loss": 0.3348,
+      "step": 15225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922363267596437e-05,
+      "loss": 0.3065,
+      "step": 15230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922292810075715e-05,
+      "loss": 0.3147,
+      "step": 15235
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9222223211031185e-05,
+      "loss": 0.3264,
+      "step": 15240
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9221518006795595e-05,
+      "loss": 0.2945,
+      "step": 15245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922081248805958e-05,
+      "loss": 0.2911,
+      "step": 15250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.922010665483226e-05,
+      "loss": 0.3106,
+      "step": 15255
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921940050712281e-05,
+      "loss": 0.3063,
+      "step": 15260
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9218694044940424e-05,
+      "loss": 0.3207,
+      "step": 15265
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9217987268294244e-05,
+      "loss": 0.2964,
+      "step": 15270
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921728017719347e-05,
+      "loss": 0.344,
+      "step": 15275
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921657277164726e-05,
+      "loss": 0.2759,
+      "step": 15280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9215865051664824e-05,
+      "loss": 0.3429,
+      "step": 15285
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9215157017255333e-05,
+      "loss": 0.3455,
+      "step": 15290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921444866842799e-05,
+      "loss": 0.3217,
+      "step": 15295
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921374000519199e-05,
+      "loss": 0.3146,
+      "step": 15300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9213031027556526e-05,
+      "loss": 0.294,
+      "step": 15305
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921232173553082e-05,
+      "loss": 0.3116,
+      "step": 15310
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921161212912406e-05,
+      "loss": 0.332,
+      "step": 15315
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9210902208345486e-05,
+      "loss": 0.3257,
+      "step": 15320
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.921019197320429e-05,
+      "loss": 0.3238,
+      "step": 15325
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920948142370972e-05,
+      "loss": 0.2883,
+      "step": 15330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9208770559870984e-05,
+      "loss": 0.3216,
+      "step": 15335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9208059381697314e-05,
+      "loss": 0.3331,
+      "step": 15340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920734788919794e-05,
+      "loss": 0.3136,
+      "step": 15345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9206636082382115e-05,
+      "loss": 0.2992,
+      "step": 15350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920592396125907e-05,
+      "loss": 0.3019,
+      "step": 15355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920521152583806e-05,
+      "loss": 0.3563,
+      "step": 15360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920449877612833e-05,
+      "loss": 0.3187,
+      "step": 15365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9203785712139126e-05,
+      "loss": 0.3117,
+      "step": 15370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920307233387973e-05,
+      "loss": 0.3155,
+      "step": 15375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9202358641359375e-05,
+      "loss": 0.305,
+      "step": 15380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9201644634587354e-05,
+      "loss": 0.29,
+      "step": 15385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.920093031357292e-05,
+      "loss": 0.3041,
+      "step": 15390
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9200215678325354e-05,
+      "loss": 0.3318,
+      "step": 15395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9199500728853944e-05,
+      "loss": 0.3196,
+      "step": 15400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919878546516795e-05,
+      "loss": 0.3327,
+      "step": 15405
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919806988727669e-05,
+      "loss": 0.2934,
+      "step": 15410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9197353995189434e-05,
+      "loss": 0.2936,
+      "step": 15415
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9196637788915486e-05,
+      "loss": 0.3291,
+      "step": 15420
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9195921268464143e-05,
+      "loss": 0.3002,
+      "step": 15425
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.91952044338447e-05,
+      "loss": 0.3251,
+      "step": 15430
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9194487285066474e-05,
+      "loss": 0.3033,
+      "step": 15435
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919376982213878e-05,
+      "loss": 0.284,
+      "step": 15440
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919305204507093e-05,
+      "loss": 0.3229,
+      "step": 15445
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919233395387224e-05,
+      "loss": 0.3252,
+      "step": 15450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919161554855204e-05,
+      "loss": 0.3176,
+      "step": 15455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919089682911965e-05,
+      "loss": 0.3115,
+      "step": 15460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.919017779558441e-05,
+      "loss": 0.3387,
+      "step": 15465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.918945844795565e-05,
+      "loss": 0.301,
+      "step": 15470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.918873878624272e-05,
+      "loss": 0.3064,
+      "step": 15475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.918801881045495e-05,
+      "loss": 0.3127,
+      "step": 15480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.91872985206017e-05,
+      "loss": 0.3176,
+      "step": 15485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9186577916692325e-05,
+      "loss": 0.3175,
+      "step": 15490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918585699873617e-05,
+      "loss": 0.3359,
+      "step": 15495
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918513576674259e-05,
+      "loss": 0.3018,
+      "step": 15500
+    },
+    {
+      "epoch": 0.55,
+      "eval_loss": 0.309041827917099,
+      "eval_runtime": 10.5379,
+      "eval_samples_per_second": 9.49,
+      "eval_steps_per_second": 9.49,
+      "step": 15500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918441422072098e-05,
+      "loss": 0.3475,
+      "step": 15505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9183692360680676e-05,
+      "loss": 0.3085,
+      "step": 15510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918297018663107e-05,
+      "loss": 0.3378,
+      "step": 15515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918224769858153e-05,
+      "loss": 0.3255,
+      "step": 15520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918152489654145e-05,
+      "loss": 0.3167,
+      "step": 15525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918080178052019e-05,
+      "loss": 0.3193,
+      "step": 15530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.918007835052717e-05,
+      "loss": 0.311,
+      "step": 15535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917935460657176e-05,
+      "loss": 0.3121,
+      "step": 15540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917863054866336e-05,
+      "loss": 0.3069,
+      "step": 15545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9177906176811384e-05,
+      "loss": 0.2918,
+      "step": 15550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917718149102523e-05,
+      "loss": 0.2929,
+      "step": 15555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9176456491314304e-05,
+      "loss": 0.3343,
+      "step": 15560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917573117768803e-05,
+      "loss": 0.2854,
+      "step": 15565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917500555015582e-05,
+      "loss": 0.3236,
+      "step": 15570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917427960872708e-05,
+      "loss": 0.3021,
+      "step": 15575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917355335341126e-05,
+      "loss": 0.3098,
+      "step": 15580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9172826784217786e-05,
+      "loss": 0.3106,
+      "step": 15585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9172099901156076e-05,
+      "loss": 0.3122,
+      "step": 15590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9171372704235584e-05,
+      "loss": 0.2936,
+      "step": 15595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.917064519346575e-05,
+      "loss": 0.329,
+      "step": 15600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916991736885602e-05,
+      "loss": 0.285,
+      "step": 15605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916918923041584e-05,
+      "loss": 0.3033,
+      "step": 15610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916846077815466e-05,
+      "loss": 0.3181,
+      "step": 15615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9167732012081944e-05,
+      "loss": 0.3313,
+      "step": 15620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9167002932207156e-05,
+      "loss": 0.2979,
+      "step": 15625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916627353853976e-05,
+      "loss": 0.3045,
+      "step": 15630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916554383108923e-05,
+      "loss": 0.3124,
+      "step": 15635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916481380986504e-05,
+      "loss": 0.3113,
+      "step": 15640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9164083474876675e-05,
+      "loss": 0.3283,
+      "step": 15645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.91633528261336e-05,
+      "loss": 0.3191,
+      "step": 15650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9162621863645317e-05,
+      "loss": 0.3268,
+      "step": 15655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916189058742131e-05,
+      "loss": 0.2921,
+      "step": 15660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9161158997471086e-05,
+      "loss": 0.3122,
+      "step": 15665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.916042709380413e-05,
+      "loss": 0.3073,
+      "step": 15670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915969487642995e-05,
+      "loss": 0.2936,
+      "step": 15675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9159108876667794e-05,
+      "loss": 0.3335,
+      "step": 15680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915837609464457e-05,
+      "loss": 0.3175,
+      "step": 15685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915764299894076e-05,
+      "loss": 0.2781,
+      "step": 15690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9156909589565866e-05,
+      "loss": 0.3107,
+      "step": 15695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915617586652943e-05,
+      "loss": 0.3128,
+      "step": 15700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915544182984096e-05,
+      "loss": 0.3059,
+      "step": 15705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915470747951001e-05,
+      "loss": 0.3112,
+      "step": 15710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.91539728155461e-05,
+      "loss": 0.3301,
+      "step": 15715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915323783795876e-05,
+      "loss": 0.3488,
+      "step": 15720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.915250254675755e-05,
+      "loss": 0.3322,
+      "step": 15725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9151766941952015e-05,
+      "loss": 0.3113,
+      "step": 15730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.91510310235517e-05,
+      "loss": 0.3019,
+      "step": 15735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9150294791566167e-05,
+      "loss": 0.2977,
+      "step": 15740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9149558246004964e-05,
+      "loss": 0.329,
+      "step": 15745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914882138687767e-05,
+      "loss": 0.3104,
+      "step": 15750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914808421419384e-05,
+      "loss": 0.3201,
+      "step": 15755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914734672796305e-05,
+      "loss": 0.2997,
+      "step": 15760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914660892819488e-05,
+      "loss": 0.3156,
+      "step": 15765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.914587081489891e-05,
+      "loss": 0.3139,
+      "step": 15770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914513238808471e-05,
+      "loss": 0.3022,
+      "step": 15775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914439364776189e-05,
+      "loss": 0.2884,
+      "step": 15780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914365459394002e-05,
+      "loss": 0.3486,
+      "step": 15785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914291522662871e-05,
+      "loss": 0.3116,
+      "step": 15790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9142175545837555e-05,
+      "loss": 0.3308,
+      "step": 15795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9141435551576165e-05,
+      "loss": 0.319,
+      "step": 15800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.914069524385414e-05,
+      "loss": 0.3092,
+      "step": 15805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9139954622681094e-05,
+      "loss": 0.3191,
+      "step": 15810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9139213688066656e-05,
+      "loss": 0.328,
+      "step": 15815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913847244002043e-05,
+      "loss": 0.3296,
+      "step": 15820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9137730878552044e-05,
+      "loss": 0.3267,
+      "step": 15825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913698900367114e-05,
+      "loss": 0.3038,
+      "step": 15830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9136246815387334e-05,
+      "loss": 0.3019,
+      "step": 15835
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913550431371027e-05,
+      "loss": 0.2831,
+      "step": 15840
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9134761498649586e-05,
+      "loss": 0.3151,
+      "step": 15845
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913401837021493e-05,
+      "loss": 0.3058,
+      "step": 15850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913327492841595e-05,
+      "loss": 0.2843,
+      "step": 15855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9132531173262306e-05,
+      "loss": 0.2963,
+      "step": 15860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913178710476365e-05,
+      "loss": 0.3378,
+      "step": 15865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913104272292963e-05,
+      "loss": 0.3084,
+      "step": 15870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.913029802776993e-05,
+      "loss": 0.3009,
+      "step": 15875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9129553019294214e-05,
+      "loss": 0.3213,
+      "step": 15880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9128807697512146e-05,
+      "loss": 0.3199,
+      "step": 15885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912806206243342e-05,
+      "loss": 0.3262,
+      "step": 15890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9127316114067715e-05,
+      "loss": 0.3089,
+      "step": 15895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91265698524247e-05,
+      "loss": 0.3164,
+      "step": 15900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9125823277514075e-05,
+      "loss": 0.322,
+      "step": 15905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9125076389345534e-05,
+      "loss": 0.3247,
+      "step": 15910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912432918792878e-05,
+      "loss": 0.2992,
+      "step": 15915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912358167327351e-05,
+      "loss": 0.3167,
+      "step": 15920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9122833845389426e-05,
+      "loss": 0.3208,
+      "step": 15925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9122085704286246e-05,
+      "loss": 0.2921,
+      "step": 15930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9121337249973675e-05,
+      "loss": 0.3092,
+      "step": 15935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.912058848246144e-05,
+      "loss": 0.3083,
+      "step": 15940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911983940175926e-05,
+      "loss": 0.2973,
+      "step": 15945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9119090007876866e-05,
+      "loss": 0.326,
+      "step": 15950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911834030082397e-05,
+      "loss": 0.3251,
+      "step": 15955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911759028061033e-05,
+      "loss": 0.3269,
+      "step": 15960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9116839947245677e-05,
+      "loss": 0.3348,
+      "step": 15965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9116089300739745e-05,
+      "loss": 0.3269,
+      "step": 15970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9115338341102287e-05,
+      "loss": 0.306,
+      "step": 15975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911458706834305e-05,
+      "loss": 0.2972,
+      "step": 15980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9113835482471796e-05,
+      "loss": 0.3247,
+      "step": 15985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911308358349828e-05,
+      "loss": 0.336,
+      "step": 15990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911233137143226e-05,
+      "loss": 0.3204,
+      "step": 15995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911157884628352e-05,
+      "loss": 0.3203,
+      "step": 16000
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 0.30903559923171997,
+      "eval_runtime": 10.5428,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 16000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9110826008061804e-05,
+      "loss": 0.2991,
+      "step": 16005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911007285677691e-05,
+      "loss": 0.3346,
+      "step": 16010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910931939243861e-05,
+      "loss": 0.3237,
+      "step": 16015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910856561505668e-05,
+      "loss": 0.2991,
+      "step": 16020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9107811524640926e-05,
+      "loss": 0.3116,
+      "step": 16025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9107057121201116e-05,
+      "loss": 0.3334,
+      "step": 16030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9106302404747064e-05,
+      "loss": 0.2986,
+      "step": 16035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.9105547375288553e-05,
+      "loss": 0.3229,
+      "step": 16040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.91047920328354e-05,
+      "loss": 0.3129,
+      "step": 16045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910403637739741e-05,
+      "loss": 0.337,
+      "step": 16050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.910328040898439e-05,
+      "loss": 0.3069,
+      "step": 16055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9102524127606156e-05,
+      "loss": 0.3133,
+      "step": 16060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.910176753327254e-05,
+      "loss": 0.3323,
+      "step": 16065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.910101062599334e-05,
+      "loss": 0.3099,
+      "step": 16070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.910025340577841e-05,
+      "loss": 0.2945,
+      "step": 16075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9099495872637574e-05,
+      "loss": 0.3001,
+      "step": 16080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909873802658067e-05,
+      "loss": 0.3157,
+      "step": 16085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909797986761753e-05,
+      "loss": 0.3145,
+      "step": 16090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9097221395758e-05,
+      "loss": 0.3083,
+      "step": 16095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909646261101193e-05,
+      "loss": 0.3221,
+      "step": 16100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909570351338917e-05,
+      "loss": 0.3278,
+      "step": 16105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9094944102899585e-05,
+      "loss": 0.322,
+      "step": 16110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909418437955303e-05,
+      "loss": 0.3157,
+      "step": 16115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909342434335937e-05,
+      "loss": 0.3012,
+      "step": 16120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909266399432847e-05,
+      "loss": 0.3021,
+      "step": 16125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9091903332470215e-05,
+      "loss": 0.3079,
+      "step": 16130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909114235779446e-05,
+      "loss": 0.3236,
+      "step": 16135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.909038107031111e-05,
+      "loss": 0.3187,
+      "step": 16140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9089619470030025e-05,
+      "loss": 0.3149,
+      "step": 16145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908885755696111e-05,
+      "loss": 0.2911,
+      "step": 16150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908809533111427e-05,
+      "loss": 0.3365,
+      "step": 16155
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908733279249937e-05,
+      "loss": 0.272,
+      "step": 16160
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9086569941126335e-05,
+      "loss": 0.3254,
+      "step": 16165
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9085806777005054e-05,
+      "loss": 0.2899,
+      "step": 16170
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9085043300145453e-05,
+      "loss": 0.3273,
+      "step": 16175
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908427951055744e-05,
+      "loss": 0.2825,
+      "step": 16180
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908351540825093e-05,
+      "loss": 0.304,
+      "step": 16185
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9082750993235835e-05,
+      "loss": 0.294,
+      "step": 16190
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.90819862655221e-05,
+      "loss": 0.3118,
+      "step": 16195
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.908122122511963e-05,
+      "loss": 0.338,
+      "step": 16200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9080455872038385e-05,
+      "loss": 0.3199,
+      "step": 16205
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9079690206288284e-05,
+      "loss": 0.3157,
+      "step": 16210
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9078924227879274e-05,
+      "loss": 0.2938,
+      "step": 16215
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907815793682131e-05,
+      "loss": 0.3026,
+      "step": 16220
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9077391333124323e-05,
+      "loss": 0.2991,
+      "step": 16225
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9076624416798285e-05,
+      "loss": 0.3261,
+      "step": 16230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907585718785314e-05,
+      "loss": 0.3206,
+      "step": 16235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907508964629886e-05,
+      "loss": 0.3247,
+      "step": 16240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9074321792145405e-05,
+      "loss": 0.2934,
+      "step": 16245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907355362540275e-05,
+      "loss": 0.3117,
+      "step": 16250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907278514608086e-05,
+      "loss": 0.3057,
+      "step": 16255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9072016354189725e-05,
+      "loss": 0.3288,
+      "step": 16260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9071247249739324e-05,
+      "loss": 0.3158,
+      "step": 16265
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.907047783273963e-05,
+      "loss": 0.3155,
+      "step": 16270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9069708103200655e-05,
+      "loss": 0.3297,
+      "step": 16275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906893806113238e-05,
+      "loss": 0.3117,
+      "step": 16280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906816770654481e-05,
+      "loss": 0.3229,
+      "step": 16285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906739703944794e-05,
+      "loss": 0.2878,
+      "step": 16290
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9066626059851775e-05,
+      "loss": 0.3256,
+      "step": 16295
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906585476776634e-05,
+      "loss": 0.31,
+      "step": 16300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906508316320163e-05,
+      "loss": 0.3165,
+      "step": 16305
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9064311246167684e-05,
+      "loss": 0.3078,
+      "step": 16310
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9063539016674517e-05,
+      "loss": 0.3042,
+      "step": 16315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9062766474732146e-05,
+      "loss": 0.2994,
+      "step": 16320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.906199362035062e-05,
+      "loss": 0.293,
+      "step": 16325
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9061220453539955e-05,
+      "loss": 0.3131,
+      "step": 16330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.9060446974310206e-05,
+      "loss": 0.3289,
+      "step": 16335
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.90596731826714e-05,
+      "loss": 0.2934,
+      "step": 16340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905889907863361e-05,
+      "loss": 0.302,
+      "step": 16345
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9058124662206864e-05,
+      "loss": 0.2956,
+      "step": 16350
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9057349933401213e-05,
+      "loss": 0.3129,
+      "step": 16355
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905657489222674e-05,
+      "loss": 0.3028,
+      "step": 16360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905579953869349e-05,
+      "loss": 0.2844,
+      "step": 16365
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905502387281154e-05,
+      "loss": 0.2775,
+      "step": 16370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905424789459096e-05,
+      "loss": 0.3449,
+      "step": 16375
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9053471604041815e-05,
+      "loss": 0.3173,
+      "step": 16380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9052695001174196e-05,
+      "loss": 0.286,
+      "step": 16385
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9051918085998184e-05,
+      "loss": 0.3029,
+      "step": 16390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905114085852387e-05,
+      "loss": 0.2968,
+      "step": 16395
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.905036331876134e-05,
+      "loss": 0.3236,
+      "step": 16400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9049585466720695e-05,
+      "loss": 0.2975,
+      "step": 16405
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904880730241202e-05,
+      "loss": 0.3245,
+      "step": 16410
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904802882584545e-05,
+      "loss": 0.3016,
+      "step": 16415
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9047250037031064e-05,
+      "loss": 0.3133,
+      "step": 16420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9046470935978986e-05,
+      "loss": 0.3239,
+      "step": 16425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904569152269932e-05,
+      "loss": 0.3015,
+      "step": 16430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904491179720221e-05,
+      "loss": 0.3449,
+      "step": 16435
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9044131759497755e-05,
+      "loss": 0.2949,
+      "step": 16440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904335140959611e-05,
+      "loss": 0.2984,
+      "step": 16445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904257074750738e-05,
+      "loss": 0.3246,
+      "step": 16450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904178977324171e-05,
+      "loss": 0.3193,
+      "step": 16455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.904100848680925e-05,
+      "loss": 0.3002,
+      "step": 16460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9040226888220144e-05,
+      "loss": 0.2971,
+      "step": 16465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903944497748453e-05,
+      "loss": 0.3081,
+      "step": 16470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9038662754612564e-05,
+      "loss": 0.3071,
+      "step": 16475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9037880219614405e-05,
+      "loss": 0.3106,
+      "step": 16480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903709737250021e-05,
+      "loss": 0.3069,
+      "step": 16485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9036314213280153e-05,
+      "loss": 0.3297,
+      "step": 16490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903553074196439e-05,
+      "loss": 0.3066,
+      "step": 16495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.90347469585631e-05,
+      "loss": 0.2976,
+      "step": 16500
+    },
+    {
+      "epoch": 0.58,
+      "eval_loss": 0.30791518092155457,
+      "eval_runtime": 10.5594,
+      "eval_samples_per_second": 9.47,
+      "eval_steps_per_second": 9.47,
+      "step": 16500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9033962863086466e-05,
+      "loss": 0.304,
+      "step": 16505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903317845554466e-05,
+      "loss": 0.3337,
+      "step": 16510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903239373594788e-05,
+      "loss": 0.3248,
+      "step": 16515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.90316087043063e-05,
+      "loss": 0.2933,
+      "step": 16520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9030823360630106e-05,
+      "loss": 0.3117,
+      "step": 16525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.903003770492952e-05,
+      "loss": 0.3175,
+      "step": 16530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902925173721473e-05,
+      "loss": 0.3121,
+      "step": 16535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9028465457495936e-05,
+      "loss": 0.2972,
+      "step": 16540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9027678865783354e-05,
+      "loss": 0.3076,
+      "step": 16545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.90268919620872e-05,
+      "loss": 0.3243,
+      "step": 16550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902610474641769e-05,
+      "loss": 0.3134,
+      "step": 16555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902531721878504e-05,
+      "loss": 0.3299,
+      "step": 16560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902452937919948e-05,
+      "loss": 0.3056,
+      "step": 16565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9023741227671236e-05,
+      "loss": 0.336,
+      "step": 16570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902295276421055e-05,
+      "loss": 0.3167,
+      "step": 16575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902216398882766e-05,
+      "loss": 0.29,
+      "step": 16580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.902137490153279e-05,
+      "loss": 0.3181,
+      "step": 16585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9020585502336205e-05,
+      "loss": 0.2852,
+      "step": 16590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9019795791248144e-05,
+      "loss": 0.3185,
+      "step": 16595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.901900576827886e-05,
+      "loss": 0.3138,
+      "step": 16600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9018215433438616e-05,
+      "loss": 0.2948,
+      "step": 16605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9017424786737674e-05,
+      "loss": 0.3113,
+      "step": 16610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9016633828186306e-05,
+      "loss": 0.2955,
+      "step": 16615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.901584255779477e-05,
+      "loss": 0.3217,
+      "step": 16620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.9015209316963526e-05,
+      "loss": 0.3111,
+      "step": 16625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9014417485285595e-05,
+      "loss": 0.2994,
+      "step": 16630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9013625341796276e-05,
+      "loss": 0.3065,
+      "step": 16635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9012832886505855e-05,
+      "loss": 0.3198,
+      "step": 16640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901204011942463e-05,
+      "loss": 0.3359,
+      "step": 16645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901124704056289e-05,
+      "loss": 0.3179,
+      "step": 16650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.901045364993092e-05,
+      "loss": 0.2868,
+      "step": 16655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900965994753904e-05,
+      "loss": 0.3325,
+      "step": 16660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9008865933397547e-05,
+      "loss": 0.3052,
+      "step": 16665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9008071607516746e-05,
+      "loss": 0.2971,
+      "step": 16670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900727696990697e-05,
+      "loss": 0.3172,
+      "step": 16675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900648202057851e-05,
+      "loss": 0.3044,
+      "step": 16680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9005686759541714e-05,
+      "loss": 0.3029,
+      "step": 16685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9004891186806894e-05,
+      "loss": 0.3203,
+      "step": 16690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900409530238438e-05,
+      "loss": 0.301,
+      "step": 16695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9003299106284504e-05,
+      "loss": 0.3246,
+      "step": 16700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900250259851762e-05,
+      "loss": 0.3039,
+      "step": 16705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900170577909404e-05,
+      "loss": 0.2734,
+      "step": 16710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.900090864802414e-05,
+      "loss": 0.2907,
+      "step": 16715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.9000111205318254e-05,
+      "loss": 0.3238,
+      "step": 16720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899931345098674e-05,
+      "loss": 0.3159,
+      "step": 16725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899851538503996e-05,
+      "loss": 0.3073,
+      "step": 16730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899771700748827e-05,
+      "loss": 0.3185,
+      "step": 16735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899691831834205e-05,
+      "loss": 0.2971,
+      "step": 16740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899611931761164e-05,
+      "loss": 0.31,
+      "step": 16745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899532000530746e-05,
+      "loss": 0.3042,
+      "step": 16750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899452038143985e-05,
+      "loss": 0.3021,
+      "step": 16755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.89937204460192e-05,
+      "loss": 0.3235,
+      "step": 16760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899292019905591e-05,
+      "loss": 0.2933,
+      "step": 16765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899211964056036e-05,
+      "loss": 0.3161,
+      "step": 16770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899131877054295e-05,
+      "loss": 0.3139,
+      "step": 16775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.899051758901407e-05,
+      "loss": 0.3109,
+      "step": 16780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898971609598414e-05,
+      "loss": 0.3411,
+      "step": 16785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898891429146355e-05,
+      "loss": 0.3293,
+      "step": 16790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898811217546272e-05,
+      "loss": 0.327,
+      "step": 16795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8987309747992064e-05,
+      "loss": 0.3188,
+      "step": 16800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8986507009061994e-05,
+      "loss": 0.2961,
+      "step": 16805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898570395868294e-05,
+      "loss": 0.3104,
+      "step": 16810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898490059686532e-05,
+      "loss": 0.3094,
+      "step": 16815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8984096923619586e-05,
+      "loss": 0.3213,
+      "step": 16820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898329293895616e-05,
+      "loss": 0.3141,
+      "step": 16825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898248864288547e-05,
+      "loss": 0.3198,
+      "step": 16830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898168403541797e-05,
+      "loss": 0.3185,
+      "step": 16835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8980879116564114e-05,
+      "loss": 0.3084,
+      "step": 16840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.898007388633434e-05,
+      "loss": 0.3077,
+      "step": 16845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897926834473912e-05,
+      "loss": 0.3165,
+      "step": 16850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.89784624917889e-05,
+      "loss": 0.3233,
+      "step": 16855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8977656327494145e-05,
+      "loss": 0.3296,
+      "step": 16860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897684985186533e-05,
+      "loss": 0.2964,
+      "step": 16865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8976043064912916e-05,
+      "loss": 0.3053,
+      "step": 16870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8975235966647376e-05,
+      "loss": 0.2974,
+      "step": 16875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8974428557079204e-05,
+      "loss": 0.3186,
+      "step": 16880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897362083621888e-05,
+      "loss": 0.294,
+      "step": 16885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.8972812804076884e-05,
+      "loss": 0.3269,
+      "step": 16890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897200446066371e-05,
+      "loss": 0.3195,
+      "step": 16895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897119580598987e-05,
+      "loss": 0.3059,
+      "step": 16900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.897038684006583e-05,
+      "loss": 0.2914,
+      "step": 16905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.896957756290213e-05,
+      "loss": 0.3063,
+      "step": 16910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896876797450926e-05,
+      "loss": 0.3383,
+      "step": 16915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8967958074897724e-05,
+      "loss": 0.3196,
+      "step": 16920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896714786407806e-05,
+      "loss": 0.3065,
+      "step": 16925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8966337342060765e-05,
+      "loss": 0.3096,
+      "step": 16930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8965526508856387e-05,
+      "loss": 0.2989,
+      "step": 16935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896471536447543e-05,
+      "loss": 0.3001,
+      "step": 16940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896390390892844e-05,
+      "loss": 0.3072,
+      "step": 16945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896309214222595e-05,
+      "loss": 0.2968,
+      "step": 16950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8962280064378504e-05,
+      "loss": 0.2986,
+      "step": 16955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896146767539664e-05,
+      "loss": 0.2951,
+      "step": 16960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.896065497529091e-05,
+      "loss": 0.3021,
+      "step": 16965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8959841964071864e-05,
+      "loss": 0.3164,
+      "step": 16970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895902864175006e-05,
+      "loss": 0.329,
+      "step": 16975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895821500833606e-05,
+      "loss": 0.3042,
+      "step": 16980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8957401063840435e-05,
+      "loss": 0.3218,
+      "step": 16985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895658680827374e-05,
+      "loss": 0.315,
+      "step": 16990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8955772241646546e-05,
+      "loss": 0.3186,
+      "step": 16995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8954957363969445e-05,
+      "loss": 0.3001,
+      "step": 17000
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 0.30690938234329224,
+      "eval_runtime": 10.5382,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 17000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895414217525301e-05,
+      "loss": 0.3164,
+      "step": 17005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895332667550783e-05,
+      "loss": 0.2993,
+      "step": 17010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895251086474447e-05,
+      "loss": 0.3298,
+      "step": 17015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895169474297356e-05,
+      "loss": 0.3277,
+      "step": 17020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895087831020567e-05,
+      "loss": 0.3215,
+      "step": 17025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.895006156645141e-05,
+      "loss": 0.332,
+      "step": 17030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8949244511721385e-05,
+      "loss": 0.3016,
+      "step": 17035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894842714602621e-05,
+      "loss": 0.3298,
+      "step": 17040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894760946937648e-05,
+      "loss": 0.3033,
+      "step": 17045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8946791481782826e-05,
+      "loss": 0.3404,
+      "step": 17050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894597318325587e-05,
+      "loss": 0.3124,
+      "step": 17055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8945154573806235e-05,
+      "loss": 0.3361,
+      "step": 17060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8944335653444544e-05,
+      "loss": 0.2948,
+      "step": 17065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8943516422181435e-05,
+      "loss": 0.3212,
+      "step": 17070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894269688002755e-05,
+      "loss": 0.3018,
+      "step": 17075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8941877026993514e-05,
+      "loss": 0.3198,
+      "step": 17080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894105686308999e-05,
+      "loss": 0.2923,
+      "step": 17085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.894023638832762e-05,
+      "loss": 0.3264,
+      "step": 17090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8939415602717056e-05,
+      "loss": 0.3026,
+      "step": 17095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893859450626896e-05,
+      "loss": 0.3145,
+      "step": 17100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893777309899399e-05,
+      "loss": 0.3186,
+      "step": 17105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8936951380902804e-05,
+      "loss": 0.3439,
+      "step": 17110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893612935200609e-05,
+      "loss": 0.3224,
+      "step": 17115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.89353070123145e-05,
+      "loss": 0.3126,
+      "step": 17120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893448436183873e-05,
+      "loss": 0.314,
+      "step": 17125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893366140058946e-05,
+      "loss": 0.3017,
+      "step": 17130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893283812857736e-05,
+      "loss": 0.3289,
+      "step": 17135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8932014545813134e-05,
+      "loss": 0.3609,
+      "step": 17140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.893119065230747e-05,
+      "loss": 0.2998,
+      "step": 17145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8930366448071066e-05,
+      "loss": 0.351,
+      "step": 17150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892954193311462e-05,
+      "loss": 0.3124,
+      "step": 17155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892871710744884e-05,
+      "loss": 0.3311,
+      "step": 17160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892789197108445e-05,
+      "loss": 0.3135,
+      "step": 17165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892706652403215e-05,
+      "loss": 0.3331,
+      "step": 17170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892624076630265e-05,
+      "loss": 0.3063,
+      "step": 17175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892541469790669e-05,
+      "loss": 0.3259,
+      "step": 17180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892458831885498e-05,
+      "loss": 0.321,
+      "step": 17185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.892376162915826e-05,
+      "loss": 0.2919,
+      "step": 17190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.8922934628827263e-05,
+      "loss": 0.314,
+      "step": 17195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.892210731787273e-05,
+      "loss": 0.3051,
+      "step": 17200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.89212796963054e-05,
+      "loss": 0.316,
+      "step": 17205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.892045176413601e-05,
+      "loss": 0.305,
+      "step": 17210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8919623521375316e-05,
+      "loss": 0.307,
+      "step": 17215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8918794968034084e-05,
+      "loss": 0.3026,
+      "step": 17220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891796610412306e-05,
+      "loss": 0.2894,
+      "step": 17225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8917136929653007e-05,
+      "loss": 0.2895,
+      "step": 17230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891630744463469e-05,
+      "loss": 0.3233,
+      "step": 17235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891547764907888e-05,
+      "loss": 0.3337,
+      "step": 17240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891464754299636e-05,
+      "loss": 0.2812,
+      "step": 17245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.89138171263979e-05,
+      "loss": 0.3134,
+      "step": 17250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8912986399294285e-05,
+      "loss": 0.3149,
+      "step": 17255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.89121553616963e-05,
+      "loss": 0.3306,
+      "step": 17260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.891132401361473e-05,
+      "loss": 0.3212,
+      "step": 17265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8910492355060386e-05,
+      "loss": 0.3005,
+      "step": 17270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8909660386044055e-05,
+      "loss": 0.3292,
+      "step": 17275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890882810657654e-05,
+      "loss": 0.337,
+      "step": 17280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890799551666865e-05,
+      "loss": 0.3018,
+      "step": 17285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890716261633119e-05,
+      "loss": 0.3002,
+      "step": 17290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890632940557498e-05,
+      "loss": 0.2851,
+      "step": 17295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890549588441083e-05,
+      "loss": 0.3158,
+      "step": 17300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8904662052849586e-05,
+      "loss": 0.3095,
+      "step": 17305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890382791090205e-05,
+      "loss": 0.3175,
+      "step": 17310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890299345857907e-05,
+      "loss": 0.3159,
+      "step": 17315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8902158695891464e-05,
+      "loss": 0.3371,
+      "step": 17320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890132362285008e-05,
+      "loss": 0.3326,
+      "step": 17325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.890048823946576e-05,
+      "loss": 0.3218,
+      "step": 17330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889965254574935e-05,
+      "loss": 0.3009,
+      "step": 17335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8898816541711706e-05,
+      "loss": 0.3419,
+      "step": 17340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8897980227363684e-05,
+      "loss": 0.3228,
+      "step": 17345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889714360271612e-05,
+      "loss": 0.3057,
+      "step": 17350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889630666777991e-05,
+      "loss": 0.2896,
+      "step": 17355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889546942256591e-05,
+      "loss": 0.3151,
+      "step": 17360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889463186708498e-05,
+      "loss": 0.3091,
+      "step": 17365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8893794001348e-05,
+      "loss": 0.3189,
+      "step": 17370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8892955825365856e-05,
+      "loss": 0.3118,
+      "step": 17375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889211733914942e-05,
+      "loss": 0.2891,
+      "step": 17380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.88912785427096e-05,
+      "loss": 0.2917,
+      "step": 17385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.889043943605726e-05,
+      "loss": 0.3112,
+      "step": 17390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888960001920331e-05,
+      "loss": 0.3095,
+      "step": 17395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8888760292158654e-05,
+      "loss": 0.3276,
+      "step": 17400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888792025493418e-05,
+      "loss": 0.3031,
+      "step": 17405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888707990754081e-05,
+      "loss": 0.315,
+      "step": 17410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8886239249989455e-05,
+      "loss": 0.3281,
+      "step": 17415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8885398282291015e-05,
+      "loss": 0.2979,
+      "step": 17420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888455700445642e-05,
+      "loss": 0.3312,
+      "step": 17425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.88837154164966e-05,
+      "loss": 0.3159,
+      "step": 17430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888287351842247e-05,
+      "loss": 0.3123,
+      "step": 17435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888203131024497e-05,
+      "loss": 0.3222,
+      "step": 17440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888118879197502e-05,
+      "loss": 0.313,
+      "step": 17445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.888034596362359e-05,
+      "loss": 0.3172,
+      "step": 17450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.887950282520159e-05,
+      "loss": 0.3118,
+      "step": 17455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8878659376719995e-05,
+      "loss": 0.2852,
+      "step": 17460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.887781561818974e-05,
+      "loss": 0.3064,
+      "step": 17465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8876971549621784e-05,
+      "loss": 0.3038,
+      "step": 17470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.8876127171027084e-05,
+      "loss": 0.3113,
+      "step": 17475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.887528248241662e-05,
+      "loss": 0.3153,
+      "step": 17480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8874437483801335e-05,
+      "loss": 0.3126,
+      "step": 17485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887359217519222e-05,
+      "loss": 0.3134,
+      "step": 17490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887274655660025e-05,
+      "loss": 0.2875,
+      "step": 17495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887190062803638e-05,
+      "loss": 0.3005,
+      "step": 17500
+    },
+    {
+      "epoch": 0.62,
+      "eval_loss": 0.3056062161922455,
+      "eval_runtime": 10.5404,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 17500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8871054389511626e-05,
+      "loss": 0.3079,
+      "step": 17505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.887020784103696e-05,
+      "loss": 0.3054,
+      "step": 17510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886936098262338e-05,
+      "loss": 0.3035,
+      "step": 17515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8868513814281866e-05,
+      "loss": 0.3326,
+      "step": 17520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8867666336023445e-05,
+      "loss": 0.3312,
+      "step": 17525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8866818547859095e-05,
+      "loss": 0.3105,
+      "step": 17530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886597044979984e-05,
+      "loss": 0.3202,
+      "step": 17535
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886512204185669e-05,
+      "loss": 0.3186,
+      "step": 17540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886427332404065e-05,
+      "loss": 0.3076,
+      "step": 17545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886342429636276e-05,
+      "loss": 0.3072,
+      "step": 17550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8862574958834016e-05,
+      "loss": 0.2953,
+      "step": 17555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8861725311465475e-05,
+      "loss": 0.313,
+      "step": 17560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.886087535426815e-05,
+      "loss": 0.3168,
+      "step": 17565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8860025087253086e-05,
+      "loss": 0.301,
+      "step": 17570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8859174510431314e-05,
+      "loss": 0.3209,
+      "step": 17575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8858323623813896e-05,
+      "loss": 0.3257,
+      "step": 17580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8857472427411864e-05,
+      "loss": 0.3161,
+      "step": 17585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8856620921236276e-05,
+      "loss": 0.2901,
+      "step": 17590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8855769105298185e-05,
+      "loss": 0.3008,
+      "step": 17595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8854916979608656e-05,
+      "loss": 0.3401,
+      "step": 17600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885406454417875e-05,
+      "loss": 0.3126,
+      "step": 17605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8853211799019534e-05,
+      "loss": 0.3105,
+      "step": 17610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8852358744142085e-05,
+      "loss": 0.3052,
+      "step": 17615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8851505379557475e-05,
+      "loss": 0.3248,
+      "step": 17620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.885065170527679e-05,
+      "loss": 0.3126,
+      "step": 17625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8849797721311106e-05,
+      "loss": 0.3261,
+      "step": 17630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8848943427671526e-05,
+      "loss": 0.3238,
+      "step": 17635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884808882436912e-05,
+      "loss": 0.3123,
+      "step": 17640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8847233911415e-05,
+      "loss": 0.3282,
+      "step": 17645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884637868882027e-05,
+      "loss": 0.3058,
+      "step": 17650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884552315659603e-05,
+      "loss": 0.3324,
+      "step": 17655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884466731475338e-05,
+      "loss": 0.3031,
+      "step": 17660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884381116330343e-05,
+      "loss": 0.3154,
+      "step": 17665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8842954702257324e-05,
+      "loss": 0.3194,
+      "step": 17670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884209793162615e-05,
+      "loss": 0.3127,
+      "step": 17675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.884124085142106e-05,
+      "loss": 0.3082,
+      "step": 17680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8840383461653164e-05,
+      "loss": 0.3087,
+      "step": 17685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883952576233359e-05,
+      "loss": 0.3167,
+      "step": 17690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8838667753473496e-05,
+      "loss": 0.3426,
+      "step": 17695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8837809435084016e-05,
+      "loss": 0.3001,
+      "step": 17700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8836950807176276e-05,
+      "loss": 0.2905,
+      "step": 17705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8836091869761444e-05,
+      "loss": 0.3193,
+      "step": 17710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883523262285068e-05,
+      "loss": 0.3067,
+      "step": 17715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8834373066455116e-05,
+      "loss": 0.3145,
+      "step": 17720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883351320058592e-05,
+      "loss": 0.3245,
+      "step": 17725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883265302525427e-05,
+      "loss": 0.3328,
+      "step": 17730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8831792540471334e-05,
+      "loss": 0.3123,
+      "step": 17735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.883093174624826e-05,
+      "loss": 0.3107,
+      "step": 17740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.8830070642596256e-05,
+      "loss": 0.3077,
+      "step": 17745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.882920922952648e-05,
+      "loss": 0.3102,
+      "step": 17750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.882834750705014e-05,
+      "loss": 0.3418,
+      "step": 17755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.88274854751784e-05,
+      "loss": 0.2913,
+      "step": 17760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882662313392247e-05,
+      "loss": 0.3132,
+      "step": 17765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8825760483293533e-05,
+      "loss": 0.2979,
+      "step": 17770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.88248975233028e-05,
+      "loss": 0.3141,
+      "step": 17775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882403425396148e-05,
+      "loss": 0.3054,
+      "step": 17780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882317067528077e-05,
+      "loss": 0.328,
+      "step": 17785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.88223067872719e-05,
+      "loss": 0.3238,
+      "step": 17790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.882144258994606e-05,
+      "loss": 0.3282,
+      "step": 17795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.88205780833145e-05,
+      "loss": 0.3001,
+      "step": 17800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8819713267388425e-05,
+      "loss": 0.307,
+      "step": 17805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881884814217907e-05,
+      "loss": 0.3345,
+      "step": 17810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881798270769767e-05,
+      "loss": 0.3091,
+      "step": 17815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8817116963955465e-05,
+      "loss": 0.2985,
+      "step": 17820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881625091096369e-05,
+      "loss": 0.2963,
+      "step": 17825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881538454873359e-05,
+      "loss": 0.34,
+      "step": 17830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881451787727642e-05,
+      "loss": 0.3245,
+      "step": 17835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881365089660342e-05,
+      "loss": 0.3092,
+      "step": 17840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881278360672587e-05,
+      "loss": 0.3129,
+      "step": 17845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.881191600765501e-05,
+      "loss": 0.3273,
+      "step": 17850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8811048099402124e-05,
+      "loss": 0.305,
+      "step": 17855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8810179881978455e-05,
+      "loss": 0.2934,
+      "step": 17860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.88093113553953e-05,
+      "loss": 0.3206,
+      "step": 17865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8808442519663936e-05,
+      "loss": 0.3365,
+      "step": 17870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880757337479562e-05,
+      "loss": 0.3166,
+      "step": 17875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8806703920801666e-05,
+      "loss": 0.3241,
+      "step": 17880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880583415769334e-05,
+      "loss": 0.299,
+      "step": 17885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880496408548196e-05,
+      "loss": 0.3023,
+      "step": 17890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.88040937041788e-05,
+      "loss": 0.3261,
+      "step": 17895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8803223013795175e-05,
+      "loss": 0.3378,
+      "step": 17900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8802352014342376e-05,
+      "loss": 0.2997,
+      "step": 17905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880148070583174e-05,
+      "loss": 0.3037,
+      "step": 17910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.880060908827455e-05,
+      "loss": 0.3039,
+      "step": 17915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879973716168214e-05,
+      "loss": 0.2984,
+      "step": 17920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8798864926065823e-05,
+      "loss": 0.3049,
+      "step": 17925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879799238143693e-05,
+      "loss": 0.2784,
+      "step": 17930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879711952780679e-05,
+      "loss": 0.2891,
+      "step": 17935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879624636518674e-05,
+      "loss": 0.297,
+      "step": 17940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879537289358811e-05,
+      "loss": 0.2995,
+      "step": 17945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879449911302225e-05,
+      "loss": 0.3221,
+      "step": 17950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879362502350049e-05,
+      "loss": 0.3157,
+      "step": 17955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.87927506250342e-05,
+      "loss": 0.3068,
+      "step": 17960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879187591763472e-05,
+      "loss": 0.3526,
+      "step": 17965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879100090131341e-05,
+      "loss": 0.2967,
+      "step": 17970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.879012557608162e-05,
+      "loss": 0.2943,
+      "step": 17975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8789249941950746e-05,
+      "loss": 0.32,
+      "step": 17980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878837399893213e-05,
+      "loss": 0.334,
+      "step": 17985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8787497747037156e-05,
+      "loss": 0.3192,
+      "step": 17990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8786621186277204e-05,
+      "loss": 0.2938,
+      "step": 17995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878574431666365e-05,
+      "loss": 0.314,
+      "step": 18000
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 0.3048493266105652,
+      "eval_runtime": 10.5346,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 18000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8784867138207877e-05,
+      "loss": 0.2992,
+      "step": 18005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878398965092129e-05,
+      "loss": 0.3117,
+      "step": 18010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878311185481526e-05,
+      "loss": 0.3081,
+      "step": 18015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8782233749901205e-05,
+      "loss": 0.3108,
+      "step": 18020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.878135533619051e-05,
+      "loss": 0.3003,
+      "step": 18025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.8780476613694594e-05,
+      "loss": 0.3065,
+      "step": 18030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.877959758242486e-05,
+      "loss": 0.307,
+      "step": 18035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.877871824239272e-05,
+      "loss": 0.3209,
+      "step": 18040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.877783859360959e-05,
+      "loss": 0.2935,
+      "step": 18045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.87769586360869e-05,
+      "loss": 0.3417,
+      "step": 18050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.877607836983608e-05,
+      "loss": 0.3528,
+      "step": 18055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8775197794868544e-05,
+      "loss": 0.3162,
+      "step": 18060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8774316911195736e-05,
+      "loss": 0.302,
+      "step": 18065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.877343571882909e-05,
+      "loss": 0.3097,
+      "step": 18070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.877255421778004e-05,
+      "loss": 0.3214,
+      "step": 18075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8771672408060056e-05,
+      "loss": 0.3215,
+      "step": 18080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8770790289680555e-05,
+      "loss": 0.321,
+      "step": 18085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876990786265302e-05,
+      "loss": 0.3224,
+      "step": 18090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8769025126988896e-05,
+      "loss": 0.3153,
+      "step": 18095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876814208269964e-05,
+      "loss": 0.33,
+      "step": 18100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8767258729796725e-05,
+      "loss": 0.3092,
+      "step": 18105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876637506829162e-05,
+      "loss": 0.3139,
+      "step": 18110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876549109819579e-05,
+      "loss": 0.3209,
+      "step": 18115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876460681952073e-05,
+      "loss": 0.2928,
+      "step": 18120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876372223227791e-05,
+      "loss": 0.2933,
+      "step": 18125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876283733647881e-05,
+      "loss": 0.3381,
+      "step": 18130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876195213213494e-05,
+      "loss": 0.331,
+      "step": 18135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.876106661925777e-05,
+      "loss": 0.3399,
+      "step": 18140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8760180797858815e-05,
+      "loss": 0.3081,
+      "step": 18145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8759294667949565e-05,
+      "loss": 0.3229,
+      "step": 18150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875840822954154e-05,
+      "loss": 0.3046,
+      "step": 18155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875752148264624e-05,
+      "loss": 0.3098,
+      "step": 18160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875663442727517e-05,
+      "loss": 0.3071,
+      "step": 18165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875574706343987e-05,
+      "loss": 0.305,
+      "step": 18170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8754859391151844e-05,
+      "loss": 0.2952,
+      "step": 18175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875397141042263e-05,
+      "loss": 0.3284,
+      "step": 18180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875308312126374e-05,
+      "loss": 0.3252,
+      "step": 18185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8752194523686726e-05,
+      "loss": 0.2949,
+      "step": 18190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875130561770312e-05,
+      "loss": 0.3216,
+      "step": 18195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.875041640332446e-05,
+      "loss": 0.2878,
+      "step": 18200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.87495268805623e-05,
+      "loss": 0.2957,
+      "step": 18205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8748637049428185e-05,
+      "loss": 0.3343,
+      "step": 18210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874774690993367e-05,
+      "loss": 0.2956,
+      "step": 18215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8746856462090305e-05,
+      "loss": 0.3084,
+      "step": 18220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874596570590967e-05,
+      "loss": 0.3132,
+      "step": 18225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874507464140331e-05,
+      "loss": 0.3079,
+      "step": 18230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8744183268582804e-05,
+      "loss": 0.3015,
+      "step": 18235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874329158745973e-05,
+      "loss": 0.3592,
+      "step": 18240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874239959804566e-05,
+      "loss": 0.3334,
+      "step": 18245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874150730035218e-05,
+      "loss": 0.3284,
+      "step": 18250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.874061469439087e-05,
+      "loss": 0.3008,
+      "step": 18255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873972178017333e-05,
+      "loss": 0.317,
+      "step": 18260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873882855771115e-05,
+      "loss": 0.3178,
+      "step": 18265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8737935027015915e-05,
+      "loss": 0.3101,
+      "step": 18270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873704118809924e-05,
+      "loss": 0.3289,
+      "step": 18275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8736147040972736e-05,
+      "loss": 0.2987,
+      "step": 18280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8735252585647995e-05,
+      "loss": 0.3261,
+      "step": 18285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873435782213665e-05,
+      "loss": 0.3093,
+      "step": 18290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8733462750450306e-05,
+      "loss": 0.284,
+      "step": 18295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873256737060059e-05,
+      "loss": 0.3229,
+      "step": 18300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8731671682599126e-05,
+      "loss": 0.3229,
+      "step": 18305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.873077568645755e-05,
+      "loss": 0.3249,
+      "step": 18310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.8729879382187485e-05,
+      "loss": 0.3393,
+      "step": 18315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.872898276980058e-05,
+      "loss": 0.33,
+      "step": 18320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.872808584930847e-05,
+      "loss": 0.3207,
+      "step": 18325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.87271886207228e-05,
+      "loss": 0.2978,
+      "step": 18330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.872629108405522e-05,
+      "loss": 0.3273,
+      "step": 18335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.872539323931739e-05,
+      "loss": 0.2903,
+      "step": 18340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8724495086520964e-05,
+      "loss": 0.3063,
+      "step": 18345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.872359662567761e-05,
+      "loss": 0.3177,
+      "step": 18350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8722697856798985e-05,
+      "loss": 0.293,
+      "step": 18355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8721798779896756e-05,
+      "loss": 0.3201,
+      "step": 18360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.872089939498261e-05,
+      "loss": 0.3389,
+      "step": 18365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871999970206822e-05,
+      "loss": 0.297,
+      "step": 18370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871909970116526e-05,
+      "loss": 0.3382,
+      "step": 18375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871819939228542e-05,
+      "loss": 0.3001,
+      "step": 18380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871729877544039e-05,
+      "loss": 0.3313,
+      "step": 18385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871639785064187e-05,
+      "loss": 0.3017,
+      "step": 18390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871549661790156e-05,
+      "loss": 0.3424,
+      "step": 18395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871459507723115e-05,
+      "loss": 0.3293,
+      "step": 18400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8713693228642356e-05,
+      "loss": 0.3195,
+      "step": 18405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871279107214687e-05,
+      "loss": 0.3149,
+      "step": 18410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871188860775643e-05,
+      "loss": 0.3221,
+      "step": 18415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871098583548274e-05,
+      "loss": 0.2928,
+      "step": 18420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.871008275533753e-05,
+      "loss": 0.2977,
+      "step": 18425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8709179367332514e-05,
+      "loss": 0.3271,
+      "step": 18430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8708275671479434e-05,
+      "loss": 0.3082,
+      "step": 18435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8707371667790016e-05,
+      "loss": 0.3163,
+      "step": 18440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8706467356276e-05,
+      "loss": 0.2826,
+      "step": 18445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870556273694913e-05,
+      "loss": 0.3069,
+      "step": 18450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8704657809821154e-05,
+      "loss": 0.3196,
+      "step": 18455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8703752574903814e-05,
+      "loss": 0.3224,
+      "step": 18460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870284703220887e-05,
+      "loss": 0.331,
+      "step": 18465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.870194118174807e-05,
+      "loss": 0.2804,
+      "step": 18470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8701035023533193e-05,
+      "loss": 0.3272,
+      "step": 18475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8700128557576e-05,
+      "loss": 0.2817,
+      "step": 18480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869922178388824e-05,
+      "loss": 0.301,
+      "step": 18485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869831470248172e-05,
+      "loss": 0.3247,
+      "step": 18490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869740731336819e-05,
+      "loss": 0.3042,
+      "step": 18495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869649961655946e-05,
+      "loss": 0.302,
+      "step": 18500
+    },
+    {
+      "epoch": 0.65,
+      "eval_loss": 0.3054150938987732,
+      "eval_runtime": 10.5494,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 18500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869559161206728e-05,
+      "loss": 0.3064,
+      "step": 18505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869468329990347e-05,
+      "loss": 0.3255,
+      "step": 18510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8693774680079806e-05,
+      "loss": 0.3148,
+      "step": 18515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.86928657526081e-05,
+      "loss": 0.302,
+      "step": 18520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869195651750014e-05,
+      "loss": 0.3206,
+      "step": 18525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869104697476774e-05,
+      "loss": 0.3158,
+      "step": 18530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.869013712442271e-05,
+      "loss": 0.2796,
+      "step": 18535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868922696647685e-05,
+      "loss": 0.321,
+      "step": 18540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8688316500942e-05,
+      "loss": 0.3074,
+      "step": 18545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868740572782997e-05,
+      "loss": 0.2912,
+      "step": 18550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868649464715259e-05,
+      "loss": 0.3335,
+      "step": 18555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868558325892168e-05,
+      "loss": 0.3184,
+      "step": 18560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8684671563149095e-05,
+      "loss": 0.3027,
+      "step": 18565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868375955984665e-05,
+      "loss": 0.3115,
+      "step": 18570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868284724902619e-05,
+      "loss": 0.3293,
+      "step": 18575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868193463069958e-05,
+      "loss": 0.3297,
+      "step": 18580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.8681021704878634e-05,
+      "loss": 0.3219,
+      "step": 18585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.868010847157525e-05,
+      "loss": 0.3026,
+      "step": 18590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.867919493080125e-05,
+      "loss": 0.3164,
+      "step": 18595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.867828108256851e-05,
+      "loss": 0.3117,
+      "step": 18600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.86773669268889e-05,
+      "loss": 0.2914,
+      "step": 18605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.867645246377428e-05,
+      "loss": 0.3312,
+      "step": 18610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.867553769323653e-05,
+      "loss": 0.3089,
+      "step": 18615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.867462261528752e-05,
+      "loss": 0.3158,
+      "step": 18620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.867370722993915e-05,
+      "loss": 0.3053,
+      "step": 18625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8672791537203286e-05,
+      "loss": 0.3143,
+      "step": 18630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8671875537091824e-05,
+      "loss": 0.333,
+      "step": 18635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.867095922961666e-05,
+      "loss": 0.2965,
+      "step": 18640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.867004261478969e-05,
+      "loss": 0.2995,
+      "step": 18645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866912569262282e-05,
+      "loss": 0.3155,
+      "step": 18650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866820846312794e-05,
+      "loss": 0.3045,
+      "step": 18655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866729092631698e-05,
+      "loss": 0.2974,
+      "step": 18660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866637308220184e-05,
+      "loss": 0.3212,
+      "step": 18665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8665454930794454e-05,
+      "loss": 0.3561,
+      "step": 18670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8664536472106715e-05,
+      "loss": 0.2988,
+      "step": 18675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866361770615058e-05,
+      "loss": 0.2998,
+      "step": 18680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8662698632937955e-05,
+      "loss": 0.3376,
+      "step": 18685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.866177925248078e-05,
+      "loss": 0.3156,
+      "step": 18690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8660859564791006e-05,
+      "loss": 0.3197,
+      "step": 18695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865993956988055e-05,
+      "loss": 0.3498,
+      "step": 18700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865901926776139e-05,
+      "loss": 0.3049,
+      "step": 18705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865809865844544e-05,
+      "loss": 0.3001,
+      "step": 18710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865717774194467e-05,
+      "loss": 0.3121,
+      "step": 18715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865625651827105e-05,
+      "loss": 0.308,
+      "step": 18720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865533498743652e-05,
+      "loss": 0.3274,
+      "step": 18725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8654413149453056e-05,
+      "loss": 0.3408,
+      "step": 18730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865349100433263e-05,
+      "loss": 0.3402,
+      "step": 18735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8652568552087206e-05,
+      "loss": 0.3118,
+      "step": 18740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.865164579272877e-05,
+      "loss": 0.3159,
+      "step": 18745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.86507227262693e-05,
+      "loss": 0.3268,
+      "step": 18750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864979935272078e-05,
+      "loss": 0.313,
+      "step": 18755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8648875672095205e-05,
+      "loss": 0.274,
+      "step": 18760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864795168440456e-05,
+      "loss": 0.2841,
+      "step": 18765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864702738966085e-05,
+      "loss": 0.3536,
+      "step": 18770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864610278787607e-05,
+      "loss": 0.3068,
+      "step": 18775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8645177879062235e-05,
+      "loss": 0.3262,
+      "step": 18780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8644252663231344e-05,
+      "loss": 0.2929,
+      "step": 18785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864332714039541e-05,
+      "loss": 0.3021,
+      "step": 18790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864240131056646e-05,
+      "loss": 0.3201,
+      "step": 18795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8641475173756514e-05,
+      "loss": 0.307,
+      "step": 18800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.864054872997759e-05,
+      "loss": 0.3242,
+      "step": 18805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863962197924172e-05,
+      "loss": 0.3069,
+      "step": 18810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863869492156093e-05,
+      "loss": 0.3135,
+      "step": 18815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863776755694728e-05,
+      "loss": 0.324,
+      "step": 18820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863683988541279e-05,
+      "loss": 0.3285,
+      "step": 18825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.86359119069695e-05,
+      "loss": 0.2937,
+      "step": 18830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863498362162948e-05,
+      "loss": 0.3272,
+      "step": 18835
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863405502940478e-05,
+      "loss": 0.2852,
+      "step": 18840
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863312613030744e-05,
+      "loss": 0.2904,
+      "step": 18845
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863219692434954e-05,
+      "loss": 0.3309,
+      "step": 18850
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863126741154313e-05,
+      "loss": 0.3213,
+      "step": 18855
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.863033759190029e-05,
+      "loss": 0.3233,
+      "step": 18860
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862940746543308e-05,
+      "loss": 0.3006,
+      "step": 18865
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862847703215361e-05,
+      "loss": 0.3136,
+      "step": 18870
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8627546292073915e-05,
+      "loss": 0.3159,
+      "step": 18875
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862661524520611e-05,
+      "loss": 0.2867,
+      "step": 18880
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862568389156228e-05,
+      "loss": 0.3126,
+      "step": 18885
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.8624752231154517e-05,
+      "loss": 0.322,
+      "step": 18890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.86238202639949e-05,
+      "loss": 0.311,
+      "step": 18895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.862288799009556e-05,
+      "loss": 0.3173,
+      "step": 18900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.862195540946858e-05,
+      "loss": 0.2893,
+      "step": 18905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.862102252212608e-05,
+      "loss": 0.305,
+      "step": 18910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8620089328080176e-05,
+      "loss": 0.299,
+      "step": 18915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861915582734297e-05,
+      "loss": 0.2985,
+      "step": 18920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861822201992659e-05,
+      "loss": 0.3384,
+      "step": 18925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861728790584317e-05,
+      "loss": 0.3381,
+      "step": 18930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861635348510482e-05,
+      "loss": 0.3576,
+      "step": 18935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861541875772369e-05,
+      "loss": 0.2984,
+      "step": 18940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861448372371191e-05,
+      "loss": 0.2902,
+      "step": 18945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861354838308162e-05,
+      "loss": 0.2975,
+      "step": 18950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861261273584497e-05,
+      "loss": 0.3032,
+      "step": 18955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8611676782014104e-05,
+      "loss": 0.3307,
+      "step": 18960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.861074052160117e-05,
+      "loss": 0.3202,
+      "step": 18965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860980395461834e-05,
+      "loss": 0.3167,
+      "step": 18970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860886708107776e-05,
+      "loss": 0.3472,
+      "step": 18975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8607929900991594e-05,
+      "loss": 0.3341,
+      "step": 18980
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860699241437202e-05,
+      "loss": 0.3142,
+      "step": 18985
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860605462123121e-05,
+      "loss": 0.327,
+      "step": 18990
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860511652158134e-05,
+      "loss": 0.3138,
+      "step": 18995
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860417811543458e-05,
+      "loss": 0.3034,
+      "step": 19000
+    },
+    {
+      "epoch": 0.67,
+      "eval_loss": 0.30461186170578003,
+      "eval_runtime": 10.5328,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 19000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860323940280313e-05,
+      "loss": 0.3134,
+      "step": 19005
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860230038369917e-05,
+      "loss": 0.3235,
+      "step": 19010
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.860136105813489e-05,
+      "loss": 0.3028,
+      "step": 19015
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.86004214261225e-05,
+      "loss": 0.3016,
+      "step": 19020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859948148767418e-05,
+      "loss": 0.2898,
+      "step": 19025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8598541242802155e-05,
+      "loss": 0.2882,
+      "step": 19030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859760069151862e-05,
+      "loss": 0.2995,
+      "step": 19035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859665983383579e-05,
+      "loss": 0.309,
+      "step": 19040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8595718669765874e-05,
+      "loss": 0.3045,
+      "step": 19045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859477719932112e-05,
+      "loss": 0.2974,
+      "step": 19050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859383542251372e-05,
+      "loss": 0.322,
+      "step": 19055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8592893339355925e-05,
+      "loss": 0.3175,
+      "step": 19060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859195094985996e-05,
+      "loss": 0.3287,
+      "step": 19065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859100825403805e-05,
+      "loss": 0.311,
+      "step": 19070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.859006525190244e-05,
+      "loss": 0.3208,
+      "step": 19075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858912194346539e-05,
+      "loss": 0.3133,
+      "step": 19080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858817832873913e-05,
+      "loss": 0.315,
+      "step": 19085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8587234407735926e-05,
+      "loss": 0.2925,
+      "step": 19090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858629018046803e-05,
+      "loss": 0.3254,
+      "step": 19095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858534564694769e-05,
+      "loss": 0.2914,
+      "step": 19100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858440080718719e-05,
+      "loss": 0.3161,
+      "step": 19105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858345566119879e-05,
+      "loss": 0.3113,
+      "step": 19110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8582510208994746e-05,
+      "loss": 0.3259,
+      "step": 19115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.858156445058736e-05,
+      "loss": 0.3197,
+      "step": 19120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.85806183859889e-05,
+      "loss": 0.3311,
+      "step": 19125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857967201521165e-05,
+      "loss": 0.3248,
+      "step": 19130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857872533826789e-05,
+      "loss": 0.3058,
+      "step": 19135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8577778355169935e-05,
+      "loss": 0.3079,
+      "step": 19140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857683106593005e-05,
+      "loss": 0.3095,
+      "step": 19145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8575883470560565e-05,
+      "loss": 0.2988,
+      "step": 19150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857493556907376e-05,
+      "loss": 0.3,
+      "step": 19155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857398736148196e-05,
+      "loss": 0.3329,
+      "step": 19160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857303884779747e-05,
+      "loss": 0.3193,
+      "step": 19165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8572090028032604e-05,
+      "loss": 0.3163,
+      "step": 19170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8571140902199684e-05,
+      "loss": 0.3055,
+      "step": 19175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.857019147031103e-05,
+      "loss": 0.3248,
+      "step": 19180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.8569241732378975e-05,
+      "loss": 0.3135,
+      "step": 19185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856829168841585e-05,
+      "loss": 0.3074,
+      "step": 19190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856734133843399e-05,
+      "loss": 0.3359,
+      "step": 19195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8566390682445734e-05,
+      "loss": 0.3111,
+      "step": 19200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8565439720463424e-05,
+      "loss": 0.3306,
+      "step": 19205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856448845249941e-05,
+      "loss": 0.3012,
+      "step": 19210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856353687856604e-05,
+      "loss": 0.3008,
+      "step": 19215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856258499867568e-05,
+      "loss": 0.3193,
+      "step": 19220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8561632812840674e-05,
+      "loss": 0.3251,
+      "step": 19225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8560680321073405e-05,
+      "loss": 0.2841,
+      "step": 19230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8559727523386214e-05,
+      "loss": 0.3252,
+      "step": 19235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8558774419791496e-05,
+      "loss": 0.3157,
+      "step": 19240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855782101030162e-05,
+      "loss": 0.3063,
+      "step": 19245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855686729492896e-05,
+      "loss": 0.284,
+      "step": 19250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85559132736859e-05,
+      "loss": 0.3275,
+      "step": 19255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855495894658484e-05,
+      "loss": 0.295,
+      "step": 19260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8554004313638154e-05,
+      "loss": 0.3287,
+      "step": 19265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855304937485825e-05,
+      "loss": 0.3224,
+      "step": 19270
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8552094130257515e-05,
+      "loss": 0.3019,
+      "step": 19275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855113857984836e-05,
+      "loss": 0.334,
+      "step": 19280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85501827236432e-05,
+      "loss": 0.295,
+      "step": 19285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8549226561654426e-05,
+      "loss": 0.2808,
+      "step": 19290
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8548270093894465e-05,
+      "loss": 0.3072,
+      "step": 19295
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854731332037574e-05,
+      "loss": 0.3187,
+      "step": 19300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8546356241110666e-05,
+      "loss": 0.3158,
+      "step": 19305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8545398856111676e-05,
+      "loss": 0.3075,
+      "step": 19310
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854444116539121e-05,
+      "loss": 0.3375,
+      "step": 19315
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854348316896168e-05,
+      "loss": 0.3051,
+      "step": 19320
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854252486683554e-05,
+      "loss": 0.293,
+      "step": 19325
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8541566259025226e-05,
+      "loss": 0.3089,
+      "step": 19330
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8540607345543194e-05,
+      "loss": 0.312,
+      "step": 19335
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853964812640188e-05,
+      "loss": 0.3218,
+      "step": 19340
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853868860161376e-05,
+      "loss": 0.2937,
+      "step": 19345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8537728771191275e-05,
+      "loss": 0.3022,
+      "step": 19350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8536768635146895e-05,
+      "loss": 0.3084,
+      "step": 19355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853580819349309e-05,
+      "loss": 0.3086,
+      "step": 19360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853484744624232e-05,
+      "loss": 0.3378,
+      "step": 19365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853388639340707e-05,
+      "loss": 0.2914,
+      "step": 19370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853292503499981e-05,
+      "loss": 0.293,
+      "step": 19375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8531963371033026e-05,
+      "loss": 0.2818,
+      "step": 19380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8531001401519214e-05,
+      "loss": 0.3251,
+      "step": 19385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853003912647085e-05,
+      "loss": 0.2689,
+      "step": 19390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852907654590044e-05,
+      "loss": 0.3265,
+      "step": 19395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8528113659820474e-05,
+      "loss": 0.306,
+      "step": 19400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8527150468243454e-05,
+      "loss": 0.2849,
+      "step": 19405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8526186971181893e-05,
+      "loss": 0.3083,
+      "step": 19410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8525223168648296e-05,
+      "loss": 0.3484,
+      "step": 19415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852425906065519e-05,
+      "loss": 0.3175,
+      "step": 19420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852329464721507e-05,
+      "loss": 0.3194,
+      "step": 19425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852232992834048e-05,
+      "loss": 0.2969,
+      "step": 19430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852136490404393e-05,
+      "loss": 0.3438,
+      "step": 19435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852039957433797e-05,
+      "loss": 0.3239,
+      "step": 19440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851943393923512e-05,
+      "loss": 0.3088,
+      "step": 19445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851846799874791e-05,
+      "loss": 0.3085,
+      "step": 19450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8517501752888895e-05,
+      "loss": 0.274,
+      "step": 19455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851653520167062e-05,
+      "loss": 0.3079,
+      "step": 19460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.851556834510564e-05,
+      "loss": 0.3116,
+      "step": 19465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.851460118320649e-05,
+      "loss": 0.2865,
+      "step": 19470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.851363371598575e-05,
+      "loss": 0.3095,
+      "step": 19475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.851266594345597e-05,
+      "loss": 0.2859,
+      "step": 19480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8511697865629715e-05,
+      "loss": 0.2908,
+      "step": 19485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.851072948251957e-05,
+      "loss": 0.3073,
+      "step": 19490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850976079413808e-05,
+      "loss": 0.3105,
+      "step": 19495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850879180049785e-05,
+      "loss": 0.3014,
+      "step": 19500
+    },
+    {
+      "epoch": 0.69,
+      "eval_loss": 0.30417418479919434,
+      "eval_runtime": 10.5331,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 19500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8507822501611444e-05,
+      "loss": 0.3377,
+      "step": 19505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850685289749146e-05,
+      "loss": 0.2887,
+      "step": 19510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850588298815048e-05,
+      "loss": 0.3038,
+      "step": 19515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850491277360111e-05,
+      "loss": 0.2823,
+      "step": 19520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8503942253855927e-05,
+      "loss": 0.3096,
+      "step": 19525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8502971428927544e-05,
+      "loss": 0.295,
+      "step": 19530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850200029882857e-05,
+      "loss": 0.2925,
+      "step": 19535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850102886357161e-05,
+      "loss": 0.3171,
+      "step": 19540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.850005712316928e-05,
+      "loss": 0.3163,
+      "step": 19545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849908507763419e-05,
+      "loss": 0.3025,
+      "step": 19550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849811272697897e-05,
+      "loss": 0.3011,
+      "step": 19555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849714007121624e-05,
+      "loss": 0.3052,
+      "step": 19560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8496167110358637e-05,
+      "loss": 0.304,
+      "step": 19565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8495193844418784e-05,
+      "loss": 0.3287,
+      "step": 19570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849422027340932e-05,
+      "loss": 0.3088,
+      "step": 19575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8493246397342896e-05,
+      "loss": 0.3191,
+      "step": 19580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849227221623215e-05,
+      "loss": 0.2954,
+      "step": 19585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849129773008972e-05,
+      "loss": 0.2997,
+      "step": 19590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849032293892828e-05,
+      "loss": 0.3194,
+      "step": 19595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848934784276048e-05,
+      "loss": 0.2893,
+      "step": 19600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848837244159897e-05,
+      "loss": 0.3139,
+      "step": 19605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848739673545643e-05,
+      "loss": 0.3236,
+      "step": 19610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8486420724345516e-05,
+      "loss": 0.3074,
+      "step": 19615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8485444408278915e-05,
+      "loss": 0.3413,
+      "step": 19620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8484467787269294e-05,
+      "loss": 0.3285,
+      "step": 19625
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848349086132933e-05,
+      "loss": 0.3101,
+      "step": 19630
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848251363047171e-05,
+      "loss": 0.3094,
+      "step": 19635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848153609470914e-05,
+      "loss": 0.3325,
+      "step": 19640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.848055825405429e-05,
+      "loss": 0.2964,
+      "step": 19645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847958010851986e-05,
+      "loss": 0.3134,
+      "step": 19650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847860165811857e-05,
+      "loss": 0.2776,
+      "step": 19655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84776229028631e-05,
+      "loss": 0.3465,
+      "step": 19660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847664384276617e-05,
+      "loss": 0.3151,
+      "step": 19665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84756644778405e-05,
+      "loss": 0.3081,
+      "step": 19670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847468480809879e-05,
+      "loss": 0.3069,
+      "step": 19675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847370483355377e-05,
+      "loss": 0.3152,
+      "step": 19680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847272455421816e-05,
+      "loss": 0.2902,
+      "step": 19685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.847174397010469e-05,
+      "loss": 0.3075,
+      "step": 19690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84707630812261e-05,
+      "loss": 0.2934,
+      "step": 19695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8469781887595115e-05,
+      "loss": 0.3203,
+      "step": 19700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846880038922447e-05,
+      "loss": 0.3175,
+      "step": 19705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846781858612693e-05,
+      "loss": 0.2932,
+      "step": 19710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846683647831522e-05,
+      "loss": 0.311,
+      "step": 19715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846585406580212e-05,
+      "loss": 0.3067,
+      "step": 19720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846487134860036e-05,
+      "loss": 0.3117,
+      "step": 19725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.8463888326722714e-05,
+      "loss": 0.3373,
+      "step": 19730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846290500018194e-05,
+      "loss": 0.3003,
+      "step": 19735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.84619213689908e-05,
+      "loss": 0.3069,
+      "step": 19740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.846093743316208e-05,
+      "loss": 0.3144,
+      "step": 19745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.845995319270855e-05,
+      "loss": 0.3183,
+      "step": 19750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845896864764299e-05,
+      "loss": 0.3005,
+      "step": 19755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845798379797818e-05,
+      "loss": 0.3114,
+      "step": 19760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84569986437269e-05,
+      "loss": 0.3258,
+      "step": 19765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845601318490196e-05,
+      "loss": 0.302,
+      "step": 19770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845502742151615e-05,
+      "loss": 0.3038,
+      "step": 19775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845404135358226e-05,
+      "loss": 0.3218,
+      "step": 19780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8453054981113106e-05,
+      "loss": 0.3271,
+      "step": 19785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845206830412149e-05,
+      "loss": 0.3154,
+      "step": 19790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8451081322620215e-05,
+      "loss": 0.3069,
+      "step": 19795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.845009403662211e-05,
+      "loss": 0.3261,
+      "step": 19800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844910644613998e-05,
+      "loss": 0.2997,
+      "step": 19805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8448118551186664e-05,
+      "loss": 0.2808,
+      "step": 19810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8447130351774986e-05,
+      "loss": 0.2954,
+      "step": 19815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844614184791776e-05,
+      "loss": 0.3093,
+      "step": 19820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844515303962784e-05,
+      "loss": 0.311,
+      "step": 19825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844416392691806e-05,
+      "loss": 0.2939,
+      "step": 19830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844317450980126e-05,
+      "loss": 0.3139,
+      "step": 19835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844218478829028e-05,
+      "loss": 0.2842,
+      "step": 19840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844119476239799e-05,
+      "loss": 0.2998,
+      "step": 19845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.844020443213724e-05,
+      "loss": 0.3093,
+      "step": 19850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8439213797520874e-05,
+      "loss": 0.2712,
+      "step": 19855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8438222858561764e-05,
+      "loss": 0.3019,
+      "step": 19860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8437231615272775e-05,
+      "loss": 0.3105,
+      "step": 19865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8436240067666784e-05,
+      "loss": 0.3299,
+      "step": 19870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843524821575666e-05,
+      "loss": 0.308,
+      "step": 19875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8434256059555284e-05,
+      "loss": 0.2992,
+      "step": 19880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843326359907553e-05,
+      "loss": 0.3057,
+      "step": 19885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.84322708343303e-05,
+      "loss": 0.2805,
+      "step": 19890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843127776533247e-05,
+      "loss": 0.2985,
+      "step": 19895
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.843028439209494e-05,
+      "loss": 0.3125,
+      "step": 19900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8429290714630605e-05,
+      "loss": 0.3265,
+      "step": 19905
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8428296732952376e-05,
+      "loss": 0.302,
+      "step": 19910
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842730244707315e-05,
+      "loss": 0.3318,
+      "step": 19915
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8426307857005845e-05,
+      "loss": 0.2854,
+      "step": 19920
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842531296276337e-05,
+      "loss": 0.3413,
+      "step": 19925
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8424317764358645e-05,
+      "loss": 0.3107,
+      "step": 19930
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842332226180459e-05,
+      "loss": 0.2836,
+      "step": 19935
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842232645511413e-05,
+      "loss": 0.3137,
+      "step": 19940
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8421330344300196e-05,
+      "loss": 0.3102,
+      "step": 19945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.842033392937573e-05,
+      "loss": 0.3285,
+      "step": 19950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.841933721035365e-05,
+      "loss": 0.3212,
+      "step": 19955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8418340187246926e-05,
+      "loss": 0.2831,
+      "step": 19960
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8417342860068476e-05,
+      "loss": 0.3049,
+      "step": 19965
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8416345228831266e-05,
+      "loss": 0.3054,
+      "step": 19970
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8415347293548245e-05,
+      "loss": 0.2954,
+      "step": 19975
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.841434905423237e-05,
+      "loss": 0.2984,
+      "step": 19980
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8413350510896605e-05,
+      "loss": 0.3073,
+      "step": 19985
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.841235166355391e-05,
+      "loss": 0.3117,
+      "step": 19990
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.841135251221726e-05,
+      "loss": 0.317,
+      "step": 19995
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8410353056899626e-05,
+      "loss": 0.3122,
+      "step": 20000
+    },
+    {
+      "epoch": 0.7,
+      "eval_loss": 0.3004976212978363,
+      "eval_runtime": 10.5473,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 20000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8409353297613994e-05,
+      "loss": 0.3367,
+      "step": 20005
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840835323437332e-05,
+      "loss": 0.2945,
+      "step": 20010
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840735286719061e-05,
+      "loss": 0.2952,
+      "step": 20015
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8406352196078864e-05,
+      "loss": 0.2975,
+      "step": 20020
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8405351221051044e-05,
+      "loss": 0.3112,
+      "step": 20025
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.840434994212017e-05,
+      "loss": 0.2915,
+      "step": 20030
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.8403348359299236e-05,
+      "loss": 0.3102,
+      "step": 20035
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.840234647260125e-05,
+      "loss": 0.3047,
+      "step": 20040
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8401344282039204e-05,
+      "loss": 0.3044,
+      "step": 20045
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.840034178762613e-05,
+      "loss": 0.282,
+      "step": 20050
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839933898937505e-05,
+      "loss": 0.3284,
+      "step": 20055
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8398335887298965e-05,
+      "loss": 0.2998,
+      "step": 20060
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839733248141092e-05,
+      "loss": 0.3012,
+      "step": 20065
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839632877172392e-05,
+      "loss": 0.291,
+      "step": 20070
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839532475825101e-05,
+      "loss": 0.3247,
+      "step": 20075
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839432044100523e-05,
+      "loss": 0.3074,
+      "step": 20080
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839331581999962e-05,
+      "loss": 0.2962,
+      "step": 20085
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.839231089524721e-05,
+      "loss": 0.3216,
+      "step": 20090
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8391305666761065e-05,
+      "loss": 0.3007,
+      "step": 20095
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8390300134554236e-05,
+      "loss": 0.2944,
+      "step": 20100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8389294298639766e-05,
+      "loss": 0.2958,
+      "step": 20105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8388288159030727e-05,
+      "loss": 0.3113,
+      "step": 20110
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838728171574018e-05,
+      "loss": 0.2786,
+      "step": 20115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.83862749687812e-05,
+      "loss": 0.3319,
+      "step": 20120
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8385267918166844e-05,
+      "loss": 0.2875,
+      "step": 20125
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838426056391019e-05,
+      "loss": 0.32,
+      "step": 20130
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838325290602434e-05,
+      "loss": 0.3277,
+      "step": 20135
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8382244944522346e-05,
+      "loss": 0.3221,
+      "step": 20140
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.838123667941731e-05,
+      "loss": 0.3474,
+      "step": 20145
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8380228110722336e-05,
+      "loss": 0.2811,
+      "step": 20150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.83792192384505e-05,
+      "loss": 0.284,
+      "step": 20155
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8378210062614916e-05,
+      "loss": 0.295,
+      "step": 20160
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837720058322868e-05,
+      "loss": 0.3091,
+      "step": 20165
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8376190800304896e-05,
+      "loss": 0.3085,
+      "step": 20170
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8375180713856684e-05,
+      "loss": 0.3105,
+      "step": 20175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837417032389715e-05,
+      "loss": 0.3035,
+      "step": 20180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837315963043943e-05,
+      "loss": 0.2796,
+      "step": 20185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8372148633496614e-05,
+      "loss": 0.299,
+      "step": 20190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8371137333081866e-05,
+      "loss": 0.3032,
+      "step": 20195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.837012572920831e-05,
+      "loss": 0.33,
+      "step": 20200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836911382188906e-05,
+      "loss": 0.2876,
+      "step": 20205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8368101611137265e-05,
+      "loss": 0.3226,
+      "step": 20210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836708909696608e-05,
+      "loss": 0.3189,
+      "step": 20215
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836607627938864e-05,
+      "loss": 0.3081,
+      "step": 20220
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836506315841809e-05,
+      "loss": 0.307,
+      "step": 20225
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8364049734067606e-05,
+      "loss": 0.3143,
+      "step": 20230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8363036006350324e-05,
+      "loss": 0.3044,
+      "step": 20235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836202197527943e-05,
+      "loss": 0.3096,
+      "step": 20240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.836100764086806e-05,
+      "loss": 0.2831,
+      "step": 20245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835999300312941e-05,
+      "loss": 0.288,
+      "step": 20250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835897806207664e-05,
+      "loss": 0.3392,
+      "step": 20255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835796281772295e-05,
+      "loss": 0.3071,
+      "step": 20260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835694727008149e-05,
+      "loss": 0.3295,
+      "step": 20265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835593141916546e-05,
+      "loss": 0.3131,
+      "step": 20270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8354915264988065e-05,
+      "loss": 0.3186,
+      "step": 20275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835389880756248e-05,
+      "loss": 0.3357,
+      "step": 20280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8352882046901915e-05,
+      "loss": 0.291,
+      "step": 20285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8351864983019556e-05,
+      "loss": 0.3085,
+      "step": 20290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.835084761592863e-05,
+      "loss": 0.3044,
+      "step": 20295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8349829945642336e-05,
+      "loss": 0.3227,
+      "step": 20300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8348811972173877e-05,
+      "loss": 0.3336,
+      "step": 20305
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.834779369553649e-05,
+      "loss": 0.286,
+      "step": 20310
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8346775115743394e-05,
+      "loss": 0.2999,
+      "step": 20315
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.8345756232807805e-05,
+      "loss": 0.3127,
+      "step": 20320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.834473704674296e-05,
+      "loss": 0.3193,
+      "step": 20325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8343717557562076e-05,
+      "loss": 0.3123,
+      "step": 20330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8342697765278424e-05,
+      "loss": 0.2938,
+      "step": 20335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.834167766990521e-05,
+      "loss": 0.3127,
+      "step": 20340
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8340657271455694e-05,
+      "loss": 0.3214,
+      "step": 20345
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833963656994313e-05,
+      "loss": 0.325,
+      "step": 20350
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833861556538076e-05,
+      "loss": 0.3356,
+      "step": 20355
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833759425778186e-05,
+      "loss": 0.3004,
+      "step": 20360
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833657264715967e-05,
+      "loss": 0.3253,
+      "step": 20365
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833555073352747e-05,
+      "loss": 0.3046,
+      "step": 20370
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8334528516898515e-05,
+      "loss": 0.3052,
+      "step": 20375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833350599728609e-05,
+      "loss": 0.3069,
+      "step": 20380
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8332483174703465e-05,
+      "loss": 0.2918,
+      "step": 20385
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.833146004916392e-05,
+      "loss": 0.2913,
+      "step": 20390
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8330436620680745e-05,
+      "loss": 0.2978,
+      "step": 20395
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832941288926723e-05,
+      "loss": 0.3143,
+      "step": 20400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8328388854936656e-05,
+      "loss": 0.288,
+      "step": 20405
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832736451770233e-05,
+      "loss": 0.302,
+      "step": 20410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832633987757755e-05,
+      "loss": 0.3215,
+      "step": 20415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832531493457561e-05,
+      "loss": 0.3222,
+      "step": 20420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832428968870984e-05,
+      "loss": 0.3041,
+      "step": 20425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832326413999354e-05,
+      "loss": 0.3167,
+      "step": 20430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832223828844003e-05,
+      "loss": 0.288,
+      "step": 20435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832121213406261e-05,
+      "loss": 0.3283,
+      "step": 20440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.832018567687463e-05,
+      "loss": 0.3207,
+      "step": 20445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.83191589168894e-05,
+      "loss": 0.3025,
+      "step": 20450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831813185412026e-05,
+      "loss": 0.2808,
+      "step": 20455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8317104488580553e-05,
+      "loss": 0.3145,
+      "step": 20460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8316076820283604e-05,
+      "loss": 0.2893,
+      "step": 20465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8315048849242774e-05,
+      "loss": 0.3034,
+      "step": 20470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8314020575471386e-05,
+      "loss": 0.2834,
+      "step": 20475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8312991998982816e-05,
+      "loss": 0.3003,
+      "step": 20480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.83119631197904e-05,
+      "loss": 0.3166,
+      "step": 20485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.831093393790751e-05,
+      "loss": 0.2674,
+      "step": 20490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8309904453347504e-05,
+      "loss": 0.2891,
+      "step": 20495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830887466612375e-05,
+      "loss": 0.2972,
+      "step": 20500
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.3011039197444916,
+      "eval_runtime": 10.5518,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 20500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8307844576249625e-05,
+      "loss": 0.2883,
+      "step": 20505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830681418373849e-05,
+      "loss": 0.2932,
+      "step": 20510
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830578348860374e-05,
+      "loss": 0.303,
+      "step": 20515
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830475249085875e-05,
+      "loss": 0.3148,
+      "step": 20520
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.83037211905169e-05,
+      "loss": 0.291,
+      "step": 20525
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.83026895875916e-05,
+      "loss": 0.2962,
+      "step": 20530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8301657682096226e-05,
+      "loss": 0.3105,
+      "step": 20535
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.830062547404418e-05,
+      "loss": 0.2883,
+      "step": 20540
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829959296344887e-05,
+      "loss": 0.2962,
+      "step": 20545
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.82985601503237e-05,
+      "loss": 0.3058,
+      "step": 20550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8297527034682086e-05,
+      "loss": 0.3153,
+      "step": 20555
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8296493616537435e-05,
+      "loss": 0.2977,
+      "step": 20560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8295459895903164e-05,
+      "loss": 0.3376,
+      "step": 20565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.82944258727927e-05,
+      "loss": 0.3026,
+      "step": 20570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8293391547219465e-05,
+      "loss": 0.2868,
+      "step": 20575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.82923569191969e-05,
+      "loss": 0.3052,
+      "step": 20580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829132198873842e-05,
+      "loss": 0.2877,
+      "step": 20585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.829028675585748e-05,
+      "loss": 0.3021,
+      "step": 20590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.828925122056752e-05,
+      "loss": 0.2839,
+      "step": 20595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8288215382881976e-05,
+      "loss": 0.3139,
+      "step": 20600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.8287179242814303e-05,
+      "loss": 0.3174,
+      "step": 20605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828614280037795e-05,
+      "loss": 0.3124,
+      "step": 20610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828510605558638e-05,
+      "loss": 0.3055,
+      "step": 20615
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828406900845306e-05,
+      "loss": 0.3304,
+      "step": 20620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8283031658991446e-05,
+      "loss": 0.2789,
+      "step": 20625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8281994007215006e-05,
+      "loss": 0.3249,
+      "step": 20630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.828095605313722e-05,
+      "loss": 0.2882,
+      "step": 20635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827991779677157e-05,
+      "loss": 0.3259,
+      "step": 20640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8278879238131514e-05,
+      "loss": 0.3139,
+      "step": 20645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8277840377230565e-05,
+      "loss": 0.2858,
+      "step": 20650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827680121408219e-05,
+      "loss": 0.3077,
+      "step": 20655
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.82757617486999e-05,
+      "loss": 0.3209,
+      "step": 20660
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827472198109717e-05,
+      "loss": 0.295,
+      "step": 20665
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8273681911287525e-05,
+      "loss": 0.3056,
+      "step": 20670
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827264153928446e-05,
+      "loss": 0.319,
+      "step": 20675
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8271600865101475e-05,
+      "loss": 0.317,
+      "step": 20680
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.827055988875209e-05,
+      "loss": 0.3192,
+      "step": 20685
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8269518610249825e-05,
+      "loss": 0.3178,
+      "step": 20690
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.82684770296082e-05,
+      "loss": 0.3107,
+      "step": 20695
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8267435146840724e-05,
+      "loss": 0.308,
+      "step": 20700
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826639296196094e-05,
+      "loss": 0.3015,
+      "step": 20705
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826535047498238e-05,
+      "loss": 0.3189,
+      "step": 20710
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826430768591858e-05,
+      "loss": 0.3386,
+      "step": 20715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8263264594783063e-05,
+      "loss": 0.3106,
+      "step": 20720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8262221201589394e-05,
+      "loss": 0.2998,
+      "step": 20725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.826117750635112e-05,
+      "loss": 0.3047,
+      "step": 20730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8260133509081776e-05,
+      "loss": 0.3295,
+      "step": 20735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8259089209794926e-05,
+      "loss": 0.2965,
+      "step": 20740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825804460850414e-05,
+      "loss": 0.3049,
+      "step": 20745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825699970522297e-05,
+      "loss": 0.3185,
+      "step": 20750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825595449996498e-05,
+      "loss": 0.3247,
+      "step": 20755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8254908992743756e-05,
+      "loss": 0.2992,
+      "step": 20760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.825386318357285e-05,
+      "loss": 0.2991,
+      "step": 20765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8252817072465864e-05,
+      "loss": 0.3054,
+      "step": 20770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8251770659436375e-05,
+      "loss": 0.3022,
+      "step": 20775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8250723944497964e-05,
+      "loss": 0.3082,
+      "step": 20780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824988635518195e-05,
+      "loss": 0.2986,
+      "step": 20785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8248839096841736e-05,
+      "loss": 0.2889,
+      "step": 20790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8247791536630674e-05,
+      "loss": 0.3032,
+      "step": 20795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824674367456235e-05,
+      "loss": 0.3027,
+      "step": 20800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824569551065039e-05,
+      "loss": 0.3137,
+      "step": 20805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824464704490838e-05,
+      "loss": 0.3047,
+      "step": 20810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824359827734997e-05,
+      "loss": 0.3319,
+      "step": 20815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8242549207988735e-05,
+      "loss": 0.2975,
+      "step": 20820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824149983683832e-05,
+      "loss": 0.3049,
+      "step": 20825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.824045016391235e-05,
+      "loss": 0.2976,
+      "step": 20830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823940018922445e-05,
+      "loss": 0.3072,
+      "step": 20835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823834991278825e-05,
+      "loss": 0.3261,
+      "step": 20840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8237299334617395e-05,
+      "loss": 0.3124,
+      "step": 20845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823624845472552e-05,
+      "loss": 0.3198,
+      "step": 20850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823519727312627e-05,
+      "loss": 0.3064,
+      "step": 20855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8234145789833296e-05,
+      "loss": 0.2938,
+      "step": 20860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8233094004860256e-05,
+      "loss": 0.3166,
+      "step": 20865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823225235968135e-05,
+      "loss": 0.3031,
+      "step": 20870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.823120003171859e-05,
+      "loss": 0.2976,
+      "step": 20875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.8230147402114024e-05,
+      "loss": 0.3039,
+      "step": 20880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.822909447088129e-05,
+      "loss": 0.2846,
+      "step": 20885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.822804123803407e-05,
+      "loss": 0.3226,
+      "step": 20890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8226987703586056e-05,
+      "loss": 0.2975,
+      "step": 20895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.82259338675509e-05,
+      "loss": 0.3286,
+      "step": 20900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8224879729942307e-05,
+      "loss": 0.2886,
+      "step": 20905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.822382529077396e-05,
+      "loss": 0.2958,
+      "step": 20910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8222770550059537e-05,
+      "loss": 0.3194,
+      "step": 20915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8221715507812746e-05,
+      "loss": 0.3349,
+      "step": 20920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.822066016404729e-05,
+      "loss": 0.2894,
+      "step": 20925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821960451877686e-05,
+      "loss": 0.2959,
+      "step": 20930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821854857201516e-05,
+      "loss": 0.2829,
+      "step": 20935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821749232377592e-05,
+      "loss": 0.3081,
+      "step": 20940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821643577407284e-05,
+      "loss": 0.3231,
+      "step": 20945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821537892291964e-05,
+      "loss": 0.3026,
+      "step": 20950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821432177033004e-05,
+      "loss": 0.3115,
+      "step": 20955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8213264316317776e-05,
+      "loss": 0.3078,
+      "step": 20960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821220656089657e-05,
+      "loss": 0.3144,
+      "step": 20965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8211148504080164e-05,
+      "loss": 0.3014,
+      "step": 20970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.821009014588229e-05,
+      "loss": 0.2936,
+      "step": 20975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820903148631669e-05,
+      "loss": 0.2923,
+      "step": 20980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820797252539711e-05,
+      "loss": 0.3053,
+      "step": 20985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.82069132631373e-05,
+      "loss": 0.3627,
+      "step": 20990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8205853699551016e-05,
+      "loss": 0.291,
+      "step": 20995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820479383465202e-05,
+      "loss": 0.3204,
+      "step": 21000
+    },
+    {
+      "epoch": 0.74,
+      "eval_loss": 0.29989683628082275,
+      "eval_runtime": 10.5493,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 21000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8203733668454066e-05,
+      "loss": 0.298,
+      "step": 21005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8202673200970925e-05,
+      "loss": 0.2888,
+      "step": 21010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820161243221636e-05,
+      "loss": 0.2988,
+      "step": 21015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.820055136220415e-05,
+      "loss": 0.3334,
+      "step": 21020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819948999094807e-05,
+      "loss": 0.3501,
+      "step": 21025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.81984283184619e-05,
+      "loss": 0.3129,
+      "step": 21030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819736634475942e-05,
+      "loss": 0.2927,
+      "step": 21035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819630406985444e-05,
+      "loss": 0.2925,
+      "step": 21040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819524149376072e-05,
+      "loss": 0.3117,
+      "step": 21045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8194178616492094e-05,
+      "loss": 0.2986,
+      "step": 21050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819311543806233e-05,
+      "loss": 0.3166,
+      "step": 21055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.819205195848526e-05,
+      "loss": 0.3084,
+      "step": 21060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8190988177774666e-05,
+      "loss": 0.3021,
+      "step": 21065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.818992409594438e-05,
+      "loss": 0.289,
+      "step": 21070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8188859713008215e-05,
+      "loss": 0.309,
+      "step": 21075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8187795028979985e-05,
+      "loss": 0.3262,
+      "step": 21080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.818673004387352e-05,
+      "loss": 0.3161,
+      "step": 21085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8185664757702654e-05,
+      "loss": 0.2769,
+      "step": 21090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.81845991704812e-05,
+      "loss": 0.3011,
+      "step": 21095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8183533282223e-05,
+      "loss": 0.3161,
+      "step": 21100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8182467092941915e-05,
+      "loss": 0.298,
+      "step": 21105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.818140060265176e-05,
+      "loss": 0.3351,
+      "step": 21110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.81803338113664e-05,
+      "loss": 0.3028,
+      "step": 21115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817926671909968e-05,
+      "loss": 0.3052,
+      "step": 21120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8178199325865457e-05,
+      "loss": 0.3117,
+      "step": 21125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817713163167759e-05,
+      "loss": 0.2919,
+      "step": 21130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817606363654995e-05,
+      "loss": 0.3248,
+      "step": 21135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.81749953404964e-05,
+      "loss": 0.2991,
+      "step": 21140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8173926743530794e-05,
+      "loss": 0.3388,
+      "step": 21145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817285784566703e-05,
+      "loss": 0.3201,
+      "step": 21150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817178864691898e-05,
+      "loss": 0.3244,
+      "step": 21155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.817071914730052e-05,
+      "loss": 0.3065,
+      "step": 21160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.816964934682555e-05,
+      "loss": 0.2745,
+      "step": 21165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.816857924550794e-05,
+      "loss": 0.3042,
+      "step": 21170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.8167508843361606e-05,
+      "loss": 0.3095,
+      "step": 21175
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.816643814040044e-05,
+      "loss": 0.289,
+      "step": 21180
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8165367136638335e-05,
+      "loss": 0.3259,
+      "step": 21185
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.81642958320892e-05,
+      "loss": 0.313,
+      "step": 21190
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8163224226766955e-05,
+      "loss": 0.2789,
+      "step": 21195
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8162152320685515e-05,
+      "loss": 0.3211,
+      "step": 21200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8161080113858786e-05,
+      "loss": 0.3257,
+      "step": 21205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8160007606300694e-05,
+      "loss": 0.3075,
+      "step": 21210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815893479802517e-05,
+      "loss": 0.3072,
+      "step": 21215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815786168904613e-05,
+      "loss": 0.3119,
+      "step": 21220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8156788279377534e-05,
+      "loss": 0.3135,
+      "step": 21225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8155714569033294e-05,
+      "loss": 0.2867,
+      "step": 21230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815464055802736e-05,
+      "loss": 0.2794,
+      "step": 21235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815356624637368e-05,
+      "loss": 0.3002,
+      "step": 21240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815249163408621e-05,
+      "loss": 0.3174,
+      "step": 21245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.815141672117888e-05,
+      "loss": 0.2666,
+      "step": 21250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8150341507665664e-05,
+      "loss": 0.3275,
+      "step": 21255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.814926599356052e-05,
+      "loss": 0.2986,
+      "step": 21260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8148190178877414e-05,
+      "loss": 0.3171,
+      "step": 21265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.814711406363032e-05,
+      "loss": 0.3302,
+      "step": 21270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.81460376478332e-05,
+      "loss": 0.3228,
+      "step": 21275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8144960931500026e-05,
+      "loss": 0.3102,
+      "step": 21280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8143883914644796e-05,
+      "loss": 0.2783,
+      "step": 21285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.814280659728149e-05,
+      "loss": 0.2945,
+      "step": 21290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8141728979424085e-05,
+      "loss": 0.2845,
+      "step": 21295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8140651061086584e-05,
+      "loss": 0.309,
+      "step": 21300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8139572842282975e-05,
+      "loss": 0.3261,
+      "step": 21305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813849432302726e-05,
+      "loss": 0.3056,
+      "step": 21310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813741550333345e-05,
+      "loss": 0.2973,
+      "step": 21315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8136336383215554e-05,
+      "loss": 0.3224,
+      "step": 21320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813525696268757e-05,
+      "loss": 0.2816,
+      "step": 21325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813417724176352e-05,
+      "loss": 0.291,
+      "step": 21330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813309722045742e-05,
+      "loss": 0.3231,
+      "step": 21335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.813201689878331e-05,
+      "loss": 0.2945,
+      "step": 21340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8130936276755204e-05,
+      "loss": 0.3041,
+      "step": 21345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8129855354387125e-05,
+      "loss": 0.2851,
+      "step": 21350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812877413169312e-05,
+      "loss": 0.3145,
+      "step": 21355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812769260868723e-05,
+      "loss": 0.3013,
+      "step": 21360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812661078538349e-05,
+      "loss": 0.2727,
+      "step": 21365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812552866179595e-05,
+      "loss": 0.3139,
+      "step": 21370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8124446237938665e-05,
+      "loss": 0.3225,
+      "step": 21375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812336351382568e-05,
+      "loss": 0.3011,
+      "step": 21380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8122280489471064e-05,
+      "loss": 0.3018,
+      "step": 21385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812119716488886e-05,
+      "loss": 0.2842,
+      "step": 21390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.812011354009317e-05,
+      "loss": 0.2962,
+      "step": 21395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8119029615098024e-05,
+      "loss": 0.3168,
+      "step": 21400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811794538991752e-05,
+      "loss": 0.2889,
+      "step": 21405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811686086456574e-05,
+      "loss": 0.3059,
+      "step": 21410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811577603905674e-05,
+      "loss": 0.2992,
+      "step": 21415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8114690913404634e-05,
+      "loss": 0.3163,
+      "step": 21420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811360548762349e-05,
+      "loss": 0.2916,
+      "step": 21425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811251976172742e-05,
+      "loss": 0.2905,
+      "step": 21430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811143373573051e-05,
+      "loss": 0.313,
+      "step": 21435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.811034740964686e-05,
+      "loss": 0.3108,
+      "step": 21440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.810926078349058e-05,
+      "loss": 0.3084,
+      "step": 21445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.8108173857275785e-05,
+      "loss": 0.3066,
+      "step": 21450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.810708663101658e-05,
+      "loss": 0.3084,
+      "step": 21455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8105999104727086e-05,
+      "loss": 0.2944,
+      "step": 21460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810491127842141e-05,
+      "loss": 0.317,
+      "step": 21465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8103823152113695e-05,
+      "loss": 0.3204,
+      "step": 21470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8102734725818056e-05,
+      "loss": 0.3106,
+      "step": 21475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810164599954864e-05,
+      "loss": 0.3186,
+      "step": 21480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810055697331957e-05,
+      "loss": 0.2905,
+      "step": 21485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809946764714499e-05,
+      "loss": 0.3014,
+      "step": 21490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809837802103905e-05,
+      "loss": 0.2906,
+      "step": 21495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80972880950159e-05,
+      "loss": 0.2881,
+      "step": 21500
+    },
+    {
+      "epoch": 0.76,
+      "eval_loss": 0.30054381489753723,
+      "eval_runtime": 10.5336,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 21500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809619786908968e-05,
+      "loss": 0.2831,
+      "step": 21505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809510734327455e-05,
+      "loss": 0.3034,
+      "step": 21510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809401651758466e-05,
+      "loss": 0.3227,
+      "step": 21515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809292539203421e-05,
+      "loss": 0.3268,
+      "step": 21520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809183396663732e-05,
+      "loss": 0.3147,
+      "step": 21525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.809074224140819e-05,
+      "loss": 0.2981,
+      "step": 21530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808965021636099e-05,
+      "loss": 0.2945,
+      "step": 21535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808855789150991e-05,
+      "loss": 0.3062,
+      "step": 21540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8087465266869115e-05,
+      "loss": 0.3048,
+      "step": 21545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808637234245279e-05,
+      "loss": 0.3082,
+      "step": 21550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808527911827514e-05,
+      "loss": 0.2964,
+      "step": 21555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808418559435036e-05,
+      "loss": 0.2923,
+      "step": 21560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8083091770692635e-05,
+      "loss": 0.3136,
+      "step": 21565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808199764731618e-05,
+      "loss": 0.3084,
+      "step": 21570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80809032242352e-05,
+      "loss": 0.3067,
+      "step": 21575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8079808501463896e-05,
+      "loss": 0.2922,
+      "step": 21580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807871347901649e-05,
+      "loss": 0.2992,
+      "step": 21585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807761815690721e-05,
+      "loss": 0.3009,
+      "step": 21590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807652253515025e-05,
+      "loss": 0.2782,
+      "step": 21595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8075426613759865e-05,
+      "loss": 0.2967,
+      "step": 21600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807433039275027e-05,
+      "loss": 0.3211,
+      "step": 21605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80732338721357e-05,
+      "loss": 0.3165,
+      "step": 21610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.807213705193039e-05,
+      "loss": 0.3044,
+      "step": 21615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8071039932148585e-05,
+      "loss": 0.3058,
+      "step": 21620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806994251280454e-05,
+      "loss": 0.3035,
+      "step": 21625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806884479391249e-05,
+      "loss": 0.3142,
+      "step": 21630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8067746775486696e-05,
+      "loss": 0.3171,
+      "step": 21635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806664845754141e-05,
+      "loss": 0.3367,
+      "step": 21640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806554984009089e-05,
+      "loss": 0.3105,
+      "step": 21645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806445092314941e-05,
+      "loss": 0.3224,
+      "step": 21650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806335170673124e-05,
+      "loss": 0.2948,
+      "step": 21655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806225219085064e-05,
+      "loss": 0.2976,
+      "step": 21660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80611523755219e-05,
+      "loss": 0.3046,
+      "step": 21665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806005226075929e-05,
+      "loss": 0.3117,
+      "step": 21670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.80589518465771e-05,
+      "loss": 0.3219,
+      "step": 21675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805785113298962e-05,
+      "loss": 0.3039,
+      "step": 21680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805675012001113e-05,
+      "loss": 0.3166,
+      "step": 21685
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805564880765594e-05,
+      "loss": 0.299,
+      "step": 21690
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8054547195938345e-05,
+      "loss": 0.3175,
+      "step": 21695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805344528487266e-05,
+      "loss": 0.2838,
+      "step": 21700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805234307447316e-05,
+      "loss": 0.2707,
+      "step": 21705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8051240564754184e-05,
+      "loss": 0.3001,
+      "step": 21710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805013775573004e-05,
+      "loss": 0.3248,
+      "step": 21715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804903464741506e-05,
+      "loss": 0.2908,
+      "step": 21720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804793123982354e-05,
+      "loss": 0.3068,
+      "step": 21725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804682753296983e-05,
+      "loss": 0.3034,
+      "step": 21730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8045723526868245e-05,
+      "loss": 0.3117,
+      "step": 21735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8044619221533135e-05,
+      "loss": 0.2958,
+      "step": 21740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.804351461697883e-05,
+      "loss": 0.3404,
+      "step": 21745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.804240971321967e-05,
+      "loss": 0.2699,
+      "step": 21750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8041304510270016e-05,
+      "loss": 0.3103,
+      "step": 21755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.804019900814419e-05,
+      "loss": 0.2829,
+      "step": 21760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803909320685658e-05,
+      "loss": 0.3489,
+      "step": 21765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8037987106421525e-05,
+      "loss": 0.3017,
+      "step": 21770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8036880706853385e-05,
+      "loss": 0.3193,
+      "step": 21775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803577400816654e-05,
+      "loss": 0.2851,
+      "step": 21780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803466701037534e-05,
+      "loss": 0.3184,
+      "step": 21785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803355971349417e-05,
+      "loss": 0.3362,
+      "step": 21790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803245211753741e-05,
+      "loss": 0.2961,
+      "step": 21795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8031344222519434e-05,
+      "loss": 0.2965,
+      "step": 21800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.803023602845465e-05,
+      "loss": 0.331,
+      "step": 21805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802912753535741e-05,
+      "loss": 0.3147,
+      "step": 21810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802801874324213e-05,
+      "loss": 0.3112,
+      "step": 21815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8026909652123196e-05,
+      "loss": 0.316,
+      "step": 21820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802580026201502e-05,
+      "loss": 0.3191,
+      "step": 21825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802469057293201e-05,
+      "loss": 0.3106,
+      "step": 21830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802358058488856e-05,
+      "loss": 0.3008,
+      "step": 21835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802247029789909e-05,
+      "loss": 0.3206,
+      "step": 21840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802135971197802e-05,
+      "loss": 0.3182,
+      "step": 21845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.802024882713976e-05,
+      "loss": 0.3298,
+      "step": 21850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8019137643398736e-05,
+      "loss": 0.3019,
+      "step": 21855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801802616076939e-05,
+      "loss": 0.2883,
+      "step": 21860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801691437926614e-05,
+      "loss": 0.2949,
+      "step": 21865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801580229890342e-05,
+      "loss": 0.2882,
+      "step": 21870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8014689919695685e-05,
+      "loss": 0.301,
+      "step": 21875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801357724165736e-05,
+      "loss": 0.3056,
+      "step": 21880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8012464264802906e-05,
+      "loss": 0.3166,
+      "step": 21885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801135098914677e-05,
+      "loss": 0.2869,
+      "step": 21890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.80102374147034e-05,
+      "loss": 0.2947,
+      "step": 21895
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800912354148727e-05,
+      "loss": 0.2838,
+      "step": 21900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800800936951283e-05,
+      "loss": 0.308,
+      "step": 21905
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8006894898794554e-05,
+      "loss": 0.3103,
+      "step": 21910
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.80057801293469e-05,
+      "loss": 0.3056,
+      "step": 21915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800466506118436e-05,
+      "loss": 0.3012,
+      "step": 21920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800354969432141e-05,
+      "loss": 0.2974,
+      "step": 21925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800243402877252e-05,
+      "loss": 0.2985,
+      "step": 21930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8001318064552184e-05,
+      "loss": 0.2776,
+      "step": 21935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.800020180167489e-05,
+      "loss": 0.2897,
+      "step": 21940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7999085240155137e-05,
+      "loss": 0.2792,
+      "step": 21945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7997968380007416e-05,
+      "loss": 0.3092,
+      "step": 21950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799685122124624e-05,
+      "loss": 0.3157,
+      "step": 21955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7995733763886095e-05,
+      "loss": 0.3111,
+      "step": 21960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.79946160079415e-05,
+      "loss": 0.2896,
+      "step": 21965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7993497953426986e-05,
+      "loss": 0.3364,
+      "step": 21970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799237960035704e-05,
+      "loss": 0.3244,
+      "step": 21975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.799126094874621e-05,
+      "loss": 0.3001,
+      "step": 21980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7990141998609e-05,
+      "loss": 0.3243,
+      "step": 21985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798902274995995e-05,
+      "loss": 0.3208,
+      "step": 21990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7987903202813586e-05,
+      "loss": 0.318,
+      "step": 21995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798678335718445e-05,
+      "loss": 0.3,
+      "step": 22000
+    },
+    {
+      "epoch": 0.77,
+      "eval_loss": 0.30042049288749695,
+      "eval_runtime": 10.5417,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 22000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7985663213087085e-05,
+      "loss": 0.3042,
+      "step": 22005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798454277053602e-05,
+      "loss": 0.3175,
+      "step": 22010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798342202954583e-05,
+      "loss": 0.3416,
+      "step": 22015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.798230099013104e-05,
+      "loss": 0.3022,
+      "step": 22020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7981179652306215e-05,
+      "loss": 0.302,
+      "step": 22025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7980058016085925e-05,
+      "loss": 0.29,
+      "step": 22030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797893608148473e-05,
+      "loss": 0.3239,
+      "step": 22035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797781384851719e-05,
+      "loss": 0.3278,
+      "step": 22040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7976691317197874e-05,
+      "loss": 0.3006,
+      "step": 22045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797556848754137e-05,
+      "loss": 0.3105,
+      "step": 22050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7974445359562256e-05,
+      "loss": 0.3179,
+      "step": 22055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.797332193327509e-05,
+      "loss": 0.3234,
+      "step": 22060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7972198208694494e-05,
+      "loss": 0.3119,
+      "step": 22065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7971074185835045e-05,
+      "loss": 0.3022,
+      "step": 22070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7969949864711335e-05,
+      "loss": 0.2895,
+      "step": 22075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796882524533796e-05,
+      "loss": 0.344,
+      "step": 22080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796770032772953e-05,
+      "loss": 0.3162,
+      "step": 22085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796657511190064e-05,
+      "loss": 0.293,
+      "step": 22090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7965449597865914e-05,
+      "loss": 0.3211,
+      "step": 22095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796432378563996e-05,
+      "loss": 0.3143,
+      "step": 22100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796319767523739e-05,
+      "loss": 0.3107,
+      "step": 22105
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.796207126667284e-05,
+      "loss": 0.2938,
+      "step": 22110
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7960944559960915e-05,
+      "loss": 0.3009,
+      "step": 22115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795981755511627e-05,
+      "loss": 0.3159,
+      "step": 22120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7958690252153514e-05,
+      "loss": 0.3194,
+      "step": 22125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7957562651087304e-05,
+      "loss": 0.3359,
+      "step": 22130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795643475193227e-05,
+      "loss": 0.3258,
+      "step": 22135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795530655470306e-05,
+      "loss": 0.304,
+      "step": 22140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795417805941432e-05,
+      "loss": 0.2929,
+      "step": 22145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795304926608071e-05,
+      "loss": 0.3054,
+      "step": 22150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795192017471688e-05,
+      "loss": 0.3056,
+      "step": 22155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.795079078533749e-05,
+      "loss": 0.2981,
+      "step": 22160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7949661097957205e-05,
+      "loss": 0.2938,
+      "step": 22165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79485311125907e-05,
+      "loss": 0.2924,
+      "step": 22170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7947400829252636e-05,
+      "loss": 0.3178,
+      "step": 22175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7946270247957704e-05,
+      "loss": 0.3124,
+      "step": 22180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7945139368720573e-05,
+      "loss": 0.3451,
+      "step": 22185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.794400819155592e-05,
+      "loss": 0.2928,
+      "step": 22190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7942876716478444e-05,
+      "loss": 0.3024,
+      "step": 22195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.794174494350284e-05,
+      "loss": 0.3331,
+      "step": 22200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79406128726438e-05,
+      "loss": 0.2988,
+      "step": 22205
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793948050391601e-05,
+      "loss": 0.2974,
+      "step": 22210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7938347837334186e-05,
+      "loss": 0.2996,
+      "step": 22215
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793721487291303e-05,
+      "loss": 0.3041,
+      "step": 22220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793608161066726e-05,
+      "loss": 0.2966,
+      "step": 22225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7934948050611586e-05,
+      "loss": 0.3155,
+      "step": 22230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793381419276073e-05,
+      "loss": 0.3225,
+      "step": 22235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.79326800371294e-05,
+      "loss": 0.2849,
+      "step": 22240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793154558373234e-05,
+      "loss": 0.2952,
+      "step": 22245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.793041083258427e-05,
+      "loss": 0.296,
+      "step": 22250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7929275783699935e-05,
+      "loss": 0.3116,
+      "step": 22255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7928140437094055e-05,
+      "loss": 0.2974,
+      "step": 22260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792700479278138e-05,
+      "loss": 0.2694,
+      "step": 22265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7925868850776664e-05,
+      "loss": 0.3148,
+      "step": 22270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792473261109465e-05,
+      "loss": 0.3181,
+      "step": 22275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792359607375009e-05,
+      "loss": 0.2937,
+      "step": 22280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7922459238757746e-05,
+      "loss": 0.294,
+      "step": 22285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7921322106132366e-05,
+      "loss": 0.3054,
+      "step": 22290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.792018467588873e-05,
+      "loss": 0.3309,
+      "step": 22295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.7919046948041605e-05,
+      "loss": 0.2885,
+      "step": 22300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.791790892260575e-05,
+      "loss": 0.3064,
+      "step": 22305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.791677059959595e-05,
+      "loss": 0.3486,
+      "step": 22310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7915631979027e-05,
+      "loss": 0.3017,
+      "step": 22315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7914493060913654e-05,
+      "loss": 0.2882,
+      "step": 22320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7913353845270725e-05,
+      "loss": 0.3062,
+      "step": 22325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7912214332113e-05,
+      "loss": 0.3241,
+      "step": 22330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7911074521455266e-05,
+      "loss": 0.2649,
+      "step": 22335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7909934413312335e-05,
+      "loss": 0.32,
+      "step": 22340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7908794007698996e-05,
+      "loss": 0.2956,
+      "step": 22345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7907653304630074e-05,
+      "loss": 0.3181,
+      "step": 22350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790651230412037e-05,
+      "loss": 0.2962,
+      "step": 22355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790537100618469e-05,
+      "loss": 0.3048,
+      "step": 22360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790422941083786e-05,
+      "loss": 0.3162,
+      "step": 22365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790308751809471e-05,
+      "loss": 0.3088,
+      "step": 22370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790194532797007e-05,
+      "loss": 0.3218,
+      "step": 22375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.790080284047876e-05,
+      "loss": 0.2781,
+      "step": 22380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789966005563561e-05,
+      "loss": 0.2965,
+      "step": 22385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7898516973455474e-05,
+      "loss": 0.3253,
+      "step": 22390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789737359395319e-05,
+      "loss": 0.2921,
+      "step": 22395
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789622991714359e-05,
+      "loss": 0.2875,
+      "step": 22400
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789508594304154e-05,
+      "loss": 0.2885,
+      "step": 22405
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789394167166188e-05,
+      "loss": 0.3002,
+      "step": 22410
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7892797103019476e-05,
+      "loss": 0.3167,
+      "step": 22415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7891652237129194e-05,
+      "loss": 0.3041,
+      "step": 22420
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.789050707400588e-05,
+      "loss": 0.3157,
+      "step": 22425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7889361613664436e-05,
+      "loss": 0.3145,
+      "step": 22430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7888215856119714e-05,
+      "loss": 0.2968,
+      "step": 22435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7887069801386585e-05,
+      "loss": 0.3106,
+      "step": 22440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7885923449479936e-05,
+      "loss": 0.3012,
+      "step": 22445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788477680041465e-05,
+      "loss": 0.28,
+      "step": 22450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788362985420564e-05,
+      "loss": 0.2875,
+      "step": 22455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788248261086776e-05,
+      "loss": 0.3307,
+      "step": 22460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788133507041592e-05,
+      "loss": 0.2875,
+      "step": 22465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.788018723286504e-05,
+      "loss": 0.2809,
+      "step": 22470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787903909822999e-05,
+      "loss": 0.2954,
+      "step": 22475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7877890666525703e-05,
+      "loss": 0.2881,
+      "step": 22480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787674193776709e-05,
+      "loss": 0.2969,
+      "step": 22485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7875592911969045e-05,
+      "loss": 0.3009,
+      "step": 22490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.787444358914651e-05,
+      "loss": 0.3125,
+      "step": 22495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78732939693144e-05,
+      "loss": 0.3025,
+      "step": 22500
+    },
+    {
+      "epoch": 0.79,
+      "eval_loss": 0.2996106445789337,
+      "eval_runtime": 10.5303,
+      "eval_samples_per_second": 9.496,
+      "eval_steps_per_second": 9.496,
+      "step": 22500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7872144052487636e-05,
+      "loss": 0.2975,
+      "step": 22505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7870993838681166e-05,
+      "loss": 0.3188,
+      "step": 22510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7869843327909904e-05,
+      "loss": 0.3126,
+      "step": 22515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78686925201888e-05,
+      "loss": 0.3167,
+      "step": 22520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7867541415532796e-05,
+      "loss": 0.3221,
+      "step": 22525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786639001395684e-05,
+      "loss": 0.2836,
+      "step": 22530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786523831547587e-05,
+      "loss": 0.2995,
+      "step": 22535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7864086320104853e-05,
+      "loss": 0.3091,
+      "step": 22540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786293402785875e-05,
+      "loss": 0.2919,
+      "step": 22545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786178143875251e-05,
+      "loss": 0.2988,
+      "step": 22550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.786062855280111e-05,
+      "loss": 0.3053,
+      "step": 22555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.78594753700195e-05,
+      "loss": 0.3258,
+      "step": 22560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785832189042268e-05,
+      "loss": 0.303,
+      "step": 22565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785716811402561e-05,
+      "loss": 0.2957,
+      "step": 22570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785601404084328e-05,
+      "loss": 0.3172,
+      "step": 22575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785485967089066e-05,
+      "loss": 0.3203,
+      "step": 22580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.7853705004182756e-05,
+      "loss": 0.3185,
+      "step": 22585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785255004073455e-05,
+      "loss": 0.3293,
+      "step": 22590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.785139478056105e-05,
+      "loss": 0.2998,
+      "step": 22595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.785023922367725e-05,
+      "loss": 0.3287,
+      "step": 22600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784908337009815e-05,
+      "loss": 0.2895,
+      "step": 22605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784792721983875e-05,
+      "loss": 0.3014,
+      "step": 22610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784677077291408e-05,
+      "loss": 0.3052,
+      "step": 22615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784561402933915e-05,
+      "loss": 0.2981,
+      "step": 22620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784445698912898e-05,
+      "loss": 0.2791,
+      "step": 22625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784329965229858e-05,
+      "loss": 0.3382,
+      "step": 22630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7842142018863e-05,
+      "loss": 0.3277,
+      "step": 22635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.784098408883725e-05,
+      "loss": 0.2887,
+      "step": 22640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783982586223638e-05,
+      "loss": 0.3109,
+      "step": 22645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783866733907543e-05,
+      "loss": 0.2923,
+      "step": 22650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7837508519369425e-05,
+      "loss": 0.3152,
+      "step": 22655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783634940313342e-05,
+      "loss": 0.3044,
+      "step": 22660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7835189990382475e-05,
+      "loss": 0.3201,
+      "step": 22665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7834030281131636e-05,
+      "loss": 0.3069,
+      "step": 22670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7832870275395956e-05,
+      "loss": 0.3071,
+      "step": 22675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783170997319051e-05,
+      "loss": 0.306,
+      "step": 22680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.783054937453035e-05,
+      "loss": 0.3189,
+      "step": 22685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782938847943056e-05,
+      "loss": 0.3142,
+      "step": 22690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7828227287906205e-05,
+      "loss": 0.2958,
+      "step": 22695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782706579997236e-05,
+      "loss": 0.3093,
+      "step": 22700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7825904015644106e-05,
+      "loss": 0.3015,
+      "step": 22705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782474193493654e-05,
+      "loss": 0.3103,
+      "step": 22710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7823579557864725e-05,
+      "loss": 0.274,
+      "step": 22715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7822416884443786e-05,
+      "loss": 0.2857,
+      "step": 22720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.78212539146888e-05,
+      "loss": 0.3126,
+      "step": 22725
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.782009064861488e-05,
+      "loss": 0.3157,
+      "step": 22730
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7818927086237106e-05,
+      "loss": 0.3064,
+      "step": 22735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7817763227570614e-05,
+      "loss": 0.3148,
+      "step": 22740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.78165990726305e-05,
+      "loss": 0.2939,
+      "step": 22745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7815434621431884e-05,
+      "loss": 0.2772,
+      "step": 22750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7814269873989895e-05,
+      "loss": 0.2954,
+      "step": 22755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7813104830319634e-05,
+      "loss": 0.3113,
+      "step": 22760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781193949043624e-05,
+      "loss": 0.3328,
+      "step": 22765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.781077385435485e-05,
+      "loss": 0.3097,
+      "step": 22770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780960792209059e-05,
+      "loss": 0.3012,
+      "step": 22775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780844169365861e-05,
+      "loss": 0.3298,
+      "step": 22780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780727516907405e-05,
+      "loss": 0.2949,
+      "step": 22785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780610834835204e-05,
+      "loss": 0.321,
+      "step": 22790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780494123150775e-05,
+      "loss": 0.3075,
+      "step": 22795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780377381855633e-05,
+      "loss": 0.3277,
+      "step": 22800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.780260610951293e-05,
+      "loss": 0.3016,
+      "step": 22805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7801438104392715e-05,
+      "loss": 0.3104,
+      "step": 22810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7800269803210856e-05,
+      "loss": 0.3077,
+      "step": 22815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779910120598252e-05,
+      "loss": 0.3111,
+      "step": 22820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779793231272288e-05,
+      "loss": 0.2961,
+      "step": 22825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779676312344711e-05,
+      "loss": 0.3126,
+      "step": 22830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.77955936381704e-05,
+      "loss": 0.3014,
+      "step": 22835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779442385690793e-05,
+      "loss": 0.3251,
+      "step": 22840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779325377967488e-05,
+      "loss": 0.3128,
+      "step": 22845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779208340648646e-05,
+      "loss": 0.3177,
+      "step": 22850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.779091273735785e-05,
+      "loss": 0.324,
+      "step": 22855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7789741772304265e-05,
+      "loss": 0.2845,
+      "step": 22860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7788570511340905e-05,
+      "loss": 0.2829,
+      "step": 22865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.778739895448296e-05,
+      "loss": 0.2705,
+      "step": 22870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7786227101745666e-05,
+      "loss": 0.3156,
+      "step": 22875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.778505495314423e-05,
+      "loss": 0.2754,
+      "step": 22880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7783882508693876e-05,
+      "loss": 0.2873,
+      "step": 22885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7782709768409815e-05,
+      "loss": 0.3089,
+      "step": 22890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.778153673230728e-05,
+      "loss": 0.3223,
+      "step": 22895
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.778036340040152e-05,
+      "loss": 0.3066,
+      "step": 22900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777918977270773e-05,
+      "loss": 0.2998,
+      "step": 22905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777801584924119e-05,
+      "loss": 0.289,
+      "step": 22910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777684163001712e-05,
+      "loss": 0.3045,
+      "step": 22915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777566711505078e-05,
+      "loss": 0.315,
+      "step": 22920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7774492304357396e-05,
+      "loss": 0.2783,
+      "step": 22925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7773317197952256e-05,
+      "loss": 0.3309,
+      "step": 22930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777214179585059e-05,
+      "loss": 0.3209,
+      "step": 22935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.777096609806767e-05,
+      "loss": 0.2981,
+      "step": 22940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7769790104618764e-05,
+      "loss": 0.3165,
+      "step": 22945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7768613815519145e-05,
+      "loss": 0.2977,
+      "step": 22950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7767437230784075e-05,
+      "loss": 0.3124,
+      "step": 22955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.776626035042884e-05,
+      "loss": 0.2972,
+      "step": 22960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.776508317446871e-05,
+      "loss": 0.2934,
+      "step": 22965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7763905702918984e-05,
+      "loss": 0.2941,
+      "step": 22970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.776272793579494e-05,
+      "loss": 0.3158,
+      "step": 22975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.776154987311188e-05,
+      "loss": 0.3337,
+      "step": 22980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7760371514885096e-05,
+      "loss": 0.3142,
+      "step": 22985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7759192861129886e-05,
+      "loss": 0.308,
+      "step": 22990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775801391186156e-05,
+      "loss": 0.3155,
+      "step": 22995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775683466709541e-05,
+      "loss": 0.3034,
+      "step": 23000
+    },
+    {
+      "epoch": 0.81,
+      "eval_loss": 0.2990756332874298,
+      "eval_runtime": 10.544,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 23000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7755655126846766e-05,
+      "loss": 0.3311,
+      "step": 23005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775447529113093e-05,
+      "loss": 0.3021,
+      "step": 23010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775329515996324e-05,
+      "loss": 0.3034,
+      "step": 23015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7752114733358996e-05,
+      "loss": 0.3255,
+      "step": 23020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.775093401133355e-05,
+      "loss": 0.3309,
+      "step": 23025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774975299390221e-05,
+      "loss": 0.3222,
+      "step": 23030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774857168108032e-05,
+      "loss": 0.3248,
+      "step": 23035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7747390072883217e-05,
+      "loss": 0.2998,
+      "step": 23040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7746208169326246e-05,
+      "loss": 0.2918,
+      "step": 23045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7745025970424756e-05,
+      "loss": 0.3339,
+      "step": 23050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7743843476194096e-05,
+      "loss": 0.2757,
+      "step": 23055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.774266068664961e-05,
+      "loss": 0.3043,
+      "step": 23060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7741477601806664e-05,
+      "loss": 0.3021,
+      "step": 23065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7740294221680616e-05,
+      "loss": 0.2808,
+      "step": 23070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7739110546286844e-05,
+      "loss": 0.2895,
+      "step": 23075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7737926575640705e-05,
+      "loss": 0.3191,
+      "step": 23080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7736742309757565e-05,
+      "loss": 0.3305,
+      "step": 23085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7735557748652815e-05,
+      "loss": 0.3037,
+      "step": 23090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.773437289234183e-05,
+      "loss": 0.2879,
+      "step": 23095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7733187740839996e-05,
+      "loss": 0.3036,
+      "step": 23100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7732002294162706e-05,
+      "loss": 0.3107,
+      "step": 23105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.773081655232534e-05,
+      "loss": 0.3171,
+      "step": 23110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.77296305153433e-05,
+      "loss": 0.3143,
+      "step": 23115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7728444183232e-05,
+      "loss": 0.3031,
+      "step": 23120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772725755600682e-05,
+      "loss": 0.3024,
+      "step": 23125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772607063368318e-05,
+      "loss": 0.2787,
+      "step": 23130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772488341627649e-05,
+      "loss": 0.3211,
+      "step": 23135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.772369590380217e-05,
+      "loss": 0.2972,
+      "step": 23140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7722508096275626e-05,
+      "loss": 0.2936,
+      "step": 23145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7721319993712286e-05,
+      "loss": 0.3176,
+      "step": 23150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.7720131596127594e-05,
+      "loss": 0.2945,
+      "step": 23155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.771894290353695e-05,
+      "loss": 0.2996,
+      "step": 23160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.771775391595582e-05,
+      "loss": 0.3121,
+      "step": 23165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.771656463339962e-05,
+      "loss": 0.3092,
+      "step": 23170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7715375055883804e-05,
+      "loss": 0.2792,
+      "step": 23175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.771418518342381e-05,
+      "loss": 0.2955,
+      "step": 23180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.771299501603509e-05,
+      "loss": 0.3136,
+      "step": 23185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7711804553733095e-05,
+      "loss": 0.3088,
+      "step": 23190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.771061379653329e-05,
+      "loss": 0.2912,
+      "step": 23195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7709422744451126e-05,
+      "loss": 0.3338,
+      "step": 23200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7708231397502076e-05,
+      "loss": 0.2899,
+      "step": 23205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770703975570161e-05,
+      "loss": 0.3269,
+      "step": 23210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7705847819065195e-05,
+      "loss": 0.3311,
+      "step": 23215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770465558760831e-05,
+      "loss": 0.3322,
+      "step": 23220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770346306134643e-05,
+      "loss": 0.3201,
+      "step": 23225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.770227024029506e-05,
+      "loss": 0.2873,
+      "step": 23230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7701077124469655e-05,
+      "loss": 0.2996,
+      "step": 23235
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769988371388573e-05,
+      "loss": 0.3076,
+      "step": 23240
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769869000855878e-05,
+      "loss": 0.3316,
+      "step": 23245
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7697496008504297e-05,
+      "loss": 0.3065,
+      "step": 23250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769630171373778e-05,
+      "loss": 0.2948,
+      "step": 23255
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769510712427475e-05,
+      "loss": 0.3184,
+      "step": 23260
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7693912240130703e-05,
+      "loss": 0.2942,
+      "step": 23265
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7692717061321165e-05,
+      "loss": 0.3225,
+      "step": 23270
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7691521587861655e-05,
+      "loss": 0.328,
+      "step": 23275
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.769032581976769e-05,
+      "loss": 0.3131,
+      "step": 23280
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7689129757054795e-05,
+      "loss": 0.3128,
+      "step": 23285
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.76879333997385e-05,
+      "loss": 0.2992,
+      "step": 23290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768673674783435e-05,
+      "loss": 0.3081,
+      "step": 23295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768553980135787e-05,
+      "loss": 0.2858,
+      "step": 23300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768434256032461e-05,
+      "loss": 0.332,
+      "step": 23305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768314502475011e-05,
+      "loss": 0.3253,
+      "step": 23310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768194719464991e-05,
+      "loss": 0.3177,
+      "step": 23315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.768074907003959e-05,
+      "loss": 0.314,
+      "step": 23320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7679550650934676e-05,
+      "loss": 0.2749,
+      "step": 23325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767835193735075e-05,
+      "loss": 0.2995,
+      "step": 23330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767715292930337e-05,
+      "loss": 0.3238,
+      "step": 23335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.76759536268081e-05,
+      "loss": 0.312,
+      "step": 23340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7674754029880524e-05,
+      "loss": 0.2929,
+      "step": 23345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.76735541385362e-05,
+      "loss": 0.3103,
+      "step": 23350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767235395279073e-05,
+      "loss": 0.3366,
+      "step": 23355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.767115347265968e-05,
+      "loss": 0.3116,
+      "step": 23360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766995269815864e-05,
+      "loss": 0.3052,
+      "step": 23365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766875162930321e-05,
+      "loss": 0.3002,
+      "step": 23370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7667550266108974e-05,
+      "loss": 0.3176,
+      "step": 23375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766634860859154e-05,
+      "loss": 0.2901,
+      "step": 23380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.76651466567665e-05,
+      "loss": 0.2979,
+      "step": 23385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766394441064948e-05,
+      "loss": 0.3301,
+      "step": 23390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766274187025607e-05,
+      "loss": 0.3071,
+      "step": 23395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766153903560189e-05,
+      "loss": 0.3096,
+      "step": 23400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.766033590670256e-05,
+      "loss": 0.313,
+      "step": 23405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765913248357371e-05,
+      "loss": 0.3152,
+      "step": 23410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7657928766230944e-05,
+      "loss": 0.2958,
+      "step": 23415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7656724754689916e-05,
+      "loss": 0.3091,
+      "step": 23420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765552044896625e-05,
+      "loss": 0.3127,
+      "step": 23425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7654315849075574e-05,
+      "loss": 0.3271,
+      "step": 23430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.7653110955033545e-05,
+      "loss": 0.2826,
+      "step": 23435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765190576685579e-05,
+      "loss": 0.3016,
+      "step": 23440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.765070028455797e-05,
+      "loss": 0.308,
+      "step": 23445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7649494508155735e-05,
+      "loss": 0.2744,
+      "step": 23450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7648288437664746e-05,
+      "loss": 0.3216,
+      "step": 23455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764708207310065e-05,
+      "loss": 0.2988,
+      "step": 23460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7645875414479115e-05,
+      "loss": 0.3203,
+      "step": 23465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764466846181582e-05,
+      "loss": 0.3257,
+      "step": 23470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7643461215126425e-05,
+      "loss": 0.3117,
+      "step": 23475
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764225367442661e-05,
+      "loss": 0.3221,
+      "step": 23480
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.764104583973205e-05,
+      "loss": 0.3148,
+      "step": 23485
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7639837711058435e-05,
+      "loss": 0.3044,
+      "step": 23490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763862928842144e-05,
+      "loss": 0.3233,
+      "step": 23495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7637420571836775e-05,
+      "loss": 0.3088,
+      "step": 23500
+    },
+    {
+      "epoch": 0.83,
+      "eval_loss": 0.29781535267829895,
+      "eval_runtime": 10.5453,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 23500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763621156132011e-05,
+      "loss": 0.2941,
+      "step": 23505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763500225688716e-05,
+      "loss": 0.3113,
+      "step": 23510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763379265855363e-05,
+      "loss": 0.3271,
+      "step": 23515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.763258276633521e-05,
+      "loss": 0.3066,
+      "step": 23520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7631372580247624e-05,
+      "loss": 0.2862,
+      "step": 23525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7630162100306576e-05,
+      "loss": 0.3283,
+      "step": 23530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7628951326527786e-05,
+      "loss": 0.3109,
+      "step": 23535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762774025892698e-05,
+      "loss": 0.2908,
+      "step": 23540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762652889751988e-05,
+      "loss": 0.3238,
+      "step": 23545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762531724232221e-05,
+      "loss": 0.3091,
+      "step": 23550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7624105293349706e-05,
+      "loss": 0.3094,
+      "step": 23555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762289305061811e-05,
+      "loss": 0.316,
+      "step": 23560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.762168051414315e-05,
+      "loss": 0.327,
+      "step": 23565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7620467683940574e-05,
+      "loss": 0.2885,
+      "step": 23570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.761925456002614e-05,
+      "loss": 0.3026,
+      "step": 23575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7618041142415596e-05,
+      "loss": 0.3034,
+      "step": 23580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7616827431124674e-05,
+      "loss": 0.2888,
+      "step": 23585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.761561342616917e-05,
+      "loss": 0.3184,
+      "step": 23590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.761439912756483e-05,
+      "loss": 0.314,
+      "step": 23595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7613184535327414e-05,
+      "loss": 0.3091,
+      "step": 23600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.76119696494727e-05,
+      "loss": 0.2994,
+      "step": 23605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.761075447001647e-05,
+      "loss": 0.3185,
+      "step": 23610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760953899697449e-05,
+      "loss": 0.3372,
+      "step": 23615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760832323036254e-05,
+      "loss": 0.3087,
+      "step": 23620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7607107170196417e-05,
+      "loss": 0.2997,
+      "step": 23625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7605890816491904e-05,
+      "loss": 0.2975,
+      "step": 23630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.760467416926481e-05,
+      "loss": 0.3262,
+      "step": 23635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.76034572285309e-05,
+      "loss": 0.3111,
+      "step": 23640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7602239994306e-05,
+      "loss": 0.2784,
+      "step": 23645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7601022466605914e-05,
+      "loss": 0.2921,
+      "step": 23650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759980464544645e-05,
+      "loss": 0.2982,
+      "step": 23655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7598586530843405e-05,
+      "loss": 0.294,
+      "step": 23660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7597368122812607e-05,
+      "loss": 0.3147,
+      "step": 23665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7596149421369886e-05,
+      "loss": 0.3126,
+      "step": 23670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7594930426531046e-05,
+      "loss": 0.3015,
+      "step": 23675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759371113831193e-05,
+      "loss": 0.3026,
+      "step": 23680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759249155672837e-05,
+      "loss": 0.3211,
+      "step": 23685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759127168179619e-05,
+      "loss": 0.2899,
+      "step": 23690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.759005151353123e-05,
+      "loss": 0.3028,
+      "step": 23695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758883105194934e-05,
+      "loss": 0.3102,
+      "step": 23700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758761029706637e-05,
+      "loss": 0.2816,
+      "step": 23705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758638924889817e-05,
+      "loss": 0.2997,
+      "step": 23710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758516790746059e-05,
+      "loss": 0.3077,
+      "step": 23715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.758394627276949e-05,
+      "loss": 0.3017,
+      "step": 23720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7582724344840726e-05,
+      "loss": 0.2612,
+      "step": 23725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.7581502123690165e-05,
+      "loss": 0.3053,
+      "step": 23730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.758027960933368e-05,
+      "loss": 0.3,
+      "step": 23735
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757905680178715e-05,
+      "loss": 0.323,
+      "step": 23740
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757783370106646e-05,
+      "loss": 0.3073,
+      "step": 23745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757661030718746e-05,
+      "loss": 0.3241,
+      "step": 23750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757538662016606e-05,
+      "loss": 0.2923,
+      "step": 23755
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.757416264001814e-05,
+      "loss": 0.3052,
+      "step": 23760
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.75729383667596e-05,
+      "loss": 0.2748,
+      "step": 23765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7571713800406324e-05,
+      "loss": 0.3155,
+      "step": 23770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7570488940974236e-05,
+      "loss": 0.3109,
+      "step": 23775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7569263788479204e-05,
+      "loss": 0.2879,
+      "step": 23780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756803834293717e-05,
+      "loss": 0.3166,
+      "step": 23785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756681260436402e-05,
+      "loss": 0.3123,
+      "step": 23790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7565586572775686e-05,
+      "loss": 0.2884,
+      "step": 23795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7564360248188086e-05,
+      "loss": 0.308,
+      "step": 23800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756313363061713e-05,
+      "loss": 0.3314,
+      "step": 23805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756190672007876e-05,
+      "loss": 0.3089,
+      "step": 23810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.756067951658889e-05,
+      "loss": 0.3132,
+      "step": 23815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755945202016348e-05,
+      "loss": 0.2933,
+      "step": 23820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755822423081844e-05,
+      "loss": 0.2671,
+      "step": 23825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755699614856973e-05,
+      "loss": 0.3004,
+      "step": 23830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755576777343329e-05,
+      "loss": 0.2954,
+      "step": 23835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7554539105425075e-05,
+      "loss": 0.3193,
+      "step": 23840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7553310144561024e-05,
+      "loss": 0.3082,
+      "step": 23845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.755208089085711e-05,
+      "loss": 0.2875,
+      "step": 23850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7550851344329286e-05,
+      "loss": 0.3087,
+      "step": 23855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.754962150499352e-05,
+      "loss": 0.3027,
+      "step": 23860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7548391372865776e-05,
+      "loss": 0.3188,
+      "step": 23865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7547160947962034e-05,
+      "loss": 0.3187,
+      "step": 23870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.754593023029827e-05,
+      "loss": 0.3014,
+      "step": 23875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7544699219890454e-05,
+      "loss": 0.2919,
+      "step": 23880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.754346791675458e-05,
+      "loss": 0.3071,
+      "step": 23885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.754223632090663e-05,
+      "loss": 0.3336,
+      "step": 23890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7541004432362596e-05,
+      "loss": 0.3384,
+      "step": 23895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753977225113847e-05,
+      "loss": 0.305,
+      "step": 23900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7538539777250266e-05,
+      "loss": 0.3004,
+      "step": 23905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753730701071397e-05,
+      "loss": 0.3281,
+      "step": 23910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.75360739515456e-05,
+      "loss": 0.2932,
+      "step": 23915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7534840599761154e-05,
+      "loss": 0.3116,
+      "step": 23920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7533606955376665e-05,
+      "loss": 0.3204,
+      "step": 23925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7532373018408125e-05,
+      "loss": 0.3119,
+      "step": 23930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.753113878887158e-05,
+      "loss": 0.2632,
+      "step": 23935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752990426678304e-05,
+      "loss": 0.3219,
+      "step": 23940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7528669452158554e-05,
+      "loss": 0.3126,
+      "step": 23945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752743434501413e-05,
+      "loss": 0.3073,
+      "step": 23950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752619894536582e-05,
+      "loss": 0.2941,
+      "step": 23955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752496325322967e-05,
+      "loss": 0.3093,
+      "step": 23960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752372726862171e-05,
+      "loss": 0.3013,
+      "step": 23965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752249099155799e-05,
+      "loss": 0.2963,
+      "step": 23970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752125442205457e-05,
+      "loss": 0.3125,
+      "step": 23975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752001756012751e-05,
+      "loss": 0.3002,
+      "step": 23980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.751878040579286e-05,
+      "loss": 0.3147,
+      "step": 23985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7517542959066684e-05,
+      "loss": 0.3145,
+      "step": 23990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.751630521996505e-05,
+      "loss": 0.2935,
+      "step": 23995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.751506718850404e-05,
+      "loss": 0.3306,
+      "step": 24000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.2971259653568268,
+      "eval_runtime": 10.5298,
+      "eval_samples_per_second": 9.497,
+      "eval_steps_per_second": 9.497,
+      "step": 24000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.751382886469972e-05,
+      "loss": 0.3354,
+      "step": 24005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7512590248568163e-05,
+      "loss": 0.2793,
+      "step": 24010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7511351340125456e-05,
+      "loss": 0.3002,
+      "step": 24015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.75101121393877e-05,
+      "loss": 0.2878,
+      "step": 24020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750887264637096e-05,
+      "loss": 0.2873,
+      "step": 24025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750763286109134e-05,
+      "loss": 0.3029,
+      "step": 24030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750639278356496e-05,
+      "loss": 0.317,
+      "step": 24035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7505152413807886e-05,
+      "loss": 0.2921,
+      "step": 24040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750391175183624e-05,
+      "loss": 0.3101,
+      "step": 24045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750267079766614e-05,
+      "loss": 0.315,
+      "step": 24050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.750142955131368e-05,
+      "loss": 0.3072,
+      "step": 24055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7500188012794985e-05,
+      "loss": 0.3223,
+      "step": 24060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749894618212618e-05,
+      "loss": 0.2978,
+      "step": 24065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749770405932339e-05,
+      "loss": 0.2927,
+      "step": 24070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749646164440273e-05,
+      "loss": 0.2951,
+      "step": 24075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7495218937380356e-05,
+      "loss": 0.3082,
+      "step": 24080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7493975938272376e-05,
+      "loss": 0.3136,
+      "step": 24085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749273264709495e-05,
+      "loss": 0.2949,
+      "step": 24090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749148906386421e-05,
+      "loss": 0.2751,
+      "step": 24095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.749024518859631e-05,
+      "loss": 0.3013,
+      "step": 24100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74890010213074e-05,
+      "loss": 0.3143,
+      "step": 24105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748775656201363e-05,
+      "loss": 0.3332,
+      "step": 24110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748651181073115e-05,
+      "loss": 0.2902,
+      "step": 24115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748526676747614e-05,
+      "loss": 0.2741,
+      "step": 24120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748402143226477e-05,
+      "loss": 0.3071,
+      "step": 24125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748277580511319e-05,
+      "loss": 0.3265,
+      "step": 24130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748152988603758e-05,
+      "loss": 0.3057,
+      "step": 24135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.748028367505413e-05,
+      "loss": 0.3049,
+      "step": 24140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.747903717217901e-05,
+      "loss": 0.3015,
+      "step": 24145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74777903774284e-05,
+      "loss": 0.3069,
+      "step": 24150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74765432908185e-05,
+      "loss": 0.3236,
+      "step": 24155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.747529591236549e-05,
+      "loss": 0.3162,
+      "step": 24160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7474048242085585e-05,
+      "loss": 0.2927,
+      "step": 24165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7472800279994964e-05,
+      "loss": 0.3169,
+      "step": 24170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7471552026109855e-05,
+      "loss": 0.3284,
+      "step": 24175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7470303480446446e-05,
+      "loss": 0.3145,
+      "step": 24180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746905464302095e-05,
+      "loss": 0.2979,
+      "step": 24185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74678055138496e-05,
+      "loss": 0.308,
+      "step": 24190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746655609294858e-05,
+      "loss": 0.3221,
+      "step": 24195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746530638033415e-05,
+      "loss": 0.3295,
+      "step": 24200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746405637602252e-05,
+      "loss": 0.3207,
+      "step": 24205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746280608002992e-05,
+      "loss": 0.3052,
+      "step": 24210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746155549237259e-05,
+      "loss": 0.2973,
+      "step": 24215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.746030461306676e-05,
+      "loss": 0.2646,
+      "step": 24220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.745905344212869e-05,
+      "loss": 0.3076,
+      "step": 24225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7457801979574604e-05,
+      "loss": 0.2814,
+      "step": 24230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.745655022542075e-05,
+      "loss": 0.3061,
+      "step": 24235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74552981796834e-05,
+      "loss": 0.3201,
+      "step": 24240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7454045842378806e-05,
+      "loss": 0.2877,
+      "step": 24245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.745279321352321e-05,
+      "loss": 0.2999,
+      "step": 24250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.74515402931329e-05,
+      "loss": 0.292,
+      "step": 24255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.745028708122414e-05,
+      "loss": 0.2876,
+      "step": 24260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744903357781319e-05,
+      "loss": 0.3125,
+      "step": 24265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7447779782916333e-05,
+      "loss": 0.3157,
+      "step": 24270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7446525696549855e-05,
+      "loss": 0.3172,
+      "step": 24275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744527131873003e-05,
+      "loss": 0.2945,
+      "step": 24280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744401664947315e-05,
+      "loss": 0.2963,
+      "step": 24285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744276168879551e-05,
+      "loss": 0.2887,
+      "step": 24290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.744150643671339e-05,
+      "loss": 0.2818,
+      "step": 24295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.7440250893243104e-05,
+      "loss": 0.3011,
+      "step": 24300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743899505840095e-05,
+      "loss": 0.3359,
+      "step": 24305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743773893220323e-05,
+      "loss": 0.2925,
+      "step": 24310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743648251466626e-05,
+      "loss": 0.2962,
+      "step": 24315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743522580580634e-05,
+      "loss": 0.3224,
+      "step": 24320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743396880563981e-05,
+      "loss": 0.3107,
+      "step": 24325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7432711514182976e-05,
+      "loss": 0.3132,
+      "step": 24330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743145393145217e-05,
+      "loss": 0.3075,
+      "step": 24335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.743019605746372e-05,
+      "loss": 0.2577,
+      "step": 24340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742893789223394e-05,
+      "loss": 0.3243,
+      "step": 24345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7427679435779196e-05,
+      "loss": 0.2945,
+      "step": 24350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7426420688115816e-05,
+      "loss": 0.3102,
+      "step": 24355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742516164926014e-05,
+      "loss": 0.3217,
+      "step": 24360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742390231922852e-05,
+      "loss": 0.2823,
+      "step": 24365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.74226426980373e-05,
+      "loss": 0.3064,
+      "step": 24370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.742138278570285e-05,
+      "loss": 0.3351,
+      "step": 24375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7420122582241514e-05,
+      "loss": 0.341,
+      "step": 24380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.741886208766966e-05,
+      "loss": 0.3188,
+      "step": 24385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.741760130200367e-05,
+      "loss": 0.3149,
+      "step": 24390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7416340225259886e-05,
+      "loss": 0.3037,
+      "step": 24395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7415078857454704e-05,
+      "loss": 0.3078,
+      "step": 24400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7413817198604496e-05,
+      "loss": 0.287,
+      "step": 24405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7412555248725646e-05,
+      "loss": 0.295,
+      "step": 24410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7411293007834525e-05,
+      "loss": 0.3377,
+      "step": 24415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.741003047594754e-05,
+      "loss": 0.2956,
+      "step": 24420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740876765308107e-05,
+      "loss": 0.3259,
+      "step": 24425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740750453925153e-05,
+      "loss": 0.2787,
+      "step": 24430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7406241134475316e-05,
+      "loss": 0.3286,
+      "step": 24435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740497743876882e-05,
+      "loss": 0.3364,
+      "step": 24440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7403713452148454e-05,
+      "loss": 0.2879,
+      "step": 24445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740244917463064e-05,
+      "loss": 0.3261,
+      "step": 24450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7401184606231786e-05,
+      "loss": 0.2984,
+      "step": 24455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7399919746968316e-05,
+      "loss": 0.3239,
+      "step": 24460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.739865459685664e-05,
+      "loss": 0.2893,
+      "step": 24465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7397389155913197e-05,
+      "loss": 0.3265,
+      "step": 24470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7396123424154426e-05,
+      "loss": 0.2947,
+      "step": 24475
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.739485740159674e-05,
+      "loss": 0.2713,
+      "step": 24480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.739359108825659e-05,
+      "loss": 0.318,
+      "step": 24485
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.739232448415043e-05,
+      "loss": 0.3234,
+      "step": 24490
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.739105758929468e-05,
+      "loss": 0.3022,
+      "step": 24495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738979040370582e-05,
+      "loss": 0.3255,
+      "step": 24500
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.297638863325119,
+      "eval_runtime": 10.5374,
+      "eval_samples_per_second": 9.49,
+      "eval_steps_per_second": 9.49,
+      "step": 24500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7388522927400276e-05,
+      "loss": 0.2881,
+      "step": 24505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7387255160394514e-05,
+      "loss": 0.32,
+      "step": 24510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7385987102705e-05,
+      "loss": 0.3053,
+      "step": 24515
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.73847187543482e-05,
+      "loss": 0.294,
+      "step": 24520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738345011534058e-05,
+      "loss": 0.2906,
+      "step": 24525
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738218118569861e-05,
+      "loss": 0.2912,
+      "step": 24530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.738091196543876e-05,
+      "loss": 0.2966,
+      "step": 24535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737964245457752e-05,
+      "loss": 0.3212,
+      "step": 24540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7378372653131375e-05,
+      "loss": 0.3061,
+      "step": 24545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7377102561116816e-05,
+      "loss": 0.3129,
+      "step": 24550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737583217855032e-05,
+      "loss": 0.3163,
+      "step": 24555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7374561505448384e-05,
+      "loss": 0.3168,
+      "step": 24560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737329054182752e-05,
+      "loss": 0.3179,
+      "step": 24565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737201928770422e-05,
+      "loss": 0.2977,
+      "step": 24570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.737074774309499e-05,
+      "loss": 0.2845,
+      "step": 24575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.736947590801635e-05,
+      "loss": 0.2973,
+      "step": 24580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.73682037824848e-05,
+      "loss": 0.3149,
+      "step": 24585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7366931366516876e-05,
+      "loss": 0.3216,
+      "step": 24590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736565866012908e-05,
+      "loss": 0.3094,
+      "step": 24595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736438566333795e-05,
+      "loss": 0.2993,
+      "step": 24600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736311237616001e-05,
+      "loss": 0.3066,
+      "step": 24605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736183879861179e-05,
+      "loss": 0.2802,
+      "step": 24610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.736056493070983e-05,
+      "loss": 0.2907,
+      "step": 24615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7359290772470675e-05,
+      "loss": 0.312,
+      "step": 24620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735801632391087e-05,
+      "loss": 0.2899,
+      "step": 24625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735674158504695e-05,
+      "loss": 0.3086,
+      "step": 24630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735546655589547e-05,
+      "loss": 0.2949,
+      "step": 24635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7354191236472995e-05,
+      "loss": 0.3053,
+      "step": 24640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735291562679608e-05,
+      "loss": 0.3149,
+      "step": 24645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.735163972688128e-05,
+      "loss": 0.2791,
+      "step": 24650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7350363536745174e-05,
+      "loss": 0.2994,
+      "step": 24655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734908705640433e-05,
+      "loss": 0.3109,
+      "step": 24660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7347810285875315e-05,
+      "loss": 0.3036,
+      "step": 24665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7346533225174705e-05,
+      "loss": 0.2824,
+      "step": 24670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7345255874319096e-05,
+      "loss": 0.3066,
+      "step": 24675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7343978233325064e-05,
+      "loss": 0.3002,
+      "step": 24680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7342700302209195e-05,
+      "loss": 0.3059,
+      "step": 24685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7341422080988094e-05,
+      "loss": 0.2977,
+      "step": 24690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734014356967835e-05,
+      "loss": 0.3186,
+      "step": 24695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.733886476829656e-05,
+      "loss": 0.2897,
+      "step": 24700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7337585676859344e-05,
+      "loss": 0.2943,
+      "step": 24705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7336306295383285e-05,
+      "loss": 0.3145,
+      "step": 24710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7335026623885015e-05,
+      "loss": 0.3302,
+      "step": 24715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.733374666238115e-05,
+      "loss": 0.2812,
+      "step": 24720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.733246641088829e-05,
+      "loss": 0.2879,
+      "step": 24725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7331185869423075e-05,
+      "loss": 0.3062,
+      "step": 24730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7329905038002136e-05,
+      "loss": 0.285,
+      "step": 24735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732862391664209e-05,
+      "loss": 0.3035,
+      "step": 24740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732734250535959e-05,
+      "loss": 0.2817,
+      "step": 24745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732606080417124e-05,
+      "loss": 0.3166,
+      "step": 24750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732477881309372e-05,
+      "loss": 0.3061,
+      "step": 24755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732349653214365e-05,
+      "loss": 0.3076,
+      "step": 24760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.732221396133769e-05,
+      "loss": 0.3195,
+      "step": 24765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.73209311006925e-05,
+      "loss": 0.3247,
+      "step": 24770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731964795022473e-05,
+      "loss": 0.3167,
+      "step": 24775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731836450995103e-05,
+      "loss": 0.3,
+      "step": 24780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7317080779888076e-05,
+      "loss": 0.2758,
+      "step": 24785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7315796760052544e-05,
+      "loss": 0.3273,
+      "step": 24790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.73145124504611e-05,
+      "loss": 0.2699,
+      "step": 24795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731322785113041e-05,
+      "loss": 0.285,
+      "step": 24800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7311942962077164e-05,
+      "loss": 0.3223,
+      "step": 24805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.731065778331805e-05,
+      "loss": 0.3058,
+      "step": 24810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730937231486974e-05,
+      "loss": 0.2904,
+      "step": 24815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7308086556748926e-05,
+      "loss": 0.2698,
+      "step": 24820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730680050897231e-05,
+      "loss": 0.3074,
+      "step": 24825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730551417155661e-05,
+      "loss": 0.2964,
+      "step": 24830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730422754451849e-05,
+      "loss": 0.2897,
+      "step": 24835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730294062787468e-05,
+      "loss": 0.3129,
+      "step": 24840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730165342164188e-05,
+      "loss": 0.3175,
+      "step": 24845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730036592583682e-05,
+      "loss": 0.2899,
+      "step": 24850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7299078140476195e-05,
+      "loss": 0.3284,
+      "step": 24855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7297790065576734e-05,
+      "loss": 0.3055,
+      "step": 24860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.729650170115517e-05,
+      "loss": 0.3146,
+      "step": 24865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.729521304722823e-05,
+      "loss": 0.307,
+      "step": 24870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7293924103812626e-05,
+      "loss": 0.2946,
+      "step": 24875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7292634870925126e-05,
+      "loss": 0.3226,
+      "step": 24880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.729134534858244e-05,
+      "loss": 0.3063,
+      "step": 24885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.729005553680133e-05,
+      "loss": 0.315,
+      "step": 24890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7288765435598536e-05,
+      "loss": 0.3197,
+      "step": 24895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7287475044990815e-05,
+      "loss": 0.306,
+      "step": 24900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7286184364994924e-05,
+      "loss": 0.3047,
+      "step": 24905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.728489339562761e-05,
+      "loss": 0.3094,
+      "step": 24910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.728360213690564e-05,
+      "loss": 0.3052,
+      "step": 24915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7282310588845786e-05,
+      "loss": 0.2992,
+      "step": 24920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7281018751464814e-05,
+      "loss": 0.3046,
+      "step": 24925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.727972662477949e-05,
+      "loss": 0.2983,
+      "step": 24930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.727843420880661e-05,
+      "loss": 0.318,
+      "step": 24935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.727714150356294e-05,
+      "loss": 0.2807,
+      "step": 24940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.727584850906526e-05,
+      "loss": 0.3074,
+      "step": 24945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7274555225330374e-05,
+      "loss": 0.2902,
+      "step": 24950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7273261652375066e-05,
+      "loss": 0.2953,
+      "step": 24955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.727196779021614e-05,
+      "loss": 0.3285,
+      "step": 24960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7270673638870386e-05,
+      "loss": 0.3097,
+      "step": 24965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7269379198354606e-05,
+      "loss": 0.3233,
+      "step": 24970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.726808446868562e-05,
+      "loss": 0.3141,
+      "step": 24975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7266789449880235e-05,
+      "loss": 0.2786,
+      "step": 24980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7265494141955255e-05,
+      "loss": 0.2944,
+      "step": 24985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7264198544927515e-05,
+      "loss": 0.313,
+      "step": 24990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.726290265881382e-05,
+      "loss": 0.2895,
+      "step": 24995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7261606483631017e-05,
+      "loss": 0.3045,
+      "step": 25000
+    },
+    {
+      "epoch": 0.88,
+      "eval_loss": 0.2971786856651306,
+      "eval_runtime": 10.5292,
+      "eval_samples_per_second": 9.497,
+      "eval_steps_per_second": 9.497,
+      "step": 25000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.726031001939592e-05,
+      "loss": 0.3188,
+      "step": 25005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725901326612536e-05,
+      "loss": 0.3274,
+      "step": 25010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.72577162238362e-05,
+      "loss": 0.3033,
+      "step": 25015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725641889254525e-05,
+      "loss": 0.3326,
+      "step": 25020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725512127226937e-05,
+      "loss": 0.3117,
+      "step": 25025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725382336302541e-05,
+      "loss": 0.3072,
+      "step": 25030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725252516483021e-05,
+      "loss": 0.2899,
+      "step": 25035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.725122667770064e-05,
+      "loss": 0.305,
+      "step": 25040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7249927901653566e-05,
+      "loss": 0.2833,
+      "step": 25045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724862883670583e-05,
+      "loss": 0.3178,
+      "step": 25050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724732948287432e-05,
+      "loss": 0.2996,
+      "step": 25055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7246029840175884e-05,
+      "loss": 0.2997,
+      "step": 25060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7244729908627426e-05,
+      "loss": 0.3063,
+      "step": 25065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.72434296882458e-05,
+      "loss": 0.2872,
+      "step": 25070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7242129179047904e-05,
+      "loss": 0.2907,
+      "step": 25075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.724082838105062e-05,
+      "loss": 0.3003,
+      "step": 25080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723952729427083e-05,
+      "loss": 0.3175,
+      "step": 25085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723822591872544e-05,
+      "loss": 0.3222,
+      "step": 25090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7236924254431344e-05,
+      "loss": 0.3183,
+      "step": 25095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723562230140544e-05,
+      "loss": 0.2934,
+      "step": 25100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7234320059664626e-05,
+      "loss": 0.3129,
+      "step": 25105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7233017529225826e-05,
+      "loss": 0.2844,
+      "step": 25110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.723171471010595e-05,
+      "loss": 0.2877,
+      "step": 25115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.72304116023219e-05,
+      "loss": 0.2901,
+      "step": 25120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.72291082058906e-05,
+      "loss": 0.2915,
+      "step": 25125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7227804520829e-05,
+      "loss": 0.3152,
+      "step": 25130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7226500547153996e-05,
+      "loss": 0.301,
+      "step": 25135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.722519628488253e-05,
+      "loss": 0.3223,
+      "step": 25140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.722389173403153e-05,
+      "loss": 0.3069,
+      "step": 25145
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7222586894617956e-05,
+      "loss": 0.3253,
+      "step": 25150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.722128176665873e-05,
+      "loss": 0.2785,
+      "step": 25155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.72199763501708e-05,
+      "loss": 0.3071,
+      "step": 25160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7218670645171125e-05,
+      "loss": 0.3289,
+      "step": 25165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721736465167666e-05,
+      "loss": 0.3063,
+      "step": 25170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721605836970435e-05,
+      "loss": 0.3212,
+      "step": 25175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7214751799271165e-05,
+      "loss": 0.2908,
+      "step": 25180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721344494039407e-05,
+      "loss": 0.3022,
+      "step": 25185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.721213779309003e-05,
+      "loss": 0.3016,
+      "step": 25190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7210830357376015e-05,
+      "loss": 0.3272,
+      "step": 25195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720952263326901e-05,
+      "loss": 0.3067,
+      "step": 25200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7208214620785986e-05,
+      "loss": 0.3113,
+      "step": 25205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7206906319943934e-05,
+      "loss": 0.277,
+      "step": 25210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720559773075984e-05,
+      "loss": 0.3084,
+      "step": 25215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720428885325069e-05,
+      "loss": 0.2904,
+      "step": 25220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720297968743348e-05,
+      "loss": 0.3018,
+      "step": 25225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.720167023332522e-05,
+      "loss": 0.3067,
+      "step": 25230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.72003604909429e-05,
+      "loss": 0.3084,
+      "step": 25235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7199050460303525e-05,
+      "loss": 0.2959,
+      "step": 25240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719774014142411e-05,
+      "loss": 0.2769,
+      "step": 25245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719642953432167e-05,
+      "loss": 0.2959,
+      "step": 25250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719511863901322e-05,
+      "loss": 0.2857,
+      "step": 25255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719380745551578e-05,
+      "loss": 0.2813,
+      "step": 25260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7192495983846385e-05,
+      "loss": 0.3073,
+      "step": 25265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.719118422402205e-05,
+      "loss": 0.2863,
+      "step": 25270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718987217605981e-05,
+      "loss": 0.3157,
+      "step": 25275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.71885598399767e-05,
+      "loss": 0.2739,
+      "step": 25280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718724721578978e-05,
+      "loss": 0.2985,
+      "step": 25285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7185934303516054e-05,
+      "loss": 0.2974,
+      "step": 25290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718462110317261e-05,
+      "loss": 0.3091,
+      "step": 25295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718330761477647e-05,
+      "loss": 0.2901,
+      "step": 25300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7181993838344704e-05,
+      "loss": 0.2989,
+      "step": 25305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.718067977389437e-05,
+      "loss": 0.2995,
+      "step": 25310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7179365421442526e-05,
+      "loss": 0.3258,
+      "step": 25315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717805078100623e-05,
+      "loss": 0.3229,
+      "step": 25320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717673585260257e-05,
+      "loss": 0.2945,
+      "step": 25325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717542063624861e-05,
+      "loss": 0.3246,
+      "step": 25330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7174105131961424e-05,
+      "loss": 0.2893,
+      "step": 25335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7172789339758095e-05,
+      "loss": 0.3087,
+      "step": 25340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717147325965571e-05,
+      "loss": 0.2858,
+      "step": 25345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.717015689167136e-05,
+      "loss": 0.2646,
+      "step": 25350
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716884023582213e-05,
+      "loss": 0.3022,
+      "step": 25355
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716752329212512e-05,
+      "loss": 0.3013,
+      "step": 25360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716620606059742e-05,
+      "loss": 0.3055,
+      "step": 25365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7164888541256154e-05,
+      "loss": 0.289,
+      "step": 25370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.716357073411841e-05,
+      "loss": 0.3143,
+      "step": 25375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7162252639201314e-05,
+      "loss": 0.3089,
+      "step": 25380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7160934256521974e-05,
+      "loss": 0.2892,
+      "step": 25385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.71596155860975e-05,
+      "loss": 0.3338,
+      "step": 25390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715829662794502e-05,
+      "loss": 0.291,
+      "step": 25395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715697738208167e-05,
+      "loss": 0.2916,
+      "step": 25400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715565784852457e-05,
+      "loss": 0.2713,
+      "step": 25405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715433802729085e-05,
+      "loss": 0.3044,
+      "step": 25410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715301791839765e-05,
+      "loss": 0.3076,
+      "step": 25415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.7151697521862104e-05,
+      "loss": 0.3235,
+      "step": 25420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.715037683770138e-05,
+      "loss": 0.362,
+      "step": 25425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.71490558659326e-05,
+      "loss": 0.2861,
+      "step": 25430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.714773460657293e-05,
+      "loss": 0.2833,
+      "step": 25435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7146413059639524e-05,
+      "loss": 0.3223,
+      "step": 25440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.714509122514953e-05,
+      "loss": 0.2809,
+      "step": 25445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7143769103120125e-05,
+      "loss": 0.3196,
+      "step": 25450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.714244669356848e-05,
+      "loss": 0.2941,
+      "step": 25455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.714112399651175e-05,
+      "loss": 0.3171,
+      "step": 25460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713980101196712e-05,
+      "loss": 0.2729,
+      "step": 25465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713847773995176e-05,
+      "loss": 0.2862,
+      "step": 25470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7137154180482854e-05,
+      "loss": 0.2835,
+      "step": 25475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.71358303335776e-05,
+      "loss": 0.31,
+      "step": 25480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713450619925317e-05,
+      "loss": 0.2939,
+      "step": 25485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713318177752677e-05,
+      "loss": 0.2882,
+      "step": 25490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.713185706841559e-05,
+      "loss": 0.3074,
+      "step": 25495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7130532071936814e-05,
+      "loss": 0.3055,
+      "step": 25500
+    },
+    {
+      "epoch": 0.9,
+      "eval_loss": 0.29553505778312683,
+      "eval_runtime": 10.5452,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 25500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712920678810768e-05,
+      "loss": 0.2973,
+      "step": 25505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712788121694538e-05,
+      "loss": 0.2921,
+      "step": 25510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712655535846713e-05,
+      "loss": 0.2836,
+      "step": 25515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712522921269012e-05,
+      "loss": 0.313,
+      "step": 25520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712390277963161e-05,
+      "loss": 0.3145,
+      "step": 25525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7122576059308796e-05,
+      "loss": 0.2959,
+      "step": 25530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.712124905173891e-05,
+      "loss": 0.3051,
+      "step": 25535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711992175693918e-05,
+      "loss": 0.3089,
+      "step": 25540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7118594174926856e-05,
+      "loss": 0.325,
+      "step": 25545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711726630571915e-05,
+      "loss": 0.3128,
+      "step": 25550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711593814933333e-05,
+      "loss": 0.3181,
+      "step": 25555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7114609705786614e-05,
+      "loss": 0.2796,
+      "step": 25560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711328097509627e-05,
+      "loss": 0.3029,
+      "step": 25565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.711195195727955e-05,
+      "loss": 0.2966,
+      "step": 25570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7110622652353696e-05,
+      "loss": 0.3051,
+      "step": 25575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7109293060335994e-05,
+      "loss": 0.3295,
+      "step": 25580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710796318124368e-05,
+      "loss": 0.2901,
+      "step": 25585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7106633015094037e-05,
+      "loss": 0.3316,
+      "step": 25590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710530256190433e-05,
+      "loss": 0.2916,
+      "step": 25595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7103971821691845e-05,
+      "loss": 0.2884,
+      "step": 25600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710264079447384e-05,
+      "loss": 0.2916,
+      "step": 25605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710130948026763e-05,
+      "loss": 0.2877,
+      "step": 25610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7099977879090464e-05,
+      "loss": 0.3071,
+      "step": 25615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7098645990959654e-05,
+      "loss": 0.3166,
+      "step": 25620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7097313815892495e-05,
+      "loss": 0.3184,
+      "step": 25625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7095981353906275e-05,
+      "loss": 0.2892,
+      "step": 25630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.70946486050183e-05,
+      "loss": 0.2976,
+      "step": 25635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7093315569245876e-05,
+      "loss": 0.3174,
+      "step": 25640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7091982246606304e-05,
+      "loss": 0.3334,
+      "step": 25645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.709064863711691e-05,
+      "loss": 0.3152,
+      "step": 25650
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708931474079499e-05,
+      "loss": 0.2769,
+      "step": 25655
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708798055765788e-05,
+      "loss": 0.3005,
+      "step": 25660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.70866460877229e-05,
+      "loss": 0.3029,
+      "step": 25665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7085311331007375e-05,
+      "loss": 0.2968,
+      "step": 25670
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.708397628752864e-05,
+      "loss": 0.2918,
+      "step": 25675
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7082640957304025e-05,
+      "loss": 0.3066,
+      "step": 25680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7081305340350875e-05,
+      "loss": 0.3111,
+      "step": 25685
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707996943668652e-05,
+      "loss": 0.2707,
+      "step": 25690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707863324632832e-05,
+      "loss": 0.3182,
+      "step": 25695
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707729676929361e-05,
+      "loss": 0.2845,
+      "step": 25700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7075960005599754e-05,
+      "loss": 0.289,
+      "step": 25705
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.707462295526411e-05,
+      "loss": 0.294,
+      "step": 25710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7073285618304034e-05,
+      "loss": 0.2875,
+      "step": 25715
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7071947994736887e-05,
+      "loss": 0.2998,
+      "step": 25720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.707061008458004e-05,
+      "loss": 0.3073,
+      "step": 25725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706927188785087e-05,
+      "loss": 0.2987,
+      "step": 25730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706793340456674e-05,
+      "loss": 0.3124,
+      "step": 25735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7066594634745055e-05,
+      "loss": 0.2986,
+      "step": 25740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7065255578403165e-05,
+      "loss": 0.3156,
+      "step": 25745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706391623555848e-05,
+      "loss": 0.3243,
+      "step": 25750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7062576606228385e-05,
+      "loss": 0.2935,
+      "step": 25755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.706123669043027e-05,
+      "loss": 0.2948,
+      "step": 25760
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705989648818154e-05,
+      "loss": 0.3191,
+      "step": 25765
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705855599949959e-05,
+      "loss": 0.3178,
+      "step": 25770
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705721522440182e-05,
+      "loss": 0.2991,
+      "step": 25775
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705587416290565e-05,
+      "loss": 0.3112,
+      "step": 25780
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7054532815028494e-05,
+      "loss": 0.3057,
+      "step": 25785
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.705319118078776e-05,
+      "loss": 0.3145,
+      "step": 25790
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7051849260200874e-05,
+      "loss": 0.2953,
+      "step": 25795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7050507053285256e-05,
+      "loss": 0.3187,
+      "step": 25800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.704916456005834e-05,
+      "loss": 0.3019,
+      "step": 25805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.704782178053755e-05,
+      "loss": 0.3052,
+      "step": 25810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.704647871474033e-05,
+      "loss": 0.3106,
+      "step": 25815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.70451353626841e-05,
+      "loss": 0.3061,
+      "step": 25820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7043791724386324e-05,
+      "loss": 0.2958,
+      "step": 25825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7042447799864444e-05,
+      "loss": 0.284,
+      "step": 25830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.70411035891359e-05,
+      "loss": 0.3004,
+      "step": 25835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703975909221815e-05,
+      "loss": 0.3132,
+      "step": 25840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703841430912865e-05,
+      "loss": 0.3095,
+      "step": 25845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703706923988488e-05,
+      "loss": 0.3004,
+      "step": 25850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703572388450427e-05,
+      "loss": 0.3101,
+      "step": 25855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703437824300432e-05,
+      "loss": 0.3068,
+      "step": 25860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703303231540248e-05,
+      "loss": 0.2825,
+      "step": 25865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7031686101716244e-05,
+      "loss": 0.3267,
+      "step": 25870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.703033960196308e-05,
+      "loss": 0.2971,
+      "step": 25875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702899281616048e-05,
+      "loss": 0.3243,
+      "step": 25880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702764574432592e-05,
+      "loss": 0.3012,
+      "step": 25885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.70262983864769e-05,
+      "loss": 0.3297,
+      "step": 25890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7024950742630906e-05,
+      "loss": 0.2954,
+      "step": 25895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7023602812805444e-05,
+      "loss": 0.2899,
+      "step": 25900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702225459701802e-05,
+      "loss": 0.319,
+      "step": 25905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702090609528612e-05,
+      "loss": 0.3112,
+      "step": 25910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7019557307627284e-05,
+      "loss": 0.2806,
+      "step": 25915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701820823405899e-05,
+      "loss": 0.2969,
+      "step": 25920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7016858874598786e-05,
+      "loss": 0.2954,
+      "step": 25925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7015509229264174e-05,
+      "loss": 0.3091,
+      "step": 25930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701415929807268e-05,
+      "loss": 0.3093,
+      "step": 25935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7012809081041845e-05,
+      "loss": 0.2918,
+      "step": 25940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.701145857818918e-05,
+      "loss": 0.2836,
+      "step": 25945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7010107789532236e-05,
+      "loss": 0.2904,
+      "step": 25950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7008756715088554e-05,
+      "loss": 0.3015,
+      "step": 25955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700740535487566e-05,
+      "loss": 0.3074,
+      "step": 25960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7006053708911115e-05,
+      "loss": 0.2716,
+      "step": 25965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7004701777212465e-05,
+      "loss": 0.3119,
+      "step": 25970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700334955979726e-05,
+      "loss": 0.3131,
+      "step": 25975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700199705668307e-05,
+      "loss": 0.2855,
+      "step": 25980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.700064426788744e-05,
+      "loss": 0.3164,
+      "step": 25985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6999291193427954e-05,
+      "loss": 0.2818,
+      "step": 25990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.699793783332216e-05,
+      "loss": 0.3179,
+      "step": 25995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.699658418758764e-05,
+      "loss": 0.2999,
+      "step": 26000
+    },
+    {
+      "epoch": 0.91,
+      "eval_loss": 0.2956346273422241,
+      "eval_runtime": 10.5342,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 26000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6995230256241984e-05,
+      "loss": 0.3159,
+      "step": 26005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.699387603930274e-05,
+      "loss": 0.3136,
+      "step": 26010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6992521536787525e-05,
+      "loss": 0.3187,
+      "step": 26015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.699116674871391e-05,
+      "loss": 0.3006,
+      "step": 26020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698981167509948e-05,
+      "loss": 0.3087,
+      "step": 26025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6988456315961835e-05,
+      "loss": 0.2858,
+      "step": 26030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698710067131859e-05,
+      "loss": 0.3003,
+      "step": 26035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6985744741187324e-05,
+      "loss": 0.2924,
+      "step": 26040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698438852558564e-05,
+      "loss": 0.3158,
+      "step": 26045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6983032024531173e-05,
+      "loss": 0.3382,
+      "step": 26050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698167523804152e-05,
+      "loss": 0.2996,
+      "step": 26055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.698031816613431e-05,
+      "loss": 0.3003,
+      "step": 26060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697896080882714e-05,
+      "loss": 0.2955,
+      "step": 26065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697760316613765e-05,
+      "loss": 0.3226,
+      "step": 26070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6976245238083474e-05,
+      "loss": 0.3099,
+      "step": 26075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6974887024682236e-05,
+      "loss": 0.3085,
+      "step": 26080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697352852595156e-05,
+      "loss": 0.3156,
+      "step": 26085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.697216974190911e-05,
+      "loss": 0.3053,
+      "step": 26090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6970810672572514e-05,
+      "loss": 0.2997,
+      "step": 26095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6969451317959415e-05,
+      "loss": 0.3295,
+      "step": 26100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696809167808748e-05,
+      "loss": 0.3008,
+      "step": 26105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696673175297433e-05,
+      "loss": 0.3188,
+      "step": 26110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696537154263767e-05,
+      "loss": 0.3141,
+      "step": 26115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696401104709512e-05,
+      "loss": 0.293,
+      "step": 26120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.696265026636437e-05,
+      "loss": 0.2927,
+      "step": 26125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6961289200463077e-05,
+      "loss": 0.2989,
+      "step": 26130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695992784940892e-05,
+      "loss": 0.2919,
+      "step": 26135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695856621321957e-05,
+      "loss": 0.3212,
+      "step": 26140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.69572042919127e-05,
+      "loss": 0.3111,
+      "step": 26145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695584208550602e-05,
+      "loss": 0.2652,
+      "step": 26150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695447959401719e-05,
+      "loss": 0.2956,
+      "step": 26155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695311681746391e-05,
+      "loss": 0.3111,
+      "step": 26160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695175375586388e-05,
+      "loss": 0.3066,
+      "step": 26165
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.695039040923479e-05,
+      "loss": 0.3114,
+      "step": 26170
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694902677759435e-05,
+      "loss": 0.3438,
+      "step": 26175
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6947662860960257e-05,
+      "loss": 0.3166,
+      "step": 26180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694629865935023e-05,
+      "loss": 0.2927,
+      "step": 26185
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694493417278197e-05,
+      "loss": 0.3508,
+      "step": 26190
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694356940127321e-05,
+      "loss": 0.3214,
+      "step": 26195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6942204344841665e-05,
+      "loss": 0.3333,
+      "step": 26200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.694083900350505e-05,
+      "loss": 0.2991,
+      "step": 26205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6939473377281105e-05,
+      "loss": 0.2859,
+      "step": 26210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6938107466187564e-05,
+      "loss": 0.3082,
+      "step": 26215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.693674127024215e-05,
+      "loss": 0.2948,
+      "step": 26220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.69353747894626e-05,
+      "loss": 0.3076,
+      "step": 26225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.693400802386667e-05,
+      "loss": 0.31,
+      "step": 26230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.69326409734721e-05,
+      "loss": 0.287,
+      "step": 26235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6931273638296635e-05,
+      "loss": 0.2953,
+      "step": 26240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6929906018358046e-05,
+      "loss": 0.2955,
+      "step": 26245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692853811367407e-05,
+      "loss": 0.302,
+      "step": 26250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6927169924262486e-05,
+      "loss": 0.3084,
+      "step": 26255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692580145014105e-05,
+      "loss": 0.3084,
+      "step": 26260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6924432691327524e-05,
+      "loss": 0.2952,
+      "step": 26265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6923063647839696e-05,
+      "loss": 0.2839,
+      "step": 26270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.692169431969533e-05,
+      "loss": 0.3139,
+      "step": 26275
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6920324706912215e-05,
+      "loss": 0.2949,
+      "step": 26280
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.691895480950813e-05,
+      "loss": 0.2875,
+      "step": 26285
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.6917584627500855e-05,
+      "loss": 0.2957,
+      "step": 26290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6916214160908194e-05,
+      "loss": 0.2871,
+      "step": 26295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.691484340974793e-05,
+      "loss": 0.2999,
+      "step": 26300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.691347237403787e-05,
+      "loss": 0.3261,
+      "step": 26305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.691210105379582e-05,
+      "loss": 0.3266,
+      "step": 26310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.691072944903957e-05,
+      "loss": 0.2979,
+      "step": 26315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6909357559786935e-05,
+      "loss": 0.3069,
+      "step": 26320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6907985386055735e-05,
+      "loss": 0.2832,
+      "step": 26325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6906612927863786e-05,
+      "loss": 0.2859,
+      "step": 26330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.69052401852289e-05,
+      "loss": 0.3004,
+      "step": 26335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.690386715816891e-05,
+      "loss": 0.2796,
+      "step": 26340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6902493846701637e-05,
+      "loss": 0.2625,
+      "step": 26345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.690112025084492e-05,
+      "loss": 0.2902,
+      "step": 26350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689974637061658e-05,
+      "loss": 0.294,
+      "step": 26355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689837220603448e-05,
+      "loss": 0.297,
+      "step": 26360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689699775711644e-05,
+      "loss": 0.2752,
+      "step": 26365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6895623023880313e-05,
+      "loss": 0.3214,
+      "step": 26370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689424800634395e-05,
+      "loss": 0.2921,
+      "step": 26375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6892872704525213e-05,
+      "loss": 0.2828,
+      "step": 26380
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6891497118441954e-05,
+      "loss": 0.2992,
+      "step": 26385
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.689012124811202e-05,
+      "loss": 0.3126,
+      "step": 26390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.68887450935533e-05,
+      "loss": 0.3218,
+      "step": 26395
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6887368654783634e-05,
+      "loss": 0.3072,
+      "step": 26400
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6885991931820925e-05,
+      "loss": 0.296,
+      "step": 26405
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6884614924683026e-05,
+      "loss": 0.2905,
+      "step": 26410
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688323763338783e-05,
+      "loss": 0.2966,
+      "step": 26415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6881860057953215e-05,
+      "loss": 0.3035,
+      "step": 26420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.688048219839707e-05,
+      "loss": 0.3109,
+      "step": 26425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6879104054737276e-05,
+      "loss": 0.3058,
+      "step": 26430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.687772562699174e-05,
+      "loss": 0.2764,
+      "step": 26435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.687634691517835e-05,
+      "loss": 0.3059,
+      "step": 26440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6874967919315016e-05,
+      "loss": 0.2983,
+      "step": 26445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.687358863941964e-05,
+      "loss": 0.3109,
+      "step": 26450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6872209075510134e-05,
+      "loss": 0.2926,
+      "step": 26455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.68708292276044e-05,
+      "loss": 0.3072,
+      "step": 26460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6869449095720364e-05,
+      "loss": 0.2999,
+      "step": 26465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.686806867987594e-05,
+      "loss": 0.2932,
+      "step": 26470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6866687980089064e-05,
+      "loss": 0.3021,
+      "step": 26475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.686530699637765e-05,
+      "loss": 0.3005,
+      "step": 26480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.686392572875964e-05,
+      "loss": 0.2977,
+      "step": 26485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.686254417725296e-05,
+      "loss": 0.2818,
+      "step": 26490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6861162341875556e-05,
+      "loss": 0.3014,
+      "step": 26495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6859780222645355e-05,
+      "loss": 0.3317,
+      "step": 26500
+    },
+    {
+      "epoch": 0.93,
+      "eval_loss": 0.29478785395622253,
+      "eval_runtime": 10.5324,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 26500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.685839781958032e-05,
+      "loss": 0.3181,
+      "step": 26505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.68570151326984e-05,
+      "loss": 0.3067,
+      "step": 26510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6855632162017535e-05,
+      "loss": 0.3111,
+      "step": 26515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6854248907555695e-05,
+      "loss": 0.3103,
+      "step": 26520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.685286536933083e-05,
+      "loss": 0.3183,
+      "step": 26525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.685148154736091e-05,
+      "loss": 0.3011,
+      "step": 26530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.685009744166391e-05,
+      "loss": 0.3188,
+      "step": 26535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6848713052257786e-05,
+      "loss": 0.303,
+      "step": 26540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684732837916053e-05,
+      "loss": 0.3034,
+      "step": 26545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6845943422390105e-05,
+      "loss": 0.2979,
+      "step": 26550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684455818196451e-05,
+      "loss": 0.2811,
+      "step": 26555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.6843172657901714e-05,
+      "loss": 0.2723,
+      "step": 26560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684178685021973e-05,
+      "loss": 0.3109,
+      "step": 26565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.684040075893652e-05,
+      "loss": 0.2876,
+      "step": 26570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.683901438407011e-05,
+      "loss": 0.2866,
+      "step": 26575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683762772563849e-05,
+      "loss": 0.2905,
+      "step": 26580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683624078365967e-05,
+      "loss": 0.2961,
+      "step": 26585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683485355815165e-05,
+      "loss": 0.2906,
+      "step": 26590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683346604913244e-05,
+      "loss": 0.3036,
+      "step": 26595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6832078256620065e-05,
+      "loss": 0.2968,
+      "step": 26600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.683069018063254e-05,
+      "loss": 0.3245,
+      "step": 26605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.68293018211879e-05,
+      "loss": 0.2988,
+      "step": 26610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.682791317830416e-05,
+      "loss": 0.2597,
+      "step": 26615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6826524251999344e-05,
+      "loss": 0.3083,
+      "step": 26620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6825135042291504e-05,
+      "loss": 0.2865,
+      "step": 26625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.682374554919866e-05,
+      "loss": 0.2915,
+      "step": 26630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6822355772738865e-05,
+      "loss": 0.2914,
+      "step": 26635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.682124374755895e-05,
+      "loss": 0.3391,
+      "step": 26640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681985346108412e-05,
+      "loss": 0.3162,
+      "step": 26645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681846289129287e-05,
+      "loss": 0.3289,
+      "step": 26650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681707203820325e-05,
+      "loss": 0.2981,
+      "step": 26655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6815680901833334e-05,
+      "loss": 0.2916,
+      "step": 26660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681428948220118e-05,
+      "loss": 0.3133,
+      "step": 26665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681289777932485e-05,
+      "loss": 0.2867,
+      "step": 26670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6811505793222426e-05,
+      "loss": 0.2839,
+      "step": 26675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.681011352391197e-05,
+      "loss": 0.2843,
+      "step": 26680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6808720971411556e-05,
+      "loss": 0.299,
+      "step": 26685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6807328135739284e-05,
+      "loss": 0.3218,
+      "step": 26690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6805935016913216e-05,
+      "loss": 0.3144,
+      "step": 26695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680454161495146e-05,
+      "loss": 0.3157,
+      "step": 26700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.68031479298721e-05,
+      "loss": 0.2977,
+      "step": 26705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.680175396169323e-05,
+      "loss": 0.2742,
+      "step": 26710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6800359710432955e-05,
+      "loss": 0.3116,
+      "step": 26715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.679896517610938e-05,
+      "loss": 0.2944,
+      "step": 26720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.67975703587406e-05,
+      "loss": 0.3056,
+      "step": 26725
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.679617525834474e-05,
+      "loss": 0.2851,
+      "step": 26730
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.67947798749399e-05,
+      "loss": 0.29,
+      "step": 26735
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.679338420854422e-05,
+      "loss": 0.2968,
+      "step": 26740
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.67919882591758e-05,
+      "loss": 0.3058,
+      "step": 26745
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.679059202685278e-05,
+      "loss": 0.3168,
+      "step": 26750
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678919551159327e-05,
+      "loss": 0.2959,
+      "step": 26755
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678779871341543e-05,
+      "loss": 0.2937,
+      "step": 26760
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6786401632337376e-05,
+      "loss": 0.2977,
+      "step": 26765
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678500426837725e-05,
+      "loss": 0.3166,
+      "step": 26770
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678360662155321e-05,
+      "loss": 0.3044,
+      "step": 26775
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.678220869188339e-05,
+      "loss": 0.3031,
+      "step": 26780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6780810479385945e-05,
+      "loss": 0.303,
+      "step": 26785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677941198407903e-05,
+      "loss": 0.307,
+      "step": 26790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677801320598081e-05,
+      "loss": 0.3096,
+      "step": 26795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677661414510943e-05,
+      "loss": 0.3295,
+      "step": 26800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677521480148307e-05,
+      "loss": 0.3012,
+      "step": 26805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.67738151751199e-05,
+      "loss": 0.2984,
+      "step": 26810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677241526603808e-05,
+      "loss": 0.3075,
+      "step": 26815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.677101507425581e-05,
+      "loss": 0.2996,
+      "step": 26820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6769614599791245e-05,
+      "loss": 0.3274,
+      "step": 26825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676821384266259e-05,
+      "loss": 0.3016,
+      "step": 26830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6766812802888016e-05,
+      "loss": 0.2927,
+      "step": 26835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676541148048573e-05,
+      "loss": 0.3203,
+      "step": 26840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6764009875473914e-05,
+      "loss": 0.2974,
+      "step": 26845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.676260798787078e-05,
+      "loss": 0.3076,
+      "step": 26850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6761205817694514e-05,
+      "loss": 0.3237,
+      "step": 26855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6759803364963336e-05,
+      "loss": 0.3246,
+      "step": 26860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6758400629695456e-05,
+      "loss": 0.3098,
+      "step": 26865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6756997611909074e-05,
+      "loss": 0.3105,
+      "step": 26870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675559431162242e-05,
+      "loss": 0.2933,
+      "step": 26875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675419072885372e-05,
+      "loss": 0.2906,
+      "step": 26880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675278686362118e-05,
+      "loss": 0.2907,
+      "step": 26885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.675138271594304e-05,
+      "loss": 0.3234,
+      "step": 26890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.674997828583753e-05,
+      "loss": 0.3188,
+      "step": 26895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6748573573322884e-05,
+      "loss": 0.3043,
+      "step": 26900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.674716857841735e-05,
+      "loss": 0.2734,
+      "step": 26905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6745763301139153e-05,
+      "loss": 0.2985,
+      "step": 26910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.674435774150656e-05,
+      "loss": 0.3309,
+      "step": 26915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.674295189953781e-05,
+      "loss": 0.3117,
+      "step": 26920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.674154577525116e-05,
+      "loss": 0.3057,
+      "step": 26925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6740139368664866e-05,
+      "loss": 0.3029,
+      "step": 26930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673873267979718e-05,
+      "loss": 0.3176,
+      "step": 26935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6737325708666394e-05,
+      "loss": 0.3005,
+      "step": 26940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673591845529074e-05,
+      "loss": 0.3125,
+      "step": 26945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6734510919688525e-05,
+      "loss": 0.3098,
+      "step": 26950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6733103101878e-05,
+      "loss": 0.3302,
+      "step": 26955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673169500187746e-05,
+      "loss": 0.2737,
+      "step": 26960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.673028661970518e-05,
+      "loss": 0.2857,
+      "step": 26965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672887795537945e-05,
+      "loss": 0.3199,
+      "step": 26970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672746900891856e-05,
+      "loss": 0.2955,
+      "step": 26975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672605978034079e-05,
+      "loss": 0.2969,
+      "step": 26980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6724650269664465e-05,
+      "loss": 0.3046,
+      "step": 26985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672324047690787e-05,
+      "loss": 0.3196,
+      "step": 26990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672183040208932e-05,
+      "loss": 0.2903,
+      "step": 26995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.672042004522711e-05,
+      "loss": 0.3032,
+      "step": 27000
+    },
+    {
+      "epoch": 0.95,
+      "eval_loss": 0.29417115449905396,
+      "eval_runtime": 10.5309,
+      "eval_samples_per_second": 9.496,
+      "eval_steps_per_second": 9.496,
+      "step": 27000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671900940633956e-05,
+      "loss": 0.3257,
+      "step": 27005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671759848544499e-05,
+      "loss": 0.3167,
+      "step": 27010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671618728256172e-05,
+      "loss": 0.2893,
+      "step": 27015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671477579770806e-05,
+      "loss": 0.2664,
+      "step": 27020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671336403090235e-05,
+      "loss": 0.3067,
+      "step": 27025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.671195198216293e-05,
+      "loss": 0.3055,
+      "step": 27030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.67105396515081e-05,
+      "loss": 0.2689,
+      "step": 27035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6709127038956235e-05,
+      "loss": 0.299,
+      "step": 27040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.670771414452566e-05,
+      "loss": 0.3221,
+      "step": 27045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.670630096823472e-05,
+      "loss": 0.3033,
+      "step": 27050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.670488751010177e-05,
+      "loss": 0.2964,
+      "step": 27055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6703473770145155e-05,
+      "loss": 0.3276,
+      "step": 27060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6702059748383244e-05,
+      "loss": 0.2896,
+      "step": 27065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.670064544483438e-05,
+      "loss": 0.3221,
+      "step": 27070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669923085951694e-05,
+      "loss": 0.3254,
+      "step": 27075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6697815992449287e-05,
+      "loss": 0.2988,
+      "step": 27080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.66964008436498e-05,
+      "loss": 0.2927,
+      "step": 27085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6694985413136835e-05,
+      "loss": 0.3356,
+      "step": 27090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6693569700928795e-05,
+      "loss": 0.3185,
+      "step": 27095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669215370704404e-05,
+      "loss": 0.2984,
+      "step": 27100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.669073743150096e-05,
+      "loss": 0.3017,
+      "step": 27105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.668932087431797e-05,
+      "loss": 0.3103,
+      "step": 27110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6687904035513416e-05,
+      "loss": 0.2788,
+      "step": 27115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.668648691510574e-05,
+      "loss": 0.2997,
+      "step": 27120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6685069513113315e-05,
+      "loss": 0.2898,
+      "step": 27125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.668365182955455e-05,
+      "loss": 0.2964,
+      "step": 27130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.6682233864447865e-05,
+      "loss": 0.2835,
+      "step": 27135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.668081561781166e-05,
+      "loss": 0.3101,
+      "step": 27140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.667939708966435e-05,
+      "loss": 0.3146,
+      "step": 27145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.667797828002436e-05,
+      "loss": 0.2676,
+      "step": 27150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.66765591889101e-05,
+      "loss": 0.2785,
+      "step": 27155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.667513981634001e-05,
+      "loss": 0.3098,
+      "step": 27160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.667372016233251e-05,
+      "loss": 0.2952,
+      "step": 27165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.667230022690605e-05,
+      "loss": 0.3123,
+      "step": 27170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.667088001007904e-05,
+      "loss": 0.2965,
+      "step": 27175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666945951186994e-05,
+      "loss": 0.3047,
+      "step": 27180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666803873229719e-05,
+      "loss": 0.2921,
+      "step": 27185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666661767137923e-05,
+      "loss": 0.3198,
+      "step": 27190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666519632913453e-05,
+      "loss": 0.2968,
+      "step": 27195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666377470558152e-05,
+      "loss": 0.2936,
+      "step": 27200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6662352800738684e-05,
+      "loss": 0.281,
+      "step": 27205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.666093061462447e-05,
+      "loss": 0.2865,
+      "step": 27210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.665950814725734e-05,
+      "loss": 0.2759,
+      "step": 27215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6658085398655784e-05,
+      "loss": 0.2907,
+      "step": 27220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6656662368838247e-05,
+      "loss": 0.291,
+      "step": 27225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.665523905782323e-05,
+      "loss": 0.3372,
+      "step": 27230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.665381546562921e-05,
+      "loss": 0.2823,
+      "step": 27235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.665239159227466e-05,
+      "loss": 0.3084,
+      "step": 27240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.665096743777807e-05,
+      "loss": 0.3115,
+      "step": 27245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664954300215795e-05,
+      "loss": 0.294,
+      "step": 27250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664811828543277e-05,
+      "loss": 0.3071,
+      "step": 27255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6646693287621045e-05,
+      "loss": 0.287,
+      "step": 27260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664526800874127e-05,
+      "loss": 0.2981,
+      "step": 27265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.664384244881196e-05,
+      "loss": 0.2819,
+      "step": 27270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6642416607851615e-05,
+      "loss": 0.3413,
+      "step": 27275
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6640990485878755e-05,
+      "loss": 0.2848,
+      "step": 27280
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6639564082911896e-05,
+      "loss": 0.3091,
+      "step": 27285
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.663813739896956e-05,
+      "loss": 0.2923,
+      "step": 27290
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.663671043407026e-05,
+      "loss": 0.3156,
+      "step": 27295
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.663528318823255e-05,
+      "loss": 0.2914,
+      "step": 27300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6633855661474935e-05,
+      "loss": 0.2957,
+      "step": 27305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6632427853815965e-05,
+      "loss": 0.2945,
+      "step": 27310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6630999765274175e-05,
+      "loss": 0.294,
+      "step": 27315
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662957139586811e-05,
+      "loss": 0.2932,
+      "step": 27320
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6628142745616313e-05,
+      "loss": 0.2979,
+      "step": 27325
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6626713814537335e-05,
+      "loss": 0.2806,
+      "step": 27330
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662528460264973e-05,
+      "loss": 0.289,
+      "step": 27335
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662385510997206e-05,
+      "loss": 0.3088,
+      "step": 27340
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.662242533652288e-05,
+      "loss": 0.2632,
+      "step": 27345
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6620995282320756e-05,
+      "loss": 0.2826,
+      "step": 27350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.661956494738426e-05,
+      "loss": 0.2917,
+      "step": 27355
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6618134331731956e-05,
+      "loss": 0.3059,
+      "step": 27360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.661670343538243e-05,
+      "loss": 0.2964,
+      "step": 27365
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6615272258354255e-05,
+      "loss": 0.3051,
+      "step": 27370
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.661384080066601e-05,
+      "loss": 0.3155,
+      "step": 27375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.661240906233629e-05,
+      "loss": 0.3184,
+      "step": 27380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.661097704338368e-05,
+      "loss": 0.2967,
+      "step": 27385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6609544743826775e-05,
+      "loss": 0.3054,
+      "step": 27390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6608112163684175e-05,
+      "loss": 0.3207,
+      "step": 27395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6606679302974474e-05,
+      "loss": 0.2907,
+      "step": 27400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660524616171629e-05,
+      "loss": 0.3062,
+      "step": 27405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660381273992822e-05,
+      "loss": 0.2796,
+      "step": 27410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660237903762887e-05,
+      "loss": 0.3121,
+      "step": 27415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.660094505483688e-05,
+      "loss": 0.2687,
+      "step": 27420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.659951079157084e-05,
+      "loss": 0.319,
+      "step": 27425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65980762478494e-05,
+      "loss": 0.3121,
+      "step": 27430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659664142369116e-05,
+      "loss": 0.3029,
+      "step": 27435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659520631911477e-05,
+      "loss": 0.2954,
+      "step": 27440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659377093413886e-05,
+      "loss": 0.2916,
+      "step": 27445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659233526878206e-05,
+      "loss": 0.2977,
+      "step": 27450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.659089932306302e-05,
+      "loss": 0.2922,
+      "step": 27455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658946309700039e-05,
+      "loss": 0.3278,
+      "step": 27460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658802659061279e-05,
+      "loss": 0.2658,
+      "step": 27465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6586589803918905e-05,
+      "loss": 0.2951,
+      "step": 27470
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658515273693738e-05,
+      "loss": 0.3163,
+      "step": 27475
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658371538968686e-05,
+      "loss": 0.2748,
+      "step": 27480
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658227776218603e-05,
+      "loss": 0.3057,
+      "step": 27485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.658083985445354e-05,
+      "loss": 0.2993,
+      "step": 27490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6579401666508074e-05,
+      "loss": 0.3059,
+      "step": 27495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.657796319836829e-05,
+      "loss": 0.3165,
+      "step": 27500
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 0.293122798204422,
+      "eval_runtime": 10.5363,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 27500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.657652445005288e-05,
+      "loss": 0.2904,
+      "step": 27505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.657508542158052e-05,
+      "loss": 0.2853,
+      "step": 27510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.657364611296989e-05,
+      "loss": 0.2936,
+      "step": 27515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6572206524239686e-05,
+      "loss": 0.2737,
+      "step": 27520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6570766655408594e-05,
+      "loss": 0.2966,
+      "step": 27525
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656932650649532e-05,
+      "loss": 0.3009,
+      "step": 27530
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656788607751855e-05,
+      "loss": 0.3059,
+      "step": 27535
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656644536849699e-05,
+      "loss": 0.2924,
+      "step": 27540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656500437944936e-05,
+      "loss": 0.3073,
+      "step": 27545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656356311039435e-05,
+      "loss": 0.295,
+      "step": 27550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6562121561350686e-05,
+      "loss": 0.2916,
+      "step": 27555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.656067973233709e-05,
+      "loss": 0.3131,
+      "step": 27560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6559237623372265e-05,
+      "loss": 0.2935,
+      "step": 27565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655779523447495e-05,
+      "loss": 0.3054,
+      "step": 27570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6556352565663885e-05,
+      "loss": 0.3058,
+      "step": 27575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6554909616957774e-05,
+      "loss": 0.2799,
+      "step": 27580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6553466388375375e-05,
+      "loss": 0.3288,
+      "step": 27585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.655202287993541e-05,
+      "loss": 0.3051,
+      "step": 27590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6550579091656646e-05,
+      "loss": 0.2936,
+      "step": 27595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65491350235578e-05,
+      "loss": 0.3018,
+      "step": 27600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654769067565764e-05,
+      "loss": 0.29,
+      "step": 27605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654624604797492e-05,
+      "loss": 0.2886,
+      "step": 27610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65448011405284e-05,
+      "loss": 0.3219,
+      "step": 27615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654335595333683e-05,
+      "loss": 0.2825,
+      "step": 27620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.654191048641898e-05,
+      "loss": 0.2895,
+      "step": 27625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6540464739793624e-05,
+      "loss": 0.3029,
+      "step": 27630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653901871347952e-05,
+      "loss": 0.3063,
+      "step": 27635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653757240749546e-05,
+      "loss": 0.3149,
+      "step": 27640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653612582186021e-05,
+      "loss": 0.3045,
+      "step": 27645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653467895659256e-05,
+      "loss": 0.2987,
+      "step": 27650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6533231811711296e-05,
+      "loss": 0.3002,
+      "step": 27655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653178438723521e-05,
+      "loss": 0.3108,
+      "step": 27660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.653033668318309e-05,
+      "loss": 0.2842,
+      "step": 27665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652888869957375e-05,
+      "loss": 0.2661,
+      "step": 27670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652744043642597e-05,
+      "loss": 0.312,
+      "step": 27675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652599189375856e-05,
+      "loss": 0.3114,
+      "step": 27680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652454307159033e-05,
+      "loss": 0.3029,
+      "step": 27685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.65230939699401e-05,
+      "loss": 0.326,
+      "step": 27690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6521644588826675e-05,
+      "loss": 0.3124,
+      "step": 27695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.652019492826888e-05,
+      "loss": 0.2903,
+      "step": 27700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6518744988285536e-05,
+      "loss": 0.3094,
+      "step": 27705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.651729476889547e-05,
+      "loss": 0.2816,
+      "step": 27710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.651584427011751e-05,
+      "loss": 0.3151,
+      "step": 27715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6514393491970496e-05,
+      "loss": 0.291,
+      "step": 27720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.651294243447326e-05,
+      "loss": 0.2991,
+      "step": 27725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.651149109764464e-05,
+      "loss": 0.301,
+      "step": 27730
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.651003948150349e-05,
+      "loss": 0.2971,
+      "step": 27735
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6508587586068644e-05,
+      "loss": 0.2991,
+      "step": 27740
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650713541135897e-05,
+      "loss": 0.3036,
+      "step": 27745
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650568295739331e-05,
+      "loss": 0.3455,
+      "step": 27750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650423022419054e-05,
+      "loss": 0.3175,
+      "step": 27755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.650277721176951e-05,
+      "loss": 0.3293,
+      "step": 27760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6501323920149086e-05,
+      "loss": 0.2894,
+      "step": 27765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.649987034934814e-05,
+      "loss": 0.3213,
+      "step": 27770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6498416499385546e-05,
+      "loss": 0.2862,
+      "step": 27775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6496962370280183e-05,
+      "loss": 0.3092,
+      "step": 27780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6495507962050936e-05,
+      "loss": 0.2987,
+      "step": 27785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.649405327471667e-05,
+      "loss": 0.3139,
+      "step": 27790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64925983082963e-05,
+      "loss": 0.2881,
+      "step": 27795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6491143062808696e-05,
+      "loss": 0.3101,
+      "step": 27800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648968753827277e-05,
+      "loss": 0.291,
+      "step": 27805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648823173470741e-05,
+      "loss": 0.2974,
+      "step": 27810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648677565213152e-05,
+      "loss": 0.3195,
+      "step": 27815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.648531929056401e-05,
+      "loss": 0.2996,
+      "step": 27820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64838626500238e-05,
+      "loss": 0.2848,
+      "step": 27825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6482405730529773e-05,
+      "loss": 0.3081,
+      "step": 27830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6480948532100875e-05,
+      "loss": 0.3071,
+      "step": 27835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647949105475602e-05,
+      "loss": 0.3182,
+      "step": 27840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6478033298514124e-05,
+      "loss": 0.3004,
+      "step": 27845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6476575263394124e-05,
+      "loss": 0.2944,
+      "step": 27850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647511694941494e-05,
+      "loss": 0.305,
+      "step": 27855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6473658356595525e-05,
+      "loss": 0.2991,
+      "step": 27860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6472199484954805e-05,
+      "loss": 0.2867,
+      "step": 27865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.647074033451172e-05,
+      "loss": 0.3062,
+      "step": 27870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6469280905285224e-05,
+      "loss": 0.2917,
+      "step": 27875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.646782119729426e-05,
+      "loss": 0.3011,
+      "step": 27880
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6466361210557794e-05,
+      "loss": 0.2834,
+      "step": 27885
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.646490094509477e-05,
+      "loss": 0.3241,
+      "step": 27890
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6463440400924156e-05,
+      "loss": 0.3104,
+      "step": 27895
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.646197957806491e-05,
+      "loss": 0.3255,
+      "step": 27900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.646051847653601e-05,
+      "loss": 0.3279,
+      "step": 27905
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6459057096356414e-05,
+      "loss": 0.2838,
+      "step": 27910
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64575954375451e-05,
+      "loss": 0.3108,
+      "step": 27915
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.645613350012106e-05,
+      "loss": 0.3067,
+      "step": 27920
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6454671284103255e-05,
+      "loss": 0.3262,
+      "step": 27925
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6453208789510695e-05,
+      "loss": 0.3154,
+      "step": 27930
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.645174601636235e-05,
+      "loss": 0.3076,
+      "step": 27935
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.645028296467722e-05,
+      "loss": 0.3014,
+      "step": 27940
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6448819634474305e-05,
+      "loss": 0.279,
+      "step": 27945
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.64473560257726e-05,
+      "loss": 0.2873,
+      "step": 27950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644589213859112e-05,
+      "loss": 0.2604,
+      "step": 27955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644442797294886e-05,
+      "loss": 0.3031,
+      "step": 27960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644296352886482e-05,
+      "loss": 0.3184,
+      "step": 27965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644149880635804e-05,
+      "loss": 0.3232,
+      "step": 27970
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.644003380544754e-05,
+      "loss": 0.3124,
+      "step": 27975
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6438568526152315e-05,
+      "loss": 0.281,
+      "step": 27980
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6437102968491416e-05,
+      "loss": 0.3167,
+      "step": 27985
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6435637132483855e-05,
+      "loss": 0.2948,
+      "step": 27990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6434171018148674e-05,
+      "loss": 0.2974,
+      "step": 27995
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6432704625504915e-05,
+      "loss": 0.31,
+      "step": 28000
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 0.2925254702568054,
+      "eval_runtime": 10.5771,
+      "eval_samples_per_second": 9.454,
+      "eval_steps_per_second": 9.454,
+      "step": 28000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6431237954571605e-05,
+      "loss": 0.3,
+      "step": 28005
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.64297710053678e-05,
+      "loss": 0.3034,
+      "step": 28010
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642830377791253e-05,
+      "loss": 0.2771,
+      "step": 28015
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642683627222486e-05,
+      "loss": 0.2768,
+      "step": 28020
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642536848832384e-05,
+      "loss": 0.2962,
+      "step": 28025
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642390042622854e-05,
+      "loss": 0.2933,
+      "step": 28030
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6422432085958014e-05,
+      "loss": 0.282,
+      "step": 28035
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.642096346753131e-05,
+      "loss": 0.3212,
+      "step": 28040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6419494570967524e-05,
+      "loss": 0.2929,
+      "step": 28045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6418025396285716e-05,
+      "loss": 0.291,
+      "step": 28050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641655594350496e-05,
+      "loss": 0.2882,
+      "step": 28055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641508621264434e-05,
+      "loss": 0.3456,
+      "step": 28060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641361620372294e-05,
+      "loss": 0.2915,
+      "step": 28065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6412145916759844e-05,
+      "loss": 0.3201,
+      "step": 28070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.641067535177414e-05,
+      "loss": 0.3373,
+      "step": 28075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.640920450878493e-05,
+      "loss": 0.303,
+      "step": 28080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6407733387811314e-05,
+      "loss": 0.278,
+      "step": 28085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.640626198887238e-05,
+      "loss": 0.2931,
+      "step": 28090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6404790311987245e-05,
+      "loss": 0.2942,
+      "step": 28095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6403318357175016e-05,
+      "loss": 0.2935,
+      "step": 28100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.640184612445481e-05,
+      "loss": 0.2873,
+      "step": 28105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.640037361384573e-05,
+      "loss": 0.3022,
+      "step": 28110
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6398900825366895e-05,
+      "loss": 0.2989,
+      "step": 28115
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6397427759037437e-05,
+      "loss": 0.3209,
+      "step": 28120
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639595441487649e-05,
+      "loss": 0.312,
+      "step": 28125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639448079290317e-05,
+      "loss": 0.2821,
+      "step": 28130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6393006893136616e-05,
+      "loss": 0.2885,
+      "step": 28135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6391532715595964e-05,
+      "loss": 0.2991,
+      "step": 28140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.639005826030035e-05,
+      "loss": 0.326,
+      "step": 28145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638858352726895e-05,
+      "loss": 0.2807,
+      "step": 28150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638710851652087e-05,
+      "loss": 0.2922,
+      "step": 28155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638563322807528e-05,
+      "loss": 0.2952,
+      "step": 28160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638415766195135e-05,
+      "loss": 0.2912,
+      "step": 28165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.638268181816821e-05,
+      "loss": 0.285,
+      "step": 28170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6381205696745046e-05,
+      "loss": 0.2991,
+      "step": 28175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637972929770101e-05,
+      "loss": 0.3202,
+      "step": 28180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637825262105528e-05,
+      "loss": 0.2775,
+      "step": 28185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6376775666827045e-05,
+      "loss": 0.3135,
+      "step": 28190
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6375298435035455e-05,
+      "loss": 0.3039,
+      "step": 28195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.63738209256997e-05,
+      "loss": 0.3044,
+      "step": 28200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637234313883896e-05,
+      "loss": 0.3249,
+      "step": 28205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6370865074472435e-05,
+      "loss": 0.3282,
+      "step": 28210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636938673261931e-05,
+      "loss": 0.3199,
+      "step": 28215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6367908113298784e-05,
+      "loss": 0.2861,
+      "step": 28220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636642921653006e-05,
+      "loss": 0.2998,
+      "step": 28225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6364950042332326e-05,
+      "loss": 0.2819,
+      "step": 28230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636347059072479e-05,
+      "loss": 0.312,
+      "step": 28235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636199086172667e-05,
+      "loss": 0.2945,
+      "step": 28240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.636051085535718e-05,
+      "loss": 0.2942,
+      "step": 28245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635903057163553e-05,
+      "loss": 0.3138,
+      "step": 28250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6357550010580954e-05,
+      "loss": 0.2827,
+      "step": 28255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635606917221266e-05,
+      "loss": 0.2791,
+      "step": 28260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6354588056549883e-05,
+      "loss": 0.3206,
+      "step": 28265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6353106663611843e-05,
+      "loss": 0.3046,
+      "step": 28270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6351624993417795e-05,
+      "loss": 0.3276,
+      "step": 28275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.635014304598697e-05,
+      "loss": 0.3149,
+      "step": 28280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6348660821338605e-05,
+      "loss": 0.2965,
+      "step": 28285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6347178319491946e-05,
+      "loss": 0.2983,
+      "step": 28290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.634569554046625e-05,
+      "loss": 0.3216,
+      "step": 28295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6344212484280756e-05,
+      "loss": 0.3091,
+      "step": 28300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.634272915095475e-05,
+      "loss": 0.3023,
+      "step": 28305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.634124554050745e-05,
+      "loss": 0.3142,
+      "step": 28310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633976165295815e-05,
+      "loss": 0.2853,
+      "step": 28315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6338277488326095e-05,
+      "loss": 0.3037,
+      "step": 28320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633679304663059e-05,
+      "loss": 0.3117,
+      "step": 28325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633530832789088e-05,
+      "loss": 0.2757,
+      "step": 28330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633382333212625e-05,
+      "loss": 0.3027,
+      "step": 28335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633233805935598e-05,
+      "loss": 0.2905,
+      "step": 28340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6330852509599364e-05,
+      "loss": 0.3119,
+      "step": 28345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632936668287568e-05,
+      "loss": 0.2912,
+      "step": 28350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632788057920423e-05,
+      "loss": 0.3334,
+      "step": 28355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632639419860431e-05,
+      "loss": 0.3038,
+      "step": 28360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632490754109521e-05,
+      "loss": 0.2781,
+      "step": 28365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6323420606696245e-05,
+      "loss": 0.2914,
+      "step": 28370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632193339542671e-05,
+      "loss": 0.304,
+      "step": 28375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.632044590730593e-05,
+      "loss": 0.2986,
+      "step": 28380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631895814235321e-05,
+      "loss": 0.3096,
+      "step": 28385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6317470100587856e-05,
+      "loss": 0.3176,
+      "step": 28390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631598178202921e-05,
+      "loss": 0.3122,
+      "step": 28395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631449318669659e-05,
+      "loss": 0.2804,
+      "step": 28400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631300431460932e-05,
+      "loss": 0.2944,
+      "step": 28405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6311515165786736e-05,
+      "loss": 0.3003,
+      "step": 28410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.631002574024818e-05,
+      "loss": 0.2809,
+      "step": 28415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630853603801296e-05,
+      "loss": 0.3108,
+      "step": 28420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6307046059100465e-05,
+      "loss": 0.2931,
+      "step": 28425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630555580353001e-05,
+      "loss": 0.2833,
+      "step": 28430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630406527132095e-05,
+      "loss": 0.2914,
+      "step": 28435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6302574462492656e-05,
+      "loss": 0.2863,
+      "step": 28440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.630108337706446e-05,
+      "loss": 0.2728,
+      "step": 28445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629959201505574e-05,
+      "loss": 0.2981,
+      "step": 28450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.629810037648585e-05,
+      "loss": 0.2896,
+      "step": 28455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6296608461374166e-05,
+      "loss": 0.2883,
+      "step": 28460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6295116269740055e-05,
+      "loss": 0.2993,
+      "step": 28465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6293623801602894e-05,
+      "loss": 0.2843,
+      "step": 28470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6292131056982056e-05,
+      "loss": 0.2819,
+      "step": 28475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6290638035896935e-05,
+      "loss": 0.2676,
+      "step": 28480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6289144738366905e-05,
+      "loss": 0.2986,
+      "step": 28485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.628765116441136e-05,
+      "loss": 0.2851,
+      "step": 28490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.62861573140497e-05,
+      "loss": 0.3139,
+      "step": 28495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.628466318730131e-05,
+      "loss": 0.2856,
+      "step": 28500
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.2926534414291382,
+      "eval_runtime": 10.5975,
+      "eval_samples_per_second": 9.436,
+      "eval_steps_per_second": 9.436,
+      "step": 28500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.62831687841856e-05,
+      "loss": 0.2937,
+      "step": 28505
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6281674104721966e-05,
+      "loss": 0.2959,
+      "step": 28510
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.628017914892982e-05,
+      "loss": 0.2815,
+      "step": 28515
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627868391682857e-05,
+      "loss": 0.2922,
+      "step": 28520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627718840843763e-05,
+      "loss": 0.3035,
+      "step": 28525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6275692623776434e-05,
+      "loss": 0.2818,
+      "step": 28530
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627419656286438e-05,
+      "loss": 0.3022,
+      "step": 28535
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6272700225720906e-05,
+      "loss": 0.2987,
+      "step": 28540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.627120361236544e-05,
+      "loss": 0.2933,
+      "step": 28545
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.626970672281741e-05,
+      "loss": 0.2709,
+      "step": 28550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6268209557096256e-05,
+      "loss": 0.2913,
+      "step": 28555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.626671211522142e-05,
+      "loss": 0.2927,
+      "step": 28560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6265214397212344e-05,
+      "loss": 0.27,
+      "step": 28565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6263716403088464e-05,
+      "loss": 0.3177,
+      "step": 28570
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.626221813286925e-05,
+      "loss": 0.3077,
+      "step": 28575
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.626071958657414e-05,
+      "loss": 0.2766,
+      "step": 28580
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.62592207642226e-05,
+      "loss": 0.3328,
+      "step": 28585
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625772166583409e-05,
+      "loss": 0.2749,
+      "step": 28590
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625622229142806e-05,
+      "loss": 0.266,
+      "step": 28595
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625472264102401e-05,
+      "loss": 0.2969,
+      "step": 28600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6253222714641386e-05,
+      "loss": 0.2833,
+      "step": 28605
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.625172251229968e-05,
+      "loss": 0.2772,
+      "step": 28610
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6250222034018354e-05,
+      "loss": 0.3067,
+      "step": 28615
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.62487212798169e-05,
+      "loss": 0.2986,
+      "step": 28620
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.62472202497148e-05,
+      "loss": 0.3008,
+      "step": 28625
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6245718943731556e-05,
+      "loss": 0.2753,
+      "step": 28630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.624421736188664e-05,
+      "loss": 0.2953,
+      "step": 28635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.624271550419957e-05,
+      "loss": 0.2925,
+      "step": 28640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.624121337068985e-05,
+      "loss": 0.302,
+      "step": 28645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623971096137696e-05,
+      "loss": 0.3007,
+      "step": 28650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6238208276280425e-05,
+      "loss": 0.2937,
+      "step": 28655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6236705315419746e-05,
+      "loss": 0.2881,
+      "step": 28660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623520207881445e-05,
+      "loss": 0.2761,
+      "step": 28665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623369856648405e-05,
+      "loss": 0.2717,
+      "step": 28670
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6232194778448064e-05,
+      "loss": 0.2863,
+      "step": 28675
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.623069071472602e-05,
+      "loss": 0.2899,
+      "step": 28680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622918637533745e-05,
+      "loss": 0.2708,
+      "step": 28685
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622768176030189e-05,
+      "loss": 0.3391,
+      "step": 28690
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622617686963887e-05,
+      "loss": 0.3078,
+      "step": 28695
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6224671703367925e-05,
+      "loss": 0.2943,
+      "step": 28700
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622316626150861e-05,
+      "loss": 0.3048,
+      "step": 28705
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622166054408047e-05,
+      "loss": 0.2815,
+      "step": 28710
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.622015455110306e-05,
+      "loss": 0.2984,
+      "step": 28715
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621864828259592e-05,
+      "loss": 0.2911,
+      "step": 28720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6217141738578605e-05,
+      "loss": 0.3086,
+      "step": 28725
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6215634919070706e-05,
+      "loss": 0.2838,
+      "step": 28730
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621412782409176e-05,
+      "loss": 0.2952,
+      "step": 28735
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621262045366134e-05,
+      "loss": 0.2666,
+      "step": 28740
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.621111280779903e-05,
+      "loss": 0.2786,
+      "step": 28745
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.62096048865244e-05,
+      "loss": 0.284,
+      "step": 28750
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.620809668985703e-05,
+      "loss": 0.2934,
+      "step": 28755
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6206588217816494e-05,
+      "loss": 0.3039,
+      "step": 28760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6205079470422385e-05,
+      "loss": 0.2786,
+      "step": 28765
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.62035704476943e-05,
+      "loss": 0.2848,
+      "step": 28770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.620206114965182e-05,
+      "loss": 0.2921,
+      "step": 28775
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6200551576314555e-05,
+      "loss": 0.2966,
+      "step": 28780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.61990417277021e-05,
+      "loss": 0.2852,
+      "step": 28785
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619753160383405e-05,
+      "loss": 0.2852,
+      "step": 28790
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619602120473003e-05,
+      "loss": 0.2634,
+      "step": 28795
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619451053040964e-05,
+      "loss": 0.2672,
+      "step": 28800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6192999580892505e-05,
+      "loss": 0.2912,
+      "step": 28805
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.619148835619823e-05,
+      "loss": 0.2948,
+      "step": 28810
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6189976856346445e-05,
+      "loss": 0.3218,
+      "step": 28815
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.618846508135678e-05,
+      "loss": 0.2818,
+      "step": 28820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6186953031248856e-05,
+      "loss": 0.2722,
+      "step": 28825
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6185440706042303e-05,
+      "loss": 0.2779,
+      "step": 28830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.618392810575677e-05,
+      "loss": 0.3148,
+      "step": 28835
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6182415230411896e-05,
+      "loss": 0.2818,
+      "step": 28840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.618090208002731e-05,
+      "loss": 0.3025,
+      "step": 28845
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.617938865462268e-05,
+      "loss": 0.279,
+      "step": 28850
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.617787495421764e-05,
+      "loss": 0.2818,
+      "step": 28855
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6176360978831854e-05,
+      "loss": 0.2788,
+      "step": 28860
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.617484672848498e-05,
+      "loss": 0.2834,
+      "step": 28865
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.617333220319666e-05,
+      "loss": 0.3027,
+      "step": 28870
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.617181740298659e-05,
+      "loss": 0.2951,
+      "step": 28875
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.617030232787441e-05,
+      "loss": 0.3074,
+      "step": 28880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616878697787982e-05,
+      "loss": 0.2972,
+      "step": 28885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616727135302248e-05,
+      "loss": 0.3025,
+      "step": 28890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616575545332207e-05,
+      "loss": 0.2992,
+      "step": 28895
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616423927879827e-05,
+      "loss": 0.2954,
+      "step": 28900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6162722829470774e-05,
+      "loss": 0.2932,
+      "step": 28905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.616120610535927e-05,
+      "loss": 0.3215,
+      "step": 28910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.615968910648345e-05,
+      "loss": 0.3091,
+      "step": 28915
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.615817183286301e-05,
+      "loss": 0.3195,
+      "step": 28920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.615665428451765e-05,
+      "loss": 0.2676,
+      "step": 28925
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.615513646146707e-05,
+      "loss": 0.2835,
+      "step": 28930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6153618363731e-05,
+      "loss": 0.2917,
+      "step": 28935
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6152099991329135e-05,
+      "loss": 0.3195,
+      "step": 28940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.615058134428118e-05,
+      "loss": 0.2848,
+      "step": 28945
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.614906242260687e-05,
+      "loss": 0.3069,
+      "step": 28950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.614754322632593e-05,
+      "loss": 0.2777,
+      "step": 28955
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6146023755458065e-05,
+      "loss": 0.3258,
+      "step": 28960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.614450401002303e-05,
+      "loss": 0.2998,
+      "step": 28965
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6143288015999886e-05,
+      "loss": 0.2975,
+      "step": 28970
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.614176777639365e-05,
+      "loss": 0.2925,
+      "step": 28975
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.614024726227548e-05,
+      "loss": 0.3256,
+      "step": 28980
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6138726473665144e-05,
+      "loss": 0.296,
+      "step": 28985
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.613720541058237e-05,
+      "loss": 0.2818,
+      "step": 28990
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.613568407304692e-05,
+      "loss": 0.2902,
+      "step": 28995
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6134162461078535e-05,
+      "loss": 0.3066,
+      "step": 29000
+    },
+    {
+      "epoch": 1.02,
+      "eval_loss": 0.2924649715423584,
+      "eval_runtime": 10.5475,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 29000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.613264057469697e-05,
+      "loss": 0.3196,
+      "step": 29005
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6131118413922015e-05,
+      "loss": 0.2835,
+      "step": 29010
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.61295959787734e-05,
+      "loss": 0.2965,
+      "step": 29015
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.61280732692709e-05,
+      "loss": 0.2943,
+      "step": 29020
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.61265502854343e-05,
+      "loss": 0.2936,
+      "step": 29025
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6125027027283375e-05,
+      "loss": 0.2984,
+      "step": 29030
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.612350349483788e-05,
+      "loss": 0.2713,
+      "step": 29035
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6121979688117626e-05,
+      "loss": 0.293,
+      "step": 29040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6120455607142385e-05,
+      "loss": 0.2994,
+      "step": 29045
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.611893125193194e-05,
+      "loss": 0.29,
+      "step": 29050
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.61174066225061e-05,
+      "loss": 0.2924,
+      "step": 29055
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.611588171888464e-05,
+      "loss": 0.3052,
+      "step": 29060
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.611435654108738e-05,
+      "loss": 0.2863,
+      "step": 29065
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.611283108913411e-05,
+      "loss": 0.2687,
+      "step": 29070
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6111305363044644e-05,
+      "loss": 0.2993,
+      "step": 29075
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6109779362838794e-05,
+      "loss": 0.293,
+      "step": 29080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.610825308853637e-05,
+      "loss": 0.2927,
+      "step": 29085
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.610672654015719e-05,
+      "loss": 0.3053,
+      "step": 29090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.610519971772107e-05,
+      "loss": 0.3047,
+      "step": 29095
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.610367262124785e-05,
+      "loss": 0.3033,
+      "step": 29100
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6102145250757345e-05,
+      "loss": 0.3131,
+      "step": 29105
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6100617606269385e-05,
+      "loss": 0.3028,
+      "step": 29110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.609908968780382e-05,
+      "loss": 0.2893,
+      "step": 29115
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.6097561495380474e-05,
+      "loss": 0.275,
+      "step": 29120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.60960330290192e-05,
+      "loss": 0.2819,
+      "step": 29125
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.609450428873984e-05,
+      "loss": 0.2878,
+      "step": 29130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.609297527456224e-05,
+      "loss": 0.3103,
+      "step": 29135
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.609144598650625e-05,
+      "loss": 0.2843,
+      "step": 29140
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608991642459175e-05,
+      "loss": 0.3022,
+      "step": 29145
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6088386588838574e-05,
+      "loss": 0.3251,
+      "step": 29150
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.60868564792666e-05,
+      "loss": 0.3163,
+      "step": 29155
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6085326095895685e-05,
+      "loss": 0.2952,
+      "step": 29160
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608379543874571e-05,
+      "loss": 0.3182,
+      "step": 29165
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.608226450783655e-05,
+      "loss": 0.2704,
+      "step": 29170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6080733303188075e-05,
+      "loss": 0.3,
+      "step": 29175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607920182482017e-05,
+      "loss": 0.2958,
+      "step": 29180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607767007275273e-05,
+      "loss": 0.2918,
+      "step": 29185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607613804700562e-05,
+      "loss": 0.307,
+      "step": 29190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607460574759876e-05,
+      "loss": 0.3087,
+      "step": 29195
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607307317455203e-05,
+      "loss": 0.2707,
+      "step": 29200
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607154032788533e-05,
+      "loss": 0.3145,
+      "step": 29205
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.607000720761856e-05,
+      "loss": 0.29,
+      "step": 29210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606847381377164e-05,
+      "loss": 0.2782,
+      "step": 29215
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606694014636447e-05,
+      "loss": 0.3053,
+      "step": 29220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6065406205416966e-05,
+      "loss": 0.2973,
+      "step": 29225
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606387199094905e-05,
+      "loss": 0.2895,
+      "step": 29230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.606233750298063e-05,
+      "loss": 0.2861,
+      "step": 29235
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6060802741531636e-05,
+      "loss": 0.2772,
+      "step": 29240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605926770662201e-05,
+      "loss": 0.3141,
+      "step": 29245
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605773239827166e-05,
+      "loss": 0.2708,
+      "step": 29250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605619681650054e-05,
+      "loss": 0.2624,
+      "step": 29255
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605466096132858e-05,
+      "loss": 0.3042,
+      "step": 29260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605312483277571e-05,
+      "loss": 0.3024,
+      "step": 29265
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.60515884308619e-05,
+      "loss": 0.3008,
+      "step": 29270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.605005175560709e-05,
+      "loss": 0.3072,
+      "step": 29275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.604851480703123e-05,
+      "loss": 0.2906,
+      "step": 29280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.604697758515427e-05,
+      "loss": 0.2937,
+      "step": 29285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.604544008999618e-05,
+      "loss": 0.2759,
+      "step": 29290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6043902321576926e-05,
+      "loss": 0.3239,
+      "step": 29295
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6042364279916464e-05,
+      "loss": 0.2735,
+      "step": 29300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6040825965034764e-05,
+      "loss": 0.3082,
+      "step": 29305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603928737695181e-05,
+      "loss": 0.2771,
+      "step": 29310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6037748515687576e-05,
+      "loss": 0.3013,
+      "step": 29315
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603620938126204e-05,
+      "loss": 0.2855,
+      "step": 29320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603466997369519e-05,
+      "loss": 0.3087,
+      "step": 29325
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603313029300701e-05,
+      "loss": 0.2584,
+      "step": 29330
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.603159033921749e-05,
+      "loss": 0.294,
+      "step": 29335
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6030050112346643e-05,
+      "loss": 0.3152,
+      "step": 29340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602850961241445e-05,
+      "loss": 0.2669,
+      "step": 29345
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602696883944091e-05,
+      "loss": 0.2989,
+      "step": 29350
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602542779344605e-05,
+      "loss": 0.2704,
+      "step": 29355
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.602388647444985e-05,
+      "loss": 0.2651,
+      "step": 29360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6022344882472354e-05,
+      "loss": 0.2857,
+      "step": 29365
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6020803017533555e-05,
+      "loss": 0.319,
+      "step": 29370
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6019260879653486e-05,
+      "loss": 0.2883,
+      "step": 29375
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6017718468852165e-05,
+      "loss": 0.2926,
+      "step": 29380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.601617578514962e-05,
+      "loss": 0.2898,
+      "step": 29385
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6014632828565887e-05,
+      "loss": 0.323,
+      "step": 29390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.6013089599121e-05,
+      "loss": 0.2977,
+      "step": 29395
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.601154609683498e-05,
+      "loss": 0.3198,
+      "step": 29400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.601000232172789e-05,
+      "loss": 0.2892,
+      "step": 29405
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.600845827381976e-05,
+      "loss": 0.2732,
+      "step": 29410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.600691395313065e-05,
+      "loss": 0.3062,
+      "step": 29415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.600536935968061e-05,
+      "loss": 0.2863,
+      "step": 29420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.6003824493489676e-05,
+      "loss": 0.2974,
+      "step": 29425
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.600227935457794e-05,
+      "loss": 0.2896,
+      "step": 29430
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.600073394296544e-05,
+      "loss": 0.3128,
+      "step": 29435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5999188258672246e-05,
+      "loss": 0.3325,
+      "step": 29440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.599764230171844e-05,
+      "loss": 0.296,
+      "step": 29445
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.599609607212408e-05,
+      "loss": 0.2799,
+      "step": 29450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.599454956990925e-05,
+      "loss": 0.3323,
+      "step": 29455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.599300279509403e-05,
+      "loss": 0.3046,
+      "step": 29460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5991455747698505e-05,
+      "loss": 0.2976,
+      "step": 29465
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598990842774276e-05,
+      "loss": 0.2796,
+      "step": 29470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598836083524689e-05,
+      "loss": 0.2941,
+      "step": 29475
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598681297023099e-05,
+      "loss": 0.2621,
+      "step": 29480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598526483271514e-05,
+      "loss": 0.313,
+      "step": 29485
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598371642271947e-05,
+      "loss": 0.3199,
+      "step": 29490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598216774026406e-05,
+      "loss": 0.3125,
+      "step": 29495
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.598061878536904e-05,
+      "loss": 0.3023,
+      "step": 29500
+    },
+    {
+      "epoch": 1.04,
+      "eval_loss": 0.2920066714286804,
+      "eval_runtime": 10.538,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 29500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.597906955805451e-05,
+      "loss": 0.2885,
+      "step": 29505
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5977520058340586e-05,
+      "loss": 0.2782,
+      "step": 29510
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5975970286247384e-05,
+      "loss": 0.2928,
+      "step": 29515
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5974420241795034e-05,
+      "loss": 0.2862,
+      "step": 29520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.597286992500366e-05,
+      "loss": 0.2886,
+      "step": 29525
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.597131933589339e-05,
+      "loss": 0.2839,
+      "step": 29530
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5969768474484364e-05,
+      "loss": 0.2895,
+      "step": 29535
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596821734079672e-05,
+      "loss": 0.2762,
+      "step": 29540
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596666593485058e-05,
+      "loss": 0.3067,
+      "step": 29545
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5965114256666104e-05,
+      "loss": 0.2999,
+      "step": 29550
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5963562306263435e-05,
+      "loss": 0.2703,
+      "step": 29555
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5962010083662724e-05,
+      "loss": 0.2873,
+      "step": 29560
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.596045758888413e-05,
+      "loss": 0.3129,
+      "step": 29565
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5958904821947805e-05,
+      "loss": 0.303,
+      "step": 29570
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5957351782873916e-05,
+      "loss": 0.2996,
+      "step": 29575
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.595579847168263e-05,
+      "loss": 0.28,
+      "step": 29580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.59542448883941e-05,
+      "loss": 0.2625,
+      "step": 29585
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.595269103302852e-05,
+      "loss": 0.2995,
+      "step": 29590
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.595113690560605e-05,
+      "loss": 0.2965,
+      "step": 29595
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594958250614687e-05,
+      "loss": 0.2867,
+      "step": 29600
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594802783467117e-05,
+      "loss": 0.2938,
+      "step": 29605
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594647289119914e-05,
+      "loss": 0.2893,
+      "step": 29610
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594491767575096e-05,
+      "loss": 0.2957,
+      "step": 29615
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594336218834683e-05,
+      "loss": 0.26,
+      "step": 29620
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.594180642900694e-05,
+      "loss": 0.2826,
+      "step": 29625
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5940250397751494e-05,
+      "loss": 0.2641,
+      "step": 29630
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5938694094600697e-05,
+      "loss": 0.3104,
+      "step": 29635
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593713751957476e-05,
+      "loss": 0.2913,
+      "step": 29640
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593558067269389e-05,
+      "loss": 0.3053,
+      "step": 29645
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.59340235539783e-05,
+      "loss": 0.301,
+      "step": 29650
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593246616344821e-05,
+      "loss": 0.2842,
+      "step": 29655
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.593090850112384e-05,
+      "loss": 0.3104,
+      "step": 29660
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592935056702542e-05,
+      "loss": 0.2777,
+      "step": 29665
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592779236117318e-05,
+      "loss": 0.3036,
+      "step": 29670
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592623388358734e-05,
+      "loss": 0.3212,
+      "step": 29675
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592467513428815e-05,
+      "loss": 0.2755,
+      "step": 29680
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.592311611329584e-05,
+      "loss": 0.2793,
+      "step": 29685
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.5921556820630656e-05,
+      "loss": 0.2717,
+      "step": 29690
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.591999725631285e-05,
+      "loss": 0.2906,
+      "step": 29695
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.591843742036266e-05,
+      "loss": 0.3059,
+      "step": 29700
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.591687731280034e-05,
+      "loss": 0.3108,
+      "step": 29705
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.591531693364617e-05,
+      "loss": 0.2996,
+      "step": 29710
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.591375628292038e-05,
+      "loss": 0.315,
+      "step": 29715
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.591219536064325e-05,
+      "loss": 0.2997,
+      "step": 29720
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5910634166835045e-05,
+      "loss": 0.2891,
+      "step": 29725
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5909072701516035e-05,
+      "loss": 0.2988,
+      "step": 29730
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.59075109647065e-05,
+      "loss": 0.2868,
+      "step": 29735
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.590594895642671e-05,
+      "loss": 0.2998,
+      "step": 29740
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5904386676696955e-05,
+      "loss": 0.2672,
+      "step": 29745
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5902824125537516e-05,
+      "loss": 0.3028,
+      "step": 29750
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5901261302968676e-05,
+      "loss": 0.3042,
+      "step": 29755
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589969820901073e-05,
+      "loss": 0.2895,
+      "step": 29760
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589813484368398e-05,
+      "loss": 0.2965,
+      "step": 29765
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589657120700872e-05,
+      "loss": 0.2915,
+      "step": 29770
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589500729900525e-05,
+      "loss": 0.2911,
+      "step": 29775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5893443119693895e-05,
+      "loss": 0.2829,
+      "step": 29780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589187866909495e-05,
+      "loss": 0.2855,
+      "step": 29785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.589031394722871e-05,
+      "loss": 0.2959,
+      "step": 29790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588874895411552e-05,
+      "loss": 0.2871,
+      "step": 29795
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5887183689775694e-05,
+      "loss": 0.2746,
+      "step": 29800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588561815422955e-05,
+      "loss": 0.2942,
+      "step": 29805
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588405234749742e-05,
+      "loss": 0.3044,
+      "step": 29810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588248626959963e-05,
+      "loss": 0.2734,
+      "step": 29815
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.588091992055653e-05,
+      "loss": 0.2695,
+      "step": 29820
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.587935330038843e-05,
+      "loss": 0.3035,
+      "step": 29825
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.58777864091157e-05,
+      "loss": 0.3184,
+      "step": 29830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.587621924675866e-05,
+      "loss": 0.3016,
+      "step": 29835
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5874651813337674e-05,
+      "loss": 0.3051,
+      "step": 29840
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.58730841088731e-05,
+      "loss": 0.2728,
+      "step": 29845
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.587151613338528e-05,
+      "loss": 0.2969,
+      "step": 29850
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5869947886894577e-05,
+      "loss": 0.2884,
+      "step": 29855
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586837936942136e-05,
+      "loss": 0.2899,
+      "step": 29860
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586681058098598e-05,
+      "loss": 0.2928,
+      "step": 29865
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586524152160881e-05,
+      "loss": 0.2819,
+      "step": 29870
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586367219131025e-05,
+      "loss": 0.2921,
+      "step": 29875
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5862102590110644e-05,
+      "loss": 0.3056,
+      "step": 29880
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.586053271803039e-05,
+      "loss": 0.2808,
+      "step": 29885
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.585896257508986e-05,
+      "loss": 0.3111,
+      "step": 29890
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.585739216130945e-05,
+      "loss": 0.3143,
+      "step": 29895
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5855821476709546e-05,
+      "loss": 0.2987,
+      "step": 29900
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5854250521310556e-05,
+      "loss": 0.2665,
+      "step": 29905
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5852679295132853e-05,
+      "loss": 0.2807,
+      "step": 29910
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5851107798196854e-05,
+      "loss": 0.279,
+      "step": 29915
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584953603052297e-05,
+      "loss": 0.3015,
+      "step": 29920
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584796399213159e-05,
+      "loss": 0.3045,
+      "step": 29925
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584639168304315e-05,
+      "loss": 0.2951,
+      "step": 29930
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584481910327804e-05,
+      "loss": 0.2686,
+      "step": 29935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5843246252856697e-05,
+      "loss": 0.2661,
+      "step": 29940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584167313179954e-05,
+      "loss": 0.2954,
+      "step": 29945
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.584009974012699e-05,
+      "loss": 0.2725,
+      "step": 29950
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.583852607785948e-05,
+      "loss": 0.3023,
+      "step": 29955
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.583695214501743e-05,
+      "loss": 0.3105,
+      "step": 29960
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.58353779416213e-05,
+      "loss": 0.279,
+      "step": 29965
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5833803467691515e-05,
+      "loss": 0.3067,
+      "step": 29970
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.583222872324853e-05,
+      "loss": 0.2747,
+      "step": 29975
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.5830653708312775e-05,
+      "loss": 0.32,
+      "step": 29980
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.582907842290472e-05,
+      "loss": 0.3127,
+      "step": 29985
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.582750286704479e-05,
+      "loss": 0.2701,
+      "step": 29990
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.582592704075348e-05,
+      "loss": 0.2854,
+      "step": 29995
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.582435094405122e-05,
+      "loss": 0.3255,
+      "step": 30000
+    },
+    {
+      "epoch": 1.06,
+      "eval_loss": 0.29047587513923645,
+      "eval_runtime": 10.5326,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 30000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.582277457695849e-05,
+      "loss": 0.3098,
+      "step": 30005
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.582119793949576e-05,
+      "loss": 0.2848,
+      "step": 30010
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.581962103168349e-05,
+      "loss": 0.2573,
+      "step": 30015
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.581804385354217e-05,
+      "loss": 0.3003,
+      "step": 30020
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5816466405092265e-05,
+      "loss": 0.2979,
+      "step": 30025
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5814888686354254e-05,
+      "loss": 0.2854,
+      "step": 30030
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5813310697348644e-05,
+      "loss": 0.3119,
+      "step": 30035
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.58117324380959e-05,
+      "loss": 0.2668,
+      "step": 30040
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5810153908616534e-05,
+      "loss": 0.3009,
+      "step": 30045
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.580857510893103e-05,
+      "loss": 0.2783,
+      "step": 30050
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5806996039059894e-05,
+      "loss": 0.2799,
+      "step": 30055
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.580541669902362e-05,
+      "loss": 0.3114,
+      "step": 30060
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5803837088842736e-05,
+      "loss": 0.2727,
+      "step": 30065
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5802257208537725e-05,
+      "loss": 0.2785,
+      "step": 30070
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5800677058129114e-05,
+      "loss": 0.2772,
+      "step": 30075
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579909663763742e-05,
+      "loss": 0.2984,
+      "step": 30080
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579751594708317e-05,
+      "loss": 0.3128,
+      "step": 30085
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579593498648688e-05,
+      "loss": 0.3048,
+      "step": 30090
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5794353755869076e-05,
+      "loss": 0.2876,
+      "step": 30095
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5792772255250285e-05,
+      "loss": 0.3073,
+      "step": 30100
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.579119048465106e-05,
+      "loss": 0.3059,
+      "step": 30105
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578960844409192e-05,
+      "loss": 0.2862,
+      "step": 30110
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5788026133593434e-05,
+      "loss": 0.2938,
+      "step": 30115
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578644355317611e-05,
+      "loss": 0.2748,
+      "step": 30120
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578486070286052e-05,
+      "loss": 0.3168,
+      "step": 30125
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578327758266721e-05,
+      "loss": 0.2866,
+      "step": 30130
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578169419261674e-05,
+      "loss": 0.293,
+      "step": 30135
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.578011053272967e-05,
+      "loss": 0.2965,
+      "step": 30140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577852660302655e-05,
+      "loss": 0.2518,
+      "step": 30145
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577694240352797e-05,
+      "loss": 0.2861,
+      "step": 30150
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577535793425447e-05,
+      "loss": 0.2965,
+      "step": 30155
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5773773195226646e-05,
+      "loss": 0.2779,
+      "step": 30160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.577218818646506e-05,
+      "loss": 0.2751,
+      "step": 30165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5770602907990314e-05,
+      "loss": 0.298,
+      "step": 30170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5769017359822965e-05,
+      "loss": 0.3175,
+      "step": 30175
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5767431541983616e-05,
+      "loss": 0.3072,
+      "step": 30180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.576584545449286e-05,
+      "loss": 0.2959,
+      "step": 30185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.576425909737128e-05,
+      "loss": 0.2951,
+      "step": 30190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.576267247063948e-05,
+      "loss": 0.2759,
+      "step": 30195
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.576108557431806e-05,
+      "loss": 0.2547,
+      "step": 30200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5759498408427634e-05,
+      "loss": 0.3103,
+      "step": 30205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.57579109729888e-05,
+      "loss": 0.2763,
+      "step": 30210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575632326802217e-05,
+      "loss": 0.2865,
+      "step": 30215
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5754735293548355e-05,
+      "loss": 0.2729,
+      "step": 30220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575314704958799e-05,
+      "loss": 0.2916,
+      "step": 30225
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.575155853616168e-05,
+      "loss": 0.3036,
+      "step": 30230
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5749969753290065e-05,
+      "loss": 0.2789,
+      "step": 30235
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574838070099376e-05,
+      "loss": 0.2888,
+      "step": 30240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574679137929341e-05,
+      "loss": 0.3114,
+      "step": 30245
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5745201788209646e-05,
+      "loss": 0.3183,
+      "step": 30250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5743611927763106e-05,
+      "loss": 0.2891,
+      "step": 30255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.5742021797974447e-05,
+      "loss": 0.3024,
+      "step": 30260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.574043139886429e-05,
+      "loss": 0.3048,
+      "step": 30265
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.57388407304533e-05,
+      "loss": 0.3093,
+      "step": 30270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.573724979276214e-05,
+      "loss": 0.2744,
+      "step": 30275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.573565858581145e-05,
+      "loss": 0.3016,
+      "step": 30280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.57340671096219e-05,
+      "loss": 0.2876,
+      "step": 30285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.573247536421415e-05,
+      "loss": 0.2879,
+      "step": 30290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5730883349608885e-05,
+      "loss": 0.3148,
+      "step": 30295
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.572929106582675e-05,
+      "loss": 0.3076,
+      "step": 30300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.572769851288843e-05,
+      "loss": 0.2992,
+      "step": 30305
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5726105690814616e-05,
+      "loss": 0.2713,
+      "step": 30310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5724512599625965e-05,
+      "loss": 0.311,
+      "step": 30315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.572291923934319e-05,
+      "loss": 0.276,
+      "step": 30320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5721325609986956e-05,
+      "loss": 0.2811,
+      "step": 30325
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571973171157796e-05,
+      "loss": 0.3065,
+      "step": 30330
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5718137544136906e-05,
+      "loss": 0.2823,
+      "step": 30335
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.57165431076845e-05,
+      "loss": 0.2673,
+      "step": 30340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571494840224143e-05,
+      "loss": 0.3034,
+      "step": 30345
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.57133534278284e-05,
+      "loss": 0.2886,
+      "step": 30350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.571175818446613e-05,
+      "loss": 0.3097,
+      "step": 30355
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5710162672175326e-05,
+      "loss": 0.301,
+      "step": 30360
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.570856689097672e-05,
+      "loss": 0.3069,
+      "step": 30365
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.570697084089101e-05,
+      "loss": 0.2981,
+      "step": 30370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.570537452193893e-05,
+      "loss": 0.321,
+      "step": 30375
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.570377793414121e-05,
+      "loss": 0.3171,
+      "step": 30380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.570218107751858e-05,
+      "loss": 0.3056,
+      "step": 30385
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5700583952091766e-05,
+      "loss": 0.2731,
+      "step": 30390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.569898655788152e-05,
+      "loss": 0.2996,
+      "step": 30395
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5697388894908565e-05,
+      "loss": 0.3146,
+      "step": 30400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5695790963193666e-05,
+      "loss": 0.2897,
+      "step": 30405
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.569419276275755e-05,
+      "loss": 0.2906,
+      "step": 30410
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.569259429362098e-05,
+      "loss": 0.2822,
+      "step": 30415
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.569131532486135e-05,
+      "loss": 0.3105,
+      "step": 30420
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5689716372116266e-05,
+      "loss": 0.2918,
+      "step": 30425
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5688117150728856e-05,
+      "loss": 0.2943,
+      "step": 30430
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5686517660719875e-05,
+      "loss": 0.3087,
+      "step": 30435
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.56849179021101e-05,
+      "loss": 0.2929,
+      "step": 30440
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.568331787492029e-05,
+      "loss": 0.2754,
+      "step": 30445
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.568171757917123e-05,
+      "loss": 0.3089,
+      "step": 30450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.568011701488371e-05,
+      "loss": 0.289,
+      "step": 30455
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567851618207848e-05,
+      "loss": 0.281,
+      "step": 30460
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567691508077635e-05,
+      "loss": 0.2843,
+      "step": 30465
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5675313710998116e-05,
+      "loss": 0.2716,
+      "step": 30470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5673712072764544e-05,
+      "loss": 0.2786,
+      "step": 30475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567211016609645e-05,
+      "loss": 0.2728,
+      "step": 30480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.567050799101463e-05,
+      "loss": 0.2905,
+      "step": 30485
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.566890554753989e-05,
+      "loss": 0.2941,
+      "step": 30490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5667302835693034e-05,
+      "loss": 0.2961,
+      "step": 30495
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5665699855494863e-05,
+      "loss": 0.3006,
+      "step": 30500
+    },
+    {
+      "epoch": 1.07,
+      "eval_loss": 0.2904548943042755,
+      "eval_runtime": 10.5425,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 30500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.56640966069662e-05,
+      "loss": 0.2878,
+      "step": 30505
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.566249309012787e-05,
+      "loss": 0.3108,
+      "step": 30510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.566088930500068e-05,
+      "loss": 0.3057,
+      "step": 30515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565928525160545e-05,
+      "loss": 0.3067,
+      "step": 30520
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565768092996302e-05,
+      "loss": 0.2811,
+      "step": 30525
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5656076340094224e-05,
+      "loss": 0.2812,
+      "step": 30530
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.5654471482019876e-05,
+      "loss": 0.2949,
+      "step": 30535
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565286635576084e-05,
+      "loss": 0.2894,
+      "step": 30540
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.565126096133794e-05,
+      "loss": 0.288,
+      "step": 30545
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.564965529877203e-05,
+      "loss": 0.3034,
+      "step": 30550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5648049368083956e-05,
+      "loss": 0.2972,
+      "step": 30555
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.564644316929456e-05,
+      "loss": 0.2935,
+      "step": 30560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.564483670242471e-05,
+      "loss": 0.2799,
+      "step": 30565
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.564322996749526e-05,
+      "loss": 0.2615,
+      "step": 30570
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5641622964527076e-05,
+      "loss": 0.3086,
+      "step": 30575
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.564001569354102e-05,
+      "loss": 0.3051,
+      "step": 30580
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563840815455797e-05,
+      "loss": 0.2889,
+      "step": 30585
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563680034759877e-05,
+      "loss": 0.2883,
+      "step": 30590
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563519227268433e-05,
+      "loss": 0.2804,
+      "step": 30595
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563358392983553e-05,
+      "loss": 0.2779,
+      "step": 30600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563197531907323e-05,
+      "loss": 0.2925,
+      "step": 30605
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.563036644041833e-05,
+      "loss": 0.312,
+      "step": 30610
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.56287572938917e-05,
+      "loss": 0.3245,
+      "step": 30615
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.562714787951427e-05,
+      "loss": 0.303,
+      "step": 30620
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5625538197306916e-05,
+      "loss": 0.2757,
+      "step": 30625
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5623928247290536e-05,
+      "loss": 0.2908,
+      "step": 30630
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5622318029486045e-05,
+      "loss": 0.2996,
+      "step": 30635
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.562070754391434e-05,
+      "loss": 0.2964,
+      "step": 30640
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561909679059635e-05,
+      "loss": 0.2906,
+      "step": 30645
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561748576955297e-05,
+      "loss": 0.2945,
+      "step": 30650
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5615874480805114e-05,
+      "loss": 0.2942,
+      "step": 30655
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561426292437372e-05,
+      "loss": 0.2847,
+      "step": 30660
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561265110027971e-05,
+      "loss": 0.2865,
+      "step": 30665
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.561103900854401e-05,
+      "loss": 0.2858,
+      "step": 30670
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.560942664918755e-05,
+      "loss": 0.2842,
+      "step": 30675
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5607814022231266e-05,
+      "loss": 0.3133,
+      "step": 30680
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5606201127696105e-05,
+      "loss": 0.2843,
+      "step": 30685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5604587965602996e-05,
+      "loss": 0.3128,
+      "step": 30690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.560297453597288e-05,
+      "loss": 0.2892,
+      "step": 30695
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.560136083882674e-05,
+      "loss": 0.3015,
+      "step": 30700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559974687418549e-05,
+      "loss": 0.3036,
+      "step": 30705
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559813264207011e-05,
+      "loss": 0.2806,
+      "step": 30710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5596518142501556e-05,
+      "loss": 0.289,
+      "step": 30715
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5594903375500786e-05,
+      "loss": 0.2716,
+      "step": 30720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559328834108876e-05,
+      "loss": 0.2837,
+      "step": 30725
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559167303928646e-05,
+      "loss": 0.2884,
+      "step": 30730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.559005747011486e-05,
+      "loss": 0.2818,
+      "step": 30735
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5588441633594926e-05,
+      "loss": 0.304,
+      "step": 30740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.558682552974765e-05,
+      "loss": 0.306,
+      "step": 30745
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5585209158594014e-05,
+      "loss": 0.2925,
+      "step": 30750
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.558359252015499e-05,
+      "loss": 0.2952,
+      "step": 30755
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.55819756144516e-05,
+      "loss": 0.2705,
+      "step": 30760
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.558035844150481e-05,
+      "loss": 0.2831,
+      "step": 30765
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.557874100133563e-05,
+      "loss": 0.2785,
+      "step": 30770
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.557712329396507e-05,
+      "loss": 0.3199,
+      "step": 30775
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5575505319414117e-05,
+      "loss": 0.3042,
+      "step": 30780
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5573887077703786e-05,
+      "loss": 0.291,
+      "step": 30785
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.557226856885509e-05,
+      "loss": 0.2976,
+      "step": 30790
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.557064979288905e-05,
+      "loss": 0.3074,
+      "step": 30795
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.5569030749826666e-05,
+      "loss": 0.28,
+      "step": 30800
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.556741143968899e-05,
+      "loss": 0.2845,
+      "step": 30805
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.556579186249702e-05,
+      "loss": 0.303,
+      "step": 30810
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.55641720182718e-05,
+      "loss": 0.3016,
+      "step": 30815
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.556255190703436e-05,
+      "loss": 0.2929,
+      "step": 30820
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.556093152880573e-05,
+      "loss": 0.28,
+      "step": 30825
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.555931088360695e-05,
+      "loss": 0.3086,
+      "step": 30830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.555768997145908e-05,
+      "loss": 0.3224,
+      "step": 30835
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.555606879238314e-05,
+      "loss": 0.3191,
+      "step": 30840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.555444734640021e-05,
+      "loss": 0.2812,
+      "step": 30845
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.555282563353132e-05,
+      "loss": 0.2703,
+      "step": 30850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5551203653797525e-05,
+      "loss": 0.2794,
+      "step": 30855
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.55495814072199e-05,
+      "loss": 0.287,
+      "step": 30860
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554795889381951e-05,
+      "loss": 0.2831,
+      "step": 30865
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554633611361741e-05,
+      "loss": 0.2908,
+      "step": 30870
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554471306663467e-05,
+      "loss": 0.2641,
+      "step": 30875
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5543089752892375e-05,
+      "loss": 0.3155,
+      "step": 30880
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.554146617241161e-05,
+      "loss": 0.2905,
+      "step": 30885
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553984232521342e-05,
+      "loss": 0.3115,
+      "step": 30890
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553821821131893e-05,
+      "loss": 0.3043,
+      "step": 30895
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553659383074921e-05,
+      "loss": 0.2819,
+      "step": 30900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5534969183525354e-05,
+      "loss": 0.2957,
+      "step": 30905
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553334426966846e-05,
+      "loss": 0.2755,
+      "step": 30910
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.553171908919961e-05,
+      "loss": 0.2928,
+      "step": 30915
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5530093642139927e-05,
+      "loss": 0.2944,
+      "step": 30920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.55284679285105e-05,
+      "loss": 0.2891,
+      "step": 30925
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5526841948332455e-05,
+      "loss": 0.2939,
+      "step": 30930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.55252157016269e-05,
+      "loss": 0.2685,
+      "step": 30935
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5523589188414926e-05,
+      "loss": 0.2761,
+      "step": 30940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.552196240871769e-05,
+      "loss": 0.2695,
+      "step": 30945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5520335362556286e-05,
+      "loss": 0.2848,
+      "step": 30950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551870804995186e-05,
+      "loss": 0.2786,
+      "step": 30955
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551708047092552e-05,
+      "loss": 0.2955,
+      "step": 30960
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551545262549842e-05,
+      "loss": 0.2877,
+      "step": 30965
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5513824513691686e-05,
+      "loss": 0.2801,
+      "step": 30970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551219613552646e-05,
+      "loss": 0.2774,
+      "step": 30975
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.551056749102389e-05,
+      "loss": 0.2859,
+      "step": 30980
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550893858020511e-05,
+      "loss": 0.2759,
+      "step": 30985
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550730940309128e-05,
+      "loss": 0.3017,
+      "step": 30990
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5505679959703565e-05,
+      "loss": 0.3095,
+      "step": 30995
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550405025006309e-05,
+      "loss": 0.2957,
+      "step": 31000
+    },
+    {
+      "epoch": 1.09,
+      "eval_loss": 0.2897602617740631,
+      "eval_runtime": 10.5439,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 31000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550242027419105e-05,
+      "loss": 0.285,
+      "step": 31005
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.550079003210859e-05,
+      "loss": 0.2983,
+      "step": 31010
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5499159523836875e-05,
+      "loss": 0.2819,
+      "step": 31015
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.549752874939709e-05,
+      "loss": 0.3313,
+      "step": 31020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5495897708810395e-05,
+      "loss": 0.266,
+      "step": 31025
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.549426640209799e-05,
+      "loss": 0.3073,
+      "step": 31030
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5492634829281026e-05,
+      "loss": 0.3227,
+      "step": 31035
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.549100299038072e-05,
+      "loss": 0.2894,
+      "step": 31040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548937088541823e-05,
+      "loss": 0.2956,
+      "step": 31045
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548773851441477e-05,
+      "loss": 0.2959,
+      "step": 31050
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548610587739152e-05,
+      "loss": 0.2917,
+      "step": 31055
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.54844729743697e-05,
+      "loss": 0.3038,
+      "step": 31060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.548283980537048e-05,
+      "loss": 0.2948,
+      "step": 31065
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5481206370415094e-05,
+      "loss": 0.2723,
+      "step": 31070
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.547989943097659e-05,
+      "loss": 0.2976,
+      "step": 31075
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5478265517353534e-05,
+      "loss": 0.3196,
+      "step": 31080
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.54766313378337e-05,
+      "loss": 0.2942,
+      "step": 31085
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5474996892438294e-05,
+      "loss": 0.2788,
+      "step": 31090
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5473362181188554e-05,
+      "loss": 0.269,
+      "step": 31095
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5471727204105695e-05,
+      "loss": 0.2889,
+      "step": 31100
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5470091961210955e-05,
+      "loss": 0.268,
+      "step": 31105
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.5468456452525556e-05,
+      "loss": 0.2981,
+      "step": 31110
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.546682067807074e-05,
+      "loss": 0.2968,
+      "step": 31115
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.546518463786775e-05,
+      "loss": 0.2593,
+      "step": 31120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.546354833193782e-05,
+      "loss": 0.3151,
+      "step": 31125
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.54619117603022e-05,
+      "loss": 0.2953,
+      "step": 31130
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.546027492298215e-05,
+      "loss": 0.2878,
+      "step": 31135
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.54586378199989e-05,
+      "loss": 0.2767,
+      "step": 31140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.545700045137373e-05,
+      "loss": 0.2915,
+      "step": 31145
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.545536281712789e-05,
+      "loss": 0.2771,
+      "step": 31150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5453724917282635e-05,
+      "loss": 0.3109,
+      "step": 31155
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5452086751859254e-05,
+      "loss": 0.2773,
+      "step": 31160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5450448320879004e-05,
+      "loss": 0.283,
+      "step": 31165
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544880962436315e-05,
+      "loss": 0.2982,
+      "step": 31170
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544717066233298e-05,
+      "loss": 0.2781,
+      "step": 31175
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544553143480978e-05,
+      "loss": 0.2913,
+      "step": 31180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544389194181483e-05,
+      "loss": 0.28,
+      "step": 31185
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544225218336942e-05,
+      "loss": 0.2506,
+      "step": 31190
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.544061215949483e-05,
+      "loss": 0.2924,
+      "step": 31195
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543897187021236e-05,
+      "loss": 0.2654,
+      "step": 31200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543733131554332e-05,
+      "loss": 0.2989,
+      "step": 31205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5435690495508984e-05,
+      "loss": 0.2829,
+      "step": 31210
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543404941013069e-05,
+      "loss": 0.2898,
+      "step": 31215
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.543240805942973e-05,
+      "loss": 0.2847,
+      "step": 31220
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5430766443427405e-05,
+      "loss": 0.3066,
+      "step": 31225
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542912456214505e-05,
+      "loss": 0.3007,
+      "step": 31230
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542748241560398e-05,
+      "loss": 0.2982,
+      "step": 31235
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542584000382552e-05,
+      "loss": 0.3101,
+      "step": 31240
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542419732683097e-05,
+      "loss": 0.2978,
+      "step": 31245
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542255438464169e-05,
+      "loss": 0.2544,
+      "step": 31250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.542091117727899e-05,
+      "loss": 0.3042,
+      "step": 31255
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5419267704764227e-05,
+      "loss": 0.3146,
+      "step": 31260
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541762396711872e-05,
+      "loss": 0.3158,
+      "step": 31265
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5415979964363825e-05,
+      "loss": 0.2852,
+      "step": 31270
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541433569652089e-05,
+      "loss": 0.2918,
+      "step": 31275
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541269116361125e-05,
+      "loss": 0.3133,
+      "step": 31280
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.541104636565627e-05,
+      "loss": 0.2998,
+      "step": 31285
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540940130267731e-05,
+      "loss": 0.2574,
+      "step": 31290
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5407755974695726e-05,
+      "loss": 0.305,
+      "step": 31295
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540611038173288e-05,
+      "loss": 0.2872,
+      "step": 31300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540446452381013e-05,
+      "loss": 0.2743,
+      "step": 31305
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540281840094886e-05,
+      "loss": 0.2836,
+      "step": 31310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.540117201317044e-05,
+      "loss": 0.2902,
+      "step": 31315
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.539952536049624e-05,
+      "loss": 0.2857,
+      "step": 31320
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.539787844294766e-05,
+      "loss": 0.2743,
+      "step": 31325
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.539623126054606e-05,
+      "loss": 0.2676,
+      "step": 31330
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.539458381331284e-05,
+      "loss": 0.2786,
+      "step": 31335
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.53929361012694e-05,
+      "loss": 0.2938,
+      "step": 31340
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5391288124437116e-05,
+      "loss": 0.3097,
+      "step": 31345
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5389639882837386e-05,
+      "loss": 0.2933,
+      "step": 31350
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.538799137649163e-05,
+      "loss": 0.2772,
+      "step": 31355
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.538634260542124e-05,
+      "loss": 0.2808,
+      "step": 31360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.538469356964763e-05,
+      "loss": 0.3264,
+      "step": 31365
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5383044269192196e-05,
+      "loss": 0.2675,
+      "step": 31370
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.5381394704076374e-05,
+      "loss": 0.3035,
+      "step": 31375
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537974487432157e-05,
+      "loss": 0.3087,
+      "step": 31380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537809477994922e-05,
+      "loss": 0.3148,
+      "step": 31385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537644442098073e-05,
+      "loss": 0.3031,
+      "step": 31390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537479379743754e-05,
+      "loss": 0.2633,
+      "step": 31395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537314290934108e-05,
+      "loss": 0.2997,
+      "step": 31400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.537149175671278e-05,
+      "loss": 0.3034,
+      "step": 31405
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5369840339574096e-05,
+      "loss": 0.2862,
+      "step": 31410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.536818865794645e-05,
+      "loss": 0.2872,
+      "step": 31415
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.53665367118513e-05,
+      "loss": 0.2752,
+      "step": 31420
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5364884501310086e-05,
+      "loss": 0.3041,
+      "step": 31425
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5363232026344284e-05,
+      "loss": 0.2794,
+      "step": 31430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.536157928697532e-05,
+      "loss": 0.295,
+      "step": 31435
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5359926283224674e-05,
+      "loss": 0.2901,
+      "step": 31440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.535827301511381e-05,
+      "loss": 0.2754,
+      "step": 31445
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.535661948266418e-05,
+      "loss": 0.2856,
+      "step": 31450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5354965685897255e-05,
+      "loss": 0.2911,
+      "step": 31455
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5353311624834516e-05,
+      "loss": 0.2583,
+      "step": 31460
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5351657299497444e-05,
+      "loss": 0.2752,
+      "step": 31465
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.535000270990751e-05,
+      "loss": 0.2796,
+      "step": 31470
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5348347856086214e-05,
+      "loss": 0.2898,
+      "step": 31475
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.534669273805502e-05,
+      "loss": 0.2915,
+      "step": 31480
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.534503735583543e-05,
+      "loss": 0.2943,
+      "step": 31485
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5343381709448943e-05,
+      "loss": 0.2976,
+      "step": 31490
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.534172579891705e-05,
+      "loss": 0.3062,
+      "step": 31495
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.534006962426125e-05,
+      "loss": 0.2871,
+      "step": 31500
+    },
+    {
+      "epoch": 1.11,
+      "eval_loss": 0.28987738490104675,
+      "eval_runtime": 10.5447,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 31500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5338413185503056e-05,
+      "loss": 0.2818,
+      "step": 31505
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533675648266397e-05,
+      "loss": 0.2519,
+      "step": 31510
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533509951576549e-05,
+      "loss": 0.2737,
+      "step": 31515
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5333442284829164e-05,
+      "loss": 0.3031,
+      "step": 31520
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533178478987647e-05,
+      "loss": 0.3129,
+      "step": 31525
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.533012703092897e-05,
+      "loss": 0.2801,
+      "step": 31530
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5328469008008154e-05,
+      "loss": 0.2885,
+      "step": 31535
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.532681072113557e-05,
+      "loss": 0.2672,
+      "step": 31540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.532515217033275e-05,
+      "loss": 0.3211,
+      "step": 31545
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5323493355621214e-05,
+      "loss": 0.3024,
+      "step": 31550
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5321834277022515e-05,
+      "loss": 0.3081,
+      "step": 31555
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.532017493455819e-05,
+      "loss": 0.3033,
+      "step": 31560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531851532824979e-05,
+      "loss": 0.3097,
+      "step": 31565
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531685545811885e-05,
+      "loss": 0.2925,
+      "step": 31570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5315195324186935e-05,
+      "loss": 0.2856,
+      "step": 31575
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.53135349264756e-05,
+      "loss": 0.2676,
+      "step": 31580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.531187426500639e-05,
+      "loss": 0.279,
+      "step": 31585
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5310213339800895e-05,
+      "loss": 0.3094,
+      "step": 31590
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5308552150880655e-05,
+      "loss": 0.2832,
+      "step": 31595
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5306890698267254e-05,
+      "loss": 0.2985,
+      "step": 31600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5305228981982254e-05,
+      "loss": 0.2949,
+      "step": 31605
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5303567002047246e-05,
+      "loss": 0.2908,
+      "step": 31610
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.530190475848379e-05,
+      "loss": 0.3151,
+      "step": 31615
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.530024225131349e-05,
+      "loss": 0.2759,
+      "step": 31620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.529857948055791e-05,
+      "loss": 0.2868,
+      "step": 31625
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.529691644623866e-05,
+      "loss": 0.2885,
+      "step": 31630
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5295253148377334e-05,
+      "loss": 0.2702,
+      "step": 31635
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.529358958699551e-05,
+      "loss": 0.2905,
+      "step": 31640
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5291925762114805e-05,
+      "loss": 0.3137,
+      "step": 31645
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.529026167375681e-05,
+      "loss": 0.2886,
+      "step": 31650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5288597321943146e-05,
+      "loss": 0.3019,
+      "step": 31655
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.528693270669542e-05,
+      "loss": 0.2911,
+      "step": 31660
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.528526782803523e-05,
+      "loss": 0.2665,
+      "step": 31665
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5283602685984215e-05,
+      "loss": 0.2793,
+      "step": 31670
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.528193728056398e-05,
+      "loss": 0.2908,
+      "step": 31675
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.5280271611796165e-05,
+      "loss": 0.2861,
+      "step": 31680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.527860567970238e-05,
+      "loss": 0.301,
+      "step": 31685
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.527693948430427e-05,
+      "loss": 0.2825,
+      "step": 31690
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.527527302562346e-05,
+      "loss": 0.298,
+      "step": 31695
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5273606303681596e-05,
+      "loss": 0.2914,
+      "step": 31700
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.527193931850031e-05,
+      "loss": 0.2769,
+      "step": 31705
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.527027207010125e-05,
+      "loss": 0.3404,
+      "step": 31710
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5268604558506075e-05,
+      "loss": 0.2766,
+      "step": 31715
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.526693678373642e-05,
+      "loss": 0.2803,
+      "step": 31720
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.526526874581395e-05,
+      "loss": 0.2841,
+      "step": 31725
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5263600444760316e-05,
+      "loss": 0.2872,
+      "step": 31730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.526193188059718e-05,
+      "loss": 0.2733,
+      "step": 31735
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.526026305334623e-05,
+      "loss": 0.2901,
+      "step": 31740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.52585939630291e-05,
+      "loss": 0.2933,
+      "step": 31745
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5256924609667484e-05,
+      "loss": 0.3002,
+      "step": 31750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.525525499328306e-05,
+      "loss": 0.2962,
+      "step": 31755
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.525358511389749e-05,
+      "loss": 0.2934,
+      "step": 31760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.525191497153246e-05,
+      "loss": 0.3064,
+      "step": 31765
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5250244566209666e-05,
+      "loss": 0.2713,
+      "step": 31770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.52485738979508e-05,
+      "loss": 0.3043,
+      "step": 31775
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5246902966777535e-05,
+      "loss": 0.2654,
+      "step": 31780
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.524523177271158e-05,
+      "loss": 0.2602,
+      "step": 31785
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5243560315774634e-05,
+      "loss": 0.2898,
+      "step": 31790
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.52418885959884e-05,
+      "loss": 0.2941,
+      "step": 31795
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.524021661337458e-05,
+      "loss": 0.3295,
+      "step": 31800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5238544367954884e-05,
+      "loss": 0.2934,
+      "step": 31805
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5236871859751026e-05,
+      "loss": 0.2955,
+      "step": 31810
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.523519908878473e-05,
+      "loss": 0.2837,
+      "step": 31815
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5233526055077705e-05,
+      "loss": 0.3061,
+      "step": 31820
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5231852758651685e-05,
+      "loss": 0.2806,
+      "step": 31825
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5230179199528384e-05,
+      "loss": 0.3167,
+      "step": 31830
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5228505377729534e-05,
+      "loss": 0.2961,
+      "step": 31835
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522683129327688e-05,
+      "loss": 0.3087,
+      "step": 31840
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522515694619215e-05,
+      "loss": 0.2866,
+      "step": 31845
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522348233649709e-05,
+      "loss": 0.2903,
+      "step": 31850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522180746421343e-05,
+      "loss": 0.2939,
+      "step": 31855
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.522013232936293e-05,
+      "loss": 0.2891,
+      "step": 31860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521845693196735e-05,
+      "loss": 0.3175,
+      "step": 31865
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5216781272048414e-05,
+      "loss": 0.3182,
+      "step": 31870
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5215105349627906e-05,
+      "loss": 0.2788,
+      "step": 31875
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5213429164727575e-05,
+      "loss": 0.2906,
+      "step": 31880
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521175271736918e-05,
+      "loss": 0.2794,
+      "step": 31885
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.521007600757451e-05,
+      "loss": 0.2859,
+      "step": 31890
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520839903536531e-05,
+      "loss": 0.2792,
+      "step": 31895
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520672180076337e-05,
+      "loss": 0.2994,
+      "step": 31900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520504430379047e-05,
+      "loss": 0.2815,
+      "step": 31905
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.520336654446838e-05,
+      "loss": 0.2966,
+      "step": 31910
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5201688522818895e-05,
+      "loss": 0.3142,
+      "step": 31915
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5200010238863794e-05,
+      "loss": 0.2811,
+      "step": 31920
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5198331692624874e-05,
+      "loss": 0.2741,
+      "step": 31925
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5196652884123926e-05,
+      "loss": 0.3182,
+      "step": 31930
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5194973813382744e-05,
+      "loss": 0.2964,
+      "step": 31935
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5193294480423146e-05,
+      "loss": 0.3003,
+      "step": 31940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5191614885266925e-05,
+      "loss": 0.2881,
+      "step": 31945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.5189935027935894e-05,
+      "loss": 0.2914,
+      "step": 31950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518825490845185e-05,
+      "loss": 0.2768,
+      "step": 31955
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518657452683663e-05,
+      "loss": 0.2805,
+      "step": 31960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518489388311205e-05,
+      "loss": 0.3325,
+      "step": 31965
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518321297729991e-05,
+      "loss": 0.3182,
+      "step": 31970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.518153180942206e-05,
+      "loss": 0.2958,
+      "step": 31975
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.517985037950032e-05,
+      "loss": 0.2981,
+      "step": 31980
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5178168687556525e-05,
+      "loss": 0.2792,
+      "step": 31985
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5176486733612497e-05,
+      "loss": 0.2914,
+      "step": 31990
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.517480451769009e-05,
+      "loss": 0.3029,
+      "step": 31995
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.517312203981114e-05,
+      "loss": 0.2882,
+      "step": 32000
+    },
+    {
+      "epoch": 1.13,
+      "eval_loss": 0.2893707752227783,
+      "eval_runtime": 10.5373,
+      "eval_samples_per_second": 9.49,
+      "eval_steps_per_second": 9.49,
+      "step": 32000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51714392999975e-05,
+      "loss": 0.2729,
+      "step": 32005
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.516975629827101e-05,
+      "loss": 0.3085,
+      "step": 32010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.516807303465353e-05,
+      "loss": 0.2839,
+      "step": 32015
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.516638950916691e-05,
+      "loss": 0.2747,
+      "step": 32020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.516470572183302e-05,
+      "loss": 0.2748,
+      "step": 32025
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5163021672673714e-05,
+      "loss": 0.2887,
+      "step": 32030
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.516133736171084e-05,
+      "loss": 0.3063,
+      "step": 32035
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5159652788966314e-05,
+      "loss": 0.2861,
+      "step": 32040
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5157967954461967e-05,
+      "loss": 0.2965,
+      "step": 32045
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.515628285821969e-05,
+      "loss": 0.3038,
+      "step": 32050
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.515459750026137e-05,
+      "loss": 0.288,
+      "step": 32055
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.515291188060888e-05,
+      "loss": 0.2806,
+      "step": 32060
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5151225999284115e-05,
+      "loss": 0.3051,
+      "step": 32065
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.514953985630895e-05,
+      "loss": 0.3021,
+      "step": 32070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51478534517053e-05,
+      "loss": 0.2987,
+      "step": 32075
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.514616678549505e-05,
+      "loss": 0.2964,
+      "step": 32080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51444798577001e-05,
+      "loss": 0.273,
+      "step": 32085
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.514279266834235e-05,
+      "loss": 0.2763,
+      "step": 32090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.514110521744371e-05,
+      "loss": 0.2772,
+      "step": 32095
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5139417505026105e-05,
+      "loss": 0.2844,
+      "step": 32100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.513772953111143e-05,
+      "loss": 0.2938,
+      "step": 32105
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5136041295721606e-05,
+      "loss": 0.3007,
+      "step": 32110
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5134352798878556e-05,
+      "loss": 0.3043,
+      "step": 32115
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51326640406042e-05,
+      "loss": 0.2907,
+      "step": 32120
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.513097502092047e-05,
+      "loss": 0.3222,
+      "step": 32125
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5129285739849304e-05,
+      "loss": 0.3052,
+      "step": 32130
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.512759619741262e-05,
+      "loss": 0.2914,
+      "step": 32135
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5125906393632376e-05,
+      "loss": 0.3115,
+      "step": 32140
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.512421632853049e-05,
+      "loss": 0.2945,
+      "step": 32145
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5122526002128924e-05,
+      "loss": 0.3217,
+      "step": 32150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5120835414449615e-05,
+      "loss": 0.3123,
+      "step": 32155
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5119144565514514e-05,
+      "loss": 0.2885,
+      "step": 32160
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5117453455345594e-05,
+      "loss": 0.2797,
+      "step": 32165
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5115762083964786e-05,
+      "loss": 0.2813,
+      "step": 32170
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.511407045139406e-05,
+      "loss": 0.3063,
+      "step": 32175
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51123785576554e-05,
+      "loss": 0.2874,
+      "step": 32180
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.511068640277075e-05,
+      "loss": 0.2988,
+      "step": 32185
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5108993986762096e-05,
+      "loss": 0.2857,
+      "step": 32190
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5107301309651404e-05,
+      "loss": 0.2855,
+      "step": 32195
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5105608371460654e-05,
+      "loss": 0.3222,
+      "step": 32200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.510391517221183e-05,
+      "loss": 0.289,
+      "step": 32205
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.510222171192692e-05,
+      "loss": 0.2758,
+      "step": 32210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.51005279906279e-05,
+      "loss": 0.3126,
+      "step": 32215
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.509883400833678e-05,
+      "loss": 0.2726,
+      "step": 32220
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.509713976507554e-05,
+      "loss": 0.2743,
+      "step": 32225
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.509544526086619e-05,
+      "loss": 0.2863,
+      "step": 32230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.509375049573072e-05,
+      "loss": 0.2754,
+      "step": 32235
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5092055469691145e-05,
+      "loss": 0.3209,
+      "step": 32240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.509036018276948e-05,
+      "loss": 0.2833,
+      "step": 32245
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.508866463498772e-05,
+      "loss": 0.2931,
+      "step": 32250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.508696882636788e-05,
+      "loss": 0.3042,
+      "step": 32255
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.5085272756932e-05,
+      "loss": 0.282,
+      "step": 32260
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.508357642670209e-05,
+      "loss": 0.2859,
+      "step": 32265
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.508187983570017e-05,
+      "loss": 0.2901,
+      "step": 32270
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5080182983948274e-05,
+      "loss": 0.3145,
+      "step": 32275
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5078485871468436e-05,
+      "loss": 0.2789,
+      "step": 32280
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50767884982827e-05,
+      "loss": 0.3104,
+      "step": 32285
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5075090864413085e-05,
+      "loss": 0.2759,
+      "step": 32290
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.507339296988165e-05,
+      "loss": 0.3297,
+      "step": 32295
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.507169481471044e-05,
+      "loss": 0.2993,
+      "step": 32300
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506999639892149e-05,
+      "loss": 0.3079,
+      "step": 32305
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506829772253687e-05,
+      "loss": 0.3061,
+      "step": 32310
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506659878557863e-05,
+      "loss": 0.2899,
+      "step": 32315
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506489958806883e-05,
+      "loss": 0.3119,
+      "step": 32320
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506320013002953e-05,
+      "loss": 0.2938,
+      "step": 32325
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.506150041148281e-05,
+      "loss": 0.3046,
+      "step": 32330
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5059800432450705e-05,
+      "loss": 0.2817,
+      "step": 32335
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5058100192955327e-05,
+      "loss": 0.279,
+      "step": 32340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.505639969301874e-05,
+      "loss": 0.2616,
+      "step": 32345
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5054698932663006e-05,
+      "loss": 0.287,
+      "step": 32350
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.505299791191023e-05,
+      "loss": 0.292,
+      "step": 32355
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50512966307825e-05,
+      "loss": 0.3002,
+      "step": 32360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504959508930189e-05,
+      "loss": 0.2942,
+      "step": 32365
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.50478932874905e-05,
+      "loss": 0.3056,
+      "step": 32370
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504619122537043e-05,
+      "loss": 0.2633,
+      "step": 32375
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504448890296377e-05,
+      "loss": 0.2736,
+      "step": 32380
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.504278632029264e-05,
+      "loss": 0.2809,
+      "step": 32385
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5041083477379134e-05,
+      "loss": 0.3094,
+      "step": 32390
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.503938037424537e-05,
+      "loss": 0.2886,
+      "step": 32395
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.503767701091346e-05,
+      "loss": 0.2882,
+      "step": 32400
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5035973387405514e-05,
+      "loss": 0.3187,
+      "step": 32405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5034269503743665e-05,
+      "loss": 0.2848,
+      "step": 32410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5032565359950016e-05,
+      "loss": 0.3036,
+      "step": 32415
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5030860956046725e-05,
+      "loss": 0.3061,
+      "step": 32420
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5029156292055895e-05,
+      "loss": 0.2894,
+      "step": 32425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.502745136799967e-05,
+      "loss": 0.2715,
+      "step": 32430
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5025746183900195e-05,
+      "loss": 0.3355,
+      "step": 32435
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.502404073977959e-05,
+      "loss": 0.2947,
+      "step": 32440
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5022335035660026e-05,
+      "loss": 0.2973,
+      "step": 32445
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.502062907156363e-05,
+      "loss": 0.304,
+      "step": 32450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5018922847512555e-05,
+      "loss": 0.3056,
+      "step": 32455
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501721636352897e-05,
+      "loss": 0.2712,
+      "step": 32460
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501550961963502e-05,
+      "loss": 0.275,
+      "step": 32465
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501380261585286e-05,
+      "loss": 0.2669,
+      "step": 32470
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501209535220468e-05,
+      "loss": 0.3307,
+      "step": 32475
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.501038782871261e-05,
+      "loss": 0.2873,
+      "step": 32480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500868004539885e-05,
+      "loss": 0.3082,
+      "step": 32485
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5006972002285574e-05,
+      "loss": 0.2735,
+      "step": 32490
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500526369939494e-05,
+      "loss": 0.3092,
+      "step": 32495
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500355513674914e-05,
+      "loss": 0.297,
+      "step": 32500
+    },
+    {
+      "epoch": 1.14,
+      "eval_loss": 0.28957316279411316,
+      "eval_runtime": 10.558,
+      "eval_samples_per_second": 9.471,
+      "eval_steps_per_second": 9.471,
+      "step": 32500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.500184631437037e-05,
+      "loss": 0.2703,
+      "step": 32505
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.5000137232280806e-05,
+      "loss": 0.2789,
+      "step": 32510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.499842789050263e-05,
+      "loss": 0.2644,
+      "step": 32515
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.499671828905806e-05,
+      "loss": 0.286,
+      "step": 32520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.499500842796927e-05,
+      "loss": 0.285,
+      "step": 32525
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.499329830725848e-05,
+      "loss": 0.2909,
+      "step": 32530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4991587926947886e-05,
+      "loss": 0.2899,
+      "step": 32535
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.4989877287059694e-05,
+      "loss": 0.2982,
+      "step": 32540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4988166387616125e-05,
+      "loss": 0.2857,
+      "step": 32545
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.498645522863938e-05,
+      "loss": 0.2861,
+      "step": 32550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.49847438101517e-05,
+      "loss": 0.2805,
+      "step": 32555
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.498303213217527e-05,
+      "loss": 0.3046,
+      "step": 32560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.498132019473236e-05,
+      "loss": 0.2983,
+      "step": 32565
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.497960799784516e-05,
+      "loss": 0.2732,
+      "step": 32570
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.497789554153593e-05,
+      "loss": 0.295,
+      "step": 32575
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4976182825826877e-05,
+      "loss": 0.2888,
+      "step": 32580
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4974469850740264e-05,
+      "loss": 0.2696,
+      "step": 32585
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4972756616298323e-05,
+      "loss": 0.2795,
+      "step": 32590
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.49710431225233e-05,
+      "loss": 0.2967,
+      "step": 32595
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.496932936943744e-05,
+      "loss": 0.2909,
+      "step": 32600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4967615357063006e-05,
+      "loss": 0.2927,
+      "step": 32605
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4965901085422235e-05,
+      "loss": 0.2756,
+      "step": 32610
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.49641865545374e-05,
+      "loss": 0.2814,
+      "step": 32615
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.496247176443077e-05,
+      "loss": 0.3063,
+      "step": 32620
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4960756715124586e-05,
+      "loss": 0.3235,
+      "step": 32625
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495904140664114e-05,
+      "loss": 0.2832,
+      "step": 32630
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495732583900269e-05,
+      "loss": 0.2814,
+      "step": 32635
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495561001223152e-05,
+      "loss": 0.299,
+      "step": 32640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4953893926349896e-05,
+      "loss": 0.2991,
+      "step": 32645
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495217758138012e-05,
+      "loss": 0.2541,
+      "step": 32650
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.495046097734447e-05,
+      "loss": 0.2964,
+      "step": 32655
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.494874411426522e-05,
+      "loss": 0.2943,
+      "step": 32660
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4947026992164677e-05,
+      "loss": 0.2958,
+      "step": 32665
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4945309611065136e-05,
+      "loss": 0.3157,
+      "step": 32670
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.49435919709889e-05,
+      "loss": 0.3251,
+      "step": 32675
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4941874071958254e-05,
+      "loss": 0.2771,
+      "step": 32680
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.494015591399553e-05,
+      "loss": 0.2911,
+      "step": 32685
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493843749712301e-05,
+      "loss": 0.3239,
+      "step": 32690
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493671882136302e-05,
+      "loss": 0.3032,
+      "step": 32695
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493499988673788e-05,
+      "loss": 0.3173,
+      "step": 32700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493328069326991e-05,
+      "loss": 0.286,
+      "step": 32705
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.493156124098141e-05,
+      "loss": 0.2682,
+      "step": 32710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4929841529894735e-05,
+      "loss": 0.2971,
+      "step": 32715
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4928121560032196e-05,
+      "loss": 0.2826,
+      "step": 32720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4926401331416135e-05,
+      "loss": 0.2779,
+      "step": 32725
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4924680844068887e-05,
+      "loss": 0.2962,
+      "step": 32730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.492296009801279e-05,
+      "loss": 0.277,
+      "step": 32735
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.492123909327018e-05,
+      "loss": 0.3074,
+      "step": 32740
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491951782986341e-05,
+      "loss": 0.3198,
+      "step": 32745
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491779630781483e-05,
+      "loss": 0.3118,
+      "step": 32750
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491607452714679e-05,
+      "loss": 0.298,
+      "step": 32755
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.491435248788165e-05,
+      "loss": 0.293,
+      "step": 32760
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4912630190041765e-05,
+      "loss": 0.2902,
+      "step": 32765
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.49109076336495e-05,
+      "loss": 0.2658,
+      "step": 32770
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4909184818727224e-05,
+      "loss": 0.2954,
+      "step": 32775
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4907461745297294e-05,
+      "loss": 0.2952,
+      "step": 32780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.49057384133821e-05,
+      "loss": 0.2875,
+      "step": 32785
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.490401482300401e-05,
+      "loss": 0.3302,
+      "step": 32790
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.490229097418541e-05,
+      "loss": 0.32,
+      "step": 32795
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.490056686694867e-05,
+      "loss": 0.3094,
+      "step": 32800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.489884250131619e-05,
+      "loss": 0.3041,
+      "step": 32805
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.489711787731036e-05,
+      "loss": 0.3089,
+      "step": 32810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.489539299495356e-05,
+      "loss": 0.286,
+      "step": 32815
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4893667854268184e-05,
+      "loss": 0.3166,
+      "step": 32820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.489194245527665e-05,
+      "loss": 0.308,
+      "step": 32825
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.489021679800135e-05,
+      "loss": 0.2992,
+      "step": 32830
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4888490882464696e-05,
+      "loss": 0.2923,
+      "step": 32835
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.48867647086891e-05,
+      "loss": 0.312,
+      "step": 32840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.488503827669696e-05,
+      "loss": 0.2999,
+      "step": 32845
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4883311586510706e-05,
+      "loss": 0.3034,
+      "step": 32850
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4881584638152756e-05,
+      "loss": 0.2815,
+      "step": 32855
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4879857431645536e-05,
+      "loss": 0.289,
+      "step": 32860
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4878129967011464e-05,
+      "loss": 0.3124,
+      "step": 32865
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4876402244272974e-05,
+      "loss": 0.2979,
+      "step": 32870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4874674263452496e-05,
+      "loss": 0.2549,
+      "step": 32875
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4872946024572474e-05,
+      "loss": 0.3019,
+      "step": 32880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4871217527655344e-05,
+      "loss": 0.2989,
+      "step": 32885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4869488772723555e-05,
+      "loss": 0.2659,
+      "step": 32890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4867759759799544e-05,
+      "loss": 0.2789,
+      "step": 32895
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4866030488905766e-05,
+      "loss": 0.2879,
+      "step": 32900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.486430096006467e-05,
+      "loss": 0.3129,
+      "step": 32905
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4862571173298727e-05,
+      "loss": 0.2921,
+      "step": 32910
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4860841128630384e-05,
+      "loss": 0.3155,
+      "step": 32915
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.485911082608211e-05,
+      "loss": 0.2996,
+      "step": 32920
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4857380265676363e-05,
+      "loss": 0.282,
+      "step": 32925
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.485564944743562e-05,
+      "loss": 0.2971,
+      "step": 32930
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4853918371382365e-05,
+      "loss": 0.2768,
+      "step": 32935
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.485218703753905e-05,
+      "loss": 0.3019,
+      "step": 32940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.485045544592818e-05,
+      "loss": 0.2769,
+      "step": 32945
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.484872359657223e-05,
+      "loss": 0.2651,
+      "step": 32950
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4846991489493674e-05,
+      "loss": 0.3109,
+      "step": 32955
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4845259124715014e-05,
+      "loss": 0.2813,
+      "step": 32960
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4843526502258747e-05,
+      "loss": 0.2924,
+      "step": 32965
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.484179362214736e-05,
+      "loss": 0.2847,
+      "step": 32970
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.484006048440337e-05,
+      "loss": 0.2839,
+      "step": 32975
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.483832708904927e-05,
+      "loss": 0.2725,
+      "step": 32980
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4836593436107555e-05,
+      "loss": 0.3009,
+      "step": 32985
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.483485952560076e-05,
+      "loss": 0.2911,
+      "step": 32990
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4833125357551375e-05,
+      "loss": 0.2847,
+      "step": 32995
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4831390931981934e-05,
+      "loss": 0.3052,
+      "step": 33000
+    },
+    {
+      "epoch": 1.16,
+      "eval_loss": 0.28887951374053955,
+      "eval_runtime": 10.5389,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 33000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4829656248914945e-05,
+      "loss": 0.3001,
+      "step": 33005
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4827921308372945e-05,
+      "loss": 0.2907,
+      "step": 33010
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.482618611037845e-05,
+      "loss": 0.2813,
+      "step": 33015
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.482445065495399e-05,
+      "loss": 0.2769,
+      "step": 33020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4822714942122105e-05,
+      "loss": 0.3057,
+      "step": 33025
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.482097897190534e-05,
+      "loss": 0.2956,
+      "step": 33030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481924274432622e-05,
+      "loss": 0.282,
+      "step": 33035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481750625940729e-05,
+      "loss": 0.2964,
+      "step": 33040
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4815769517171105e-05,
+      "loss": 0.3054,
+      "step": 33045
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481403251764021e-05,
+      "loss": 0.2802,
+      "step": 33050
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.481229526083715e-05,
+      "loss": 0.302,
+      "step": 33055
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4810557746784514e-05,
+      "loss": 0.2758,
+      "step": 33060
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4808819975504835e-05,
+      "loss": 0.2762,
+      "step": 33065
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480708194702068e-05,
+      "loss": 0.3195,
+      "step": 33070
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480534366135462e-05,
+      "loss": 0.2812,
+      "step": 33075
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480360511852922e-05,
+      "loss": 0.2912,
+      "step": 33080
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480186631856707e-05,
+      "loss": 0.2869,
+      "step": 33085
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.480012726149072e-05,
+      "loss": 0.2889,
+      "step": 33090
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.479838794732277e-05,
+      "loss": 0.302,
+      "step": 33095
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.4796648376085815e-05,
+      "loss": 0.2824,
+      "step": 33100
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.479490854780242e-05,
+      "loss": 0.2879,
+      "step": 33105
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.479316846249517e-05,
+      "loss": 0.31,
+      "step": 33110
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.479142812018669e-05,
+      "loss": 0.2899,
+      "step": 33115
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.478968752089955e-05,
+      "loss": 0.3027,
+      "step": 33120
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.478794666465636e-05,
+      "loss": 0.2736,
+      "step": 33125
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.478620555147972e-05,
+      "loss": 0.2686,
+      "step": 33130
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.478446418139224e-05,
+      "loss": 0.3045,
+      "step": 33135
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.478272255441653e-05,
+      "loss": 0.273,
+      "step": 33140
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.478098067057522e-05,
+      "loss": 0.2836,
+      "step": 33145
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.47792385298909e-05,
+      "loss": 0.2809,
+      "step": 33150
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477749613238621e-05,
+      "loss": 0.3096,
+      "step": 33155
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4775753478083756e-05,
+      "loss": 0.3059,
+      "step": 33160
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477401056700618e-05,
+      "loss": 0.2698,
+      "step": 33165
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477226739917611e-05,
+      "loss": 0.2845,
+      "step": 33170
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.477052397461617e-05,
+      "loss": 0.3061,
+      "step": 33175
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.476878029334902e-05,
+      "loss": 0.2727,
+      "step": 33180
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4767036355397276e-05,
+      "loss": 0.2941,
+      "step": 33185
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.47652921607836e-05,
+      "loss": 0.284,
+      "step": 33190
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.476354770953062e-05,
+      "loss": 0.2993,
+      "step": 33195
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.476180300166101e-05,
+      "loss": 0.2959,
+      "step": 33200
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.476005803719741e-05,
+      "loss": 0.2831,
+      "step": 33205
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4758312816162466e-05,
+      "loss": 0.3011,
+      "step": 33210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4756567338578856e-05,
+      "loss": 0.3106,
+      "step": 33215
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4754821604469245e-05,
+      "loss": 0.2985,
+      "step": 33220
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.47530756138563e-05,
+      "loss": 0.3007,
+      "step": 33225
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4751329366762677e-05,
+      "loss": 0.3026,
+      "step": 33230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474958286321105e-05,
+      "loss": 0.2842,
+      "step": 33235
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4747836103224115e-05,
+      "loss": 0.2882,
+      "step": 33240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474608908682455e-05,
+      "loss": 0.2853,
+      "step": 33245
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.474434181403502e-05,
+      "loss": 0.3047,
+      "step": 33250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4742594284878225e-05,
+      "loss": 0.2728,
+      "step": 33255
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4740846499376856e-05,
+      "loss": 0.2977,
+      "step": 33260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4739098457553606e-05,
+      "loss": 0.3427,
+      "step": 33265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473735015943117e-05,
+      "loss": 0.2878,
+      "step": 33270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473560160503225e-05,
+      "loss": 0.2875,
+      "step": 33275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473385279437955e-05,
+      "loss": 0.2912,
+      "step": 33280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4732103727495775e-05,
+      "loss": 0.2663,
+      "step": 33285
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.473035440440364e-05,
+      "loss": 0.2863,
+      "step": 33290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.472860482512585e-05,
+      "loss": 0.2773,
+      "step": 33295
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.472685498968513e-05,
+      "loss": 0.2813,
+      "step": 33300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.47251048981042e-05,
+      "loss": 0.302,
+      "step": 33305
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4723354550405784e-05,
+      "loss": 0.2719,
+      "step": 33310
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4721603946612604e-05,
+      "loss": 0.2746,
+      "step": 33315
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.47198530867474e-05,
+      "loss": 0.3027,
+      "step": 33320
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4718101970832895e-05,
+      "loss": 0.2748,
+      "step": 33325
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4716350598891834e-05,
+      "loss": 0.2948,
+      "step": 33330
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4714598970946944e-05,
+      "loss": 0.3105,
+      "step": 33335
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.471284708702098e-05,
+      "loss": 0.3052,
+      "step": 33340
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.47110949471367e-05,
+      "loss": 0.3049,
+      "step": 33345
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4709342551316836e-05,
+      "loss": 0.2955,
+      "step": 33350
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470758989958415e-05,
+      "loss": 0.276,
+      "step": 33355
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470583699196139e-05,
+      "loss": 0.2762,
+      "step": 33360
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470408382847133e-05,
+      "loss": 0.3054,
+      "step": 33365
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4702330409136716e-05,
+      "loss": 0.2919,
+      "step": 33370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.470057673398034e-05,
+      "loss": 0.2929,
+      "step": 33375
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.469882280302495e-05,
+      "loss": 0.3022,
+      "step": 33380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.469706861629333e-05,
+      "loss": 0.3106,
+      "step": 33385
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.469531417380826e-05,
+      "loss": 0.2885,
+      "step": 33390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.46935594755925e-05,
+      "loss": 0.2992,
+      "step": 33395
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4691804521668865e-05,
+      "loss": 0.2822,
+      "step": 33400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4690049312060126e-05,
+      "loss": 0.3032,
+      "step": 33405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.468829384678907e-05,
+      "loss": 0.2915,
+      "step": 33410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4686538125878485e-05,
+      "loss": 0.2831,
+      "step": 33415
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.468478214935118e-05,
+      "loss": 0.2915,
+      "step": 33420
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4683025917229956e-05,
+      "loss": 0.3001,
+      "step": 33425
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.468126942953761e-05,
+      "loss": 0.3102,
+      "step": 33430
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.467951268629695e-05,
+      "loss": 0.2899,
+      "step": 33435
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4677755687530787e-05,
+      "loss": 0.2786,
+      "step": 33440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.467599843326193e-05,
+      "loss": 0.2695,
+      "step": 33445
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.467424092351321e-05,
+      "loss": 0.3129,
+      "step": 33450
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.467248315830743e-05,
+      "loss": 0.273,
+      "step": 33455
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4670725137667425e-05,
+      "loss": 0.2893,
+      "step": 33460
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466896686161601e-05,
+      "loss": 0.2975,
+      "step": 33465
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466720833017604e-05,
+      "loss": 0.2891,
+      "step": 33470
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4665449543370316e-05,
+      "loss": 0.2937,
+      "step": 33475
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466369050122169e-05,
+      "loss": 0.2957,
+      "step": 33480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.466193120375301e-05,
+      "loss": 0.2962,
+      "step": 33485
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46601716509871e-05,
+      "loss": 0.3038,
+      "step": 33490
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.465841184294682e-05,
+      "loss": 0.2816,
+      "step": 33495
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.465665177965502e-05,
+      "loss": 0.3047,
+      "step": 33500
+    },
+    {
+      "epoch": 1.18,
+      "eval_loss": 0.2884669601917267,
+      "eval_runtime": 10.5481,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 33500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4654891461134556e-05,
+      "loss": 0.2963,
+      "step": 33505
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.465313088740828e-05,
+      "loss": 0.2943,
+      "step": 33510
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4651370058499045e-05,
+      "loss": 0.3003,
+      "step": 33515
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.464960897442973e-05,
+      "loss": 0.2907,
+      "step": 33520
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4647847635223175e-05,
+      "loss": 0.2912,
+      "step": 33525
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.464608604090228e-05,
+      "loss": 0.2866,
+      "step": 33530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46443241914899e-05,
+      "loss": 0.2882,
+      "step": 33535
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.464256208700892e-05,
+      "loss": 0.2638,
+      "step": 33540
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.464079972748221e-05,
+      "loss": 0.3069,
+      "step": 33545
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.463903711293267e-05,
+      "loss": 0.283,
+      "step": 33550
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.463727424338316e-05,
+      "loss": 0.2771,
+      "step": 33555
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46355111188566e-05,
+      "loss": 0.2906,
+      "step": 33560
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.463374773937587e-05,
+      "loss": 0.3029,
+      "step": 33565
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4631984104963854e-05,
+      "loss": 0.2964,
+      "step": 33570
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.463022021564346e-05,
+      "loss": 0.2757,
+      "step": 33575
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46284560714376e-05,
+      "loss": 0.2749,
+      "step": 33580
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.462669167236917e-05,
+      "loss": 0.3006,
+      "step": 33585
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.462492701846109e-05,
+      "loss": 0.2939,
+      "step": 33590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.462316210973626e-05,
+      "loss": 0.2729,
+      "step": 33595
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.46213969462176e-05,
+      "loss": 0.2856,
+      "step": 33600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4619631527928043e-05,
+      "loss": 0.3168,
+      "step": 33605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4617865854890495e-05,
+      "loss": 0.2675,
+      "step": 33610
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.461609992712788e-05,
+      "loss": 0.2795,
+      "step": 33615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.461433374466314e-05,
+      "loss": 0.2885,
+      "step": 33620
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4612567307519205e-05,
+      "loss": 0.2828,
+      "step": 33625
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4610800615719007e-05,
+      "loss": 0.299,
+      "step": 33630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4609033669285485e-05,
+      "loss": 0.2943,
+      "step": 33635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.460726646824159e-05,
+      "loss": 0.2989,
+      "step": 33640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.460549901261025e-05,
+      "loss": 0.2855,
+      "step": 33645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.4603731302414445e-05,
+      "loss": 0.2958,
+      "step": 33650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.460196333767709e-05,
+      "loss": 0.2847,
+      "step": 33655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.460019511842117e-05,
+      "loss": 0.2869,
+      "step": 33660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.459842664466962e-05,
+      "loss": 0.2747,
+      "step": 33665
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.459665791644543e-05,
+      "loss": 0.2878,
+      "step": 33670
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.459488893377155e-05,
+      "loss": 0.3042,
+      "step": 33675
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.459311969667095e-05,
+      "loss": 0.2931,
+      "step": 33680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.459135020516659e-05,
+      "loss": 0.2853,
+      "step": 33685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4589580459281474e-05,
+      "loss": 0.287,
+      "step": 33690
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4587810459038556e-05,
+      "loss": 0.2946,
+      "step": 33695
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.458604020446084e-05,
+      "loss": 0.292,
+      "step": 33700
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.458426969557129e-05,
+      "loss": 0.3181,
+      "step": 33705
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4582498932392906e-05,
+      "loss": 0.2827,
+      "step": 33710
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4580727914948685e-05,
+      "loss": 0.2826,
+      "step": 33715
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.45789566432616e-05,
+      "loss": 0.2858,
+      "step": 33720
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.457718511735468e-05,
+      "loss": 0.2845,
+      "step": 33725
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4575413337250914e-05,
+      "loss": 0.3087,
+      "step": 33730
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.45736413029733e-05,
+      "loss": 0.2915,
+      "step": 33735
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.457186901454485e-05,
+      "loss": 0.2721,
+      "step": 33740
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4570096471988596e-05,
+      "loss": 0.3164,
+      "step": 33745
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4568323675327525e-05,
+      "loss": 0.2786,
+      "step": 33750
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.456655062458467e-05,
+      "loss": 0.2895,
+      "step": 33755
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4564777319783047e-05,
+      "loss": 0.2568,
+      "step": 33760
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.456300376094568e-05,
+      "loss": 0.2961,
+      "step": 33765
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.456122994809561e-05,
+      "loss": 0.3038,
+      "step": 33770
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.455945588125585e-05,
+      "loss": 0.2928,
+      "step": 33775
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4557681560449456e-05,
+      "loss": 0.2999,
+      "step": 33780
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.455590698569945e-05,
+      "loss": 0.3193,
+      "step": 33785
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4554132157028885e-05,
+      "loss": 0.2992,
+      "step": 33790
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4552357074460796e-05,
+      "loss": 0.3181,
+      "step": 33795
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4550581738018236e-05,
+      "loss": 0.2944,
+      "step": 33800
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.454880614772426e-05,
+      "loss": 0.2957,
+      "step": 33805
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.454703030360192e-05,
+      "loss": 0.3041,
+      "step": 33810
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4545254205674266e-05,
+      "loss": 0.2972,
+      "step": 33815
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4543477853964375e-05,
+      "loss": 0.2742,
+      "step": 33820
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.45417012484953e-05,
+      "loss": 0.279,
+      "step": 33825
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.453992438929011e-05,
+      "loss": 0.2805,
+      "step": 33830
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4538147276371884e-05,
+      "loss": 0.2776,
+      "step": 33835
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4536369909763683e-05,
+      "loss": 0.2843,
+      "step": 33840
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.453459228948861e-05,
+      "loss": 0.2555,
+      "step": 33845
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.453281441556972e-05,
+      "loss": 0.3132,
+      "step": 33850
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.45310362880301e-05,
+      "loss": 0.3091,
+      "step": 33855
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4529257906892864e-05,
+      "loss": 0.2793,
+      "step": 33860
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4527479272181074e-05,
+      "loss": 0.2814,
+      "step": 33865
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.452570038391783e-05,
+      "loss": 0.3016,
+      "step": 33870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.452392124212624e-05,
+      "loss": 0.2923,
+      "step": 33875
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.45221418468294e-05,
+      "loss": 0.2731,
+      "step": 33880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.452036219805041e-05,
+      "loss": 0.3242,
+      "step": 33885
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.451858229581238e-05,
+      "loss": 0.2567,
+      "step": 33890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.451680214013842e-05,
+      "loss": 0.3106,
+      "step": 33895
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4515021731051654e-05,
+      "loss": 0.3036,
+      "step": 33900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4513241068575196e-05,
+      "loss": 0.2825,
+      "step": 33905
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4511460152732156e-05,
+      "loss": 0.2796,
+      "step": 33910
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4509678983545654e-05,
+      "loss": 0.2792,
+      "step": 33915
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4507897561038845e-05,
+      "loss": 0.266,
+      "step": 33920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4506115885234836e-05,
+      "loss": 0.2884,
+      "step": 33925
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.450433395615676e-05,
+      "loss": 0.283,
+      "step": 33930
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4502551773827775e-05,
+      "loss": 0.2885,
+      "step": 33935
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4500769338271e-05,
+      "loss": 0.2917,
+      "step": 33940
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4498986649509584e-05,
+      "loss": 0.2898,
+      "step": 33945
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.449720370756668e-05,
+      "loss": 0.2922,
+      "step": 33950
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.449542051246544e-05,
+      "loss": 0.3001,
+      "step": 33955
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.4493637064229e-05,
+      "loss": 0.2668,
+      "step": 33960
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.449185336288054e-05,
+      "loss": 0.2862,
+      "step": 33965
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44900694084432e-05,
+      "loss": 0.2975,
+      "step": 33970
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.448828520094017e-05,
+      "loss": 0.293,
+      "step": 33975
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.448650074039459e-05,
+      "loss": 0.2925,
+      "step": 33980
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4484716026829635e-05,
+      "loss": 0.2922,
+      "step": 33985
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.448293106026849e-05,
+      "loss": 0.2837,
+      "step": 33990
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.448114584073432e-05,
+      "loss": 0.308,
+      "step": 33995
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.447936036825031e-05,
+      "loss": 0.3173,
+      "step": 34000
+    },
+    {
+      "epoch": 1.2,
+      "eval_loss": 0.28807368874549866,
+      "eval_runtime": 10.551,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 34000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.447757464283965e-05,
+      "loss": 0.2874,
+      "step": 34005
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.447578866452552e-05,
+      "loss": 0.2941,
+      "step": 34010
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44740024333311e-05,
+      "loss": 0.2793,
+      "step": 34015
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4472215949279594e-05,
+      "loss": 0.2921,
+      "step": 34020
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44704292123942e-05,
+      "loss": 0.2979,
+      "step": 34025
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.446864222269812e-05,
+      "loss": 0.3073,
+      "step": 34030
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.446685498021455e-05,
+      "loss": 0.3038,
+      "step": 34035
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.446506748496668e-05,
+      "loss": 0.2936,
+      "step": 34040
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4463279736977754e-05,
+      "loss": 0.3041,
+      "step": 34045
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.446149173627096e-05,
+      "loss": 0.3029,
+      "step": 34050
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.445970348286952e-05,
+      "loss": 0.3014,
+      "step": 34055
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.445791497679666e-05,
+      "loss": 0.2813,
+      "step": 34060
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.445612621807559e-05,
+      "loss": 0.2693,
+      "step": 34065
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.445433720672955e-05,
+      "loss": 0.2814,
+      "step": 34070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4452547942781765e-05,
+      "loss": 0.2809,
+      "step": 34075
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4450758426255456e-05,
+      "loss": 0.285,
+      "step": 34080
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.444896865717387e-05,
+      "loss": 0.2821,
+      "step": 34085
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.444717863556025e-05,
+      "loss": 0.3087,
+      "step": 34090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.444538836143782e-05,
+      "loss": 0.2888,
+      "step": 34095
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.444359783482985e-05,
+      "loss": 0.2941,
+      "step": 34100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4441807055759574e-05,
+      "loss": 0.3051,
+      "step": 34105
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.444001602425024e-05,
+      "loss": 0.2595,
+      "step": 34110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.443822474032511e-05,
+      "loss": 0.2836,
+      "step": 34115
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.443643320400744e-05,
+      "loss": 0.3046,
+      "step": 34120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.44346414153205e-05,
+      "loss": 0.2819,
+      "step": 34125
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.443284937428754e-05,
+      "loss": 0.2821,
+      "step": 34130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.443105708093185e-05,
+      "loss": 0.3211,
+      "step": 34135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4429264535276684e-05,
+      "loss": 0.2882,
+      "step": 34140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.442747173734532e-05,
+      "loss": 0.3001,
+      "step": 34145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4425678687161046e-05,
+      "loss": 0.2723,
+      "step": 34150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4423885384747124e-05,
+      "loss": 0.2883,
+      "step": 34155
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.442209183012685e-05,
+      "loss": 0.2654,
+      "step": 34160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.442029802332353e-05,
+      "loss": 0.2737,
+      "step": 34165
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4418503964360424e-05,
+      "loss": 0.29,
+      "step": 34170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.441670965326085e-05,
+      "loss": 0.2884,
+      "step": 34175
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4414915090048095e-05,
+      "loss": 0.3039,
+      "step": 34180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4413120274745455e-05,
+      "loss": 0.2871,
+      "step": 34185
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.441132520737625e-05,
+      "loss": 0.2728,
+      "step": 34190
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.440952988796378e-05,
+      "loss": 0.3025,
+      "step": 34195
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.440773431653136e-05,
+      "loss": 0.2866,
+      "step": 34200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.440593849310229e-05,
+      "loss": 0.2843,
+      "step": 34205
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4404142417699904e-05,
+      "loss": 0.2922,
+      "step": 34210
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4402346090347515e-05,
+      "loss": 0.2985,
+      "step": 34215
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4400549511068445e-05,
+      "loss": 0.2766,
+      "step": 34220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4398752679886026e-05,
+      "loss": 0.2857,
+      "step": 34225
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4396955596823594e-05,
+      "loss": 0.3026,
+      "step": 34230
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.439515826190447e-05,
+      "loss": 0.2888,
+      "step": 34235
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.4393360675151996e-05,
+      "loss": 0.3029,
+      "step": 34240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.439156283658952e-05,
+      "loss": 0.2839,
+      "step": 34245
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.438976474624038e-05,
+      "loss": 0.2773,
+      "step": 34250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4387966404127926e-05,
+      "loss": 0.3019,
+      "step": 34255
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4386167810275493e-05,
+      "loss": 0.3107,
+      "step": 34260
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.438436896470646e-05,
+      "loss": 0.2845,
+      "step": 34265
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4382569867444164e-05,
+      "loss": 0.2895,
+      "step": 34270
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.438077051851197e-05,
+      "loss": 0.2912,
+      "step": 34275
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.437897091793325e-05,
+      "loss": 0.2887,
+      "step": 34280
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.437717106573136e-05,
+      "loss": 0.2649,
+      "step": 34285
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.437537096192967e-05,
+      "loss": 0.2903,
+      "step": 34290
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.437357060655156e-05,
+      "loss": 0.2955,
+      "step": 34295
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4371769999620406e-05,
+      "loss": 0.2914,
+      "step": 34300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4369969141159585e-05,
+      "loss": 0.2676,
+      "step": 34305
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.436816803119248e-05,
+      "loss": 0.2737,
+      "step": 34310
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4366366669742476e-05,
+      "loss": 0.3046,
+      "step": 34315
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4364565056832966e-05,
+      "loss": 0.3038,
+      "step": 34320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4362763192487336e-05,
+      "loss": 0.2929,
+      "step": 34325
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.436096107672899e-05,
+      "loss": 0.2979,
+      "step": 34330
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435915870958132e-05,
+      "loss": 0.3026,
+      "step": 34335
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4357356091067734e-05,
+      "loss": 0.3056,
+      "step": 34340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435555322121164e-05,
+      "loss": 0.3093,
+      "step": 34345
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435375010003644e-05,
+      "loss": 0.2713,
+      "step": 34350
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.435194672756555e-05,
+      "loss": 0.295,
+      "step": 34355
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4350143103822385e-05,
+      "loss": 0.2824,
+      "step": 34360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.434833922883036e-05,
+      "loss": 0.2898,
+      "step": 34365
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.434653510261291e-05,
+      "loss": 0.2953,
+      "step": 34370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.434473072519344e-05,
+      "loss": 0.3097,
+      "step": 34375
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4342926096595396e-05,
+      "loss": 0.301,
+      "step": 34380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.434112121684221e-05,
+      "loss": 0.2811,
+      "step": 34385
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4339316085957305e-05,
+      "loss": 0.2814,
+      "step": 34390
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433751070396412e-05,
+      "loss": 0.2852,
+      "step": 34395
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433570507088611e-05,
+      "loss": 0.283,
+      "step": 34400
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433389918674671e-05,
+      "loss": 0.2712,
+      "step": 34405
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4332093051569374e-05,
+      "loss": 0.3063,
+      "step": 34410
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.433028666537755e-05,
+      "loss": 0.2805,
+      "step": 34415
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4328480028194684e-05,
+      "loss": 0.3026,
+      "step": 34420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.432667314004425e-05,
+      "loss": 0.2853,
+      "step": 34425
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.432486600094971e-05,
+      "loss": 0.3051,
+      "step": 34430
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.432305861093451e-05,
+      "loss": 0.2682,
+      "step": 34435
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4321250970022126e-05,
+      "loss": 0.2969,
+      "step": 34440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431944307823604e-05,
+      "loss": 0.2952,
+      "step": 34445
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431763493559971e-05,
+      "loss": 0.2963,
+      "step": 34450
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431582654213663e-05,
+      "loss": 0.296,
+      "step": 34455
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431401789787026e-05,
+      "loss": 0.2775,
+      "step": 34460
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.43122090028241e-05,
+      "loss": 0.2827,
+      "step": 34465
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.431039985702164e-05,
+      "loss": 0.3046,
+      "step": 34470
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.430859046048636e-05,
+      "loss": 0.2818,
+      "step": 34475
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.430678081324175e-05,
+      "loss": 0.2823,
+      "step": 34480
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4304970915311326e-05,
+      "loss": 0.2867,
+      "step": 34485
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4303160766718574e-05,
+      "loss": 0.3043,
+      "step": 34490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.4301350367487005e-05,
+      "loss": 0.3266,
+      "step": 34495
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429953971764012e-05,
+      "loss": 0.287,
+      "step": 34500
+    },
+    {
+      "epoch": 1.21,
+      "eval_loss": 0.2872275710105896,
+      "eval_runtime": 10.5666,
+      "eval_samples_per_second": 9.464,
+      "eval_steps_per_second": 9.464,
+      "step": 34500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429772881720142e-05,
+      "loss": 0.2876,
+      "step": 34505
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429591766619444e-05,
+      "loss": 0.2798,
+      "step": 34510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429410626464268e-05,
+      "loss": 0.285,
+      "step": 34515
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429229461256967e-05,
+      "loss": 0.2681,
+      "step": 34520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.429048270999892e-05,
+      "loss": 0.2791,
+      "step": 34525
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.428867055695398e-05,
+      "loss": 0.3229,
+      "step": 34530
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4286858153458344e-05,
+      "loss": 0.3223,
+      "step": 34535
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4285045499535584e-05,
+      "loss": 0.3156,
+      "step": 34540
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.428323259520921e-05,
+      "loss": 0.2721,
+      "step": 34545
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.428141944050277e-05,
+      "loss": 0.2948,
+      "step": 34550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.42796060354398e-05,
+      "loss": 0.2897,
+      "step": 34555
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.427779238004386e-05,
+      "loss": 0.2914,
+      "step": 34560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4275978474338485e-05,
+      "loss": 0.2896,
+      "step": 34565
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4274164318347235e-05,
+      "loss": 0.2832,
+      "step": 34570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.427234991209366e-05,
+      "loss": 0.3122,
+      "step": 34575
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4270535255601324e-05,
+      "loss": 0.2831,
+      "step": 34580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4268720348893786e-05,
+      "loss": 0.2953,
+      "step": 34585
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4266905191994615e-05,
+      "loss": 0.2978,
+      "step": 34590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.426508978492738e-05,
+      "loss": 0.2993,
+      "step": 34595
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.426327412771565e-05,
+      "loss": 0.3028,
+      "step": 34600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4261458220383e-05,
+      "loss": 0.2977,
+      "step": 34605
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.425964206295301e-05,
+      "loss": 0.2806,
+      "step": 34610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4257825655449257e-05,
+      "loss": 0.2709,
+      "step": 34615
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.425600899789533e-05,
+      "loss": 0.2556,
+      "step": 34620
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.425419209031482e-05,
+      "loss": 0.2851,
+      "step": 34625
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.42523749327313e-05,
+      "loss": 0.2911,
+      "step": 34630
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.425055752516839e-05,
+      "loss": 0.2835,
+      "step": 34635
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.424873986764968e-05,
+      "loss": 0.2993,
+      "step": 34640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.424692196019877e-05,
+      "loss": 0.289,
+      "step": 34645
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4245103802839256e-05,
+      "loss": 0.2912,
+      "step": 34650
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4243285395594754e-05,
+      "loss": 0.2923,
+      "step": 34655
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.424146673848887e-05,
+      "loss": 0.291,
+      "step": 34660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4239647831545225e-05,
+      "loss": 0.3019,
+      "step": 34665
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423782867478743e-05,
+      "loss": 0.2934,
+      "step": 34670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423600926823911e-05,
+      "loss": 0.2817,
+      "step": 34675
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423418961192389e-05,
+      "loss": 0.3051,
+      "step": 34680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423236970586538e-05,
+      "loss": 0.2773,
+      "step": 34685
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.423054955008724e-05,
+      "loss": 0.2957,
+      "step": 34690
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.422872914461308e-05,
+      "loss": 0.2919,
+      "step": 34695
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4226908489466544e-05,
+      "loss": 0.2951,
+      "step": 34700
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.422508758467128e-05,
+      "loss": 0.278,
+      "step": 34705
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4223266430250915e-05,
+      "loss": 0.2756,
+      "step": 34710
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.42214450262291e-05,
+      "loss": 0.2845,
+      "step": 34715
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.42196233726295e-05,
+      "loss": 0.2854,
+      "step": 34720
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.421780146947575e-05,
+      "loss": 0.287,
+      "step": 34725
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4215979316791516e-05,
+      "loss": 0.2951,
+      "step": 34730
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4214156914600454e-05,
+      "loss": 0.2786,
+      "step": 34735
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4212334262926226e-05,
+      "loss": 0.3042,
+      "step": 34740
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.42105113617925e-05,
+      "loss": 0.2986,
+      "step": 34745
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420868821122295e-05,
+      "loss": 0.2869,
+      "step": 34750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420686481124123e-05,
+      "loss": 0.2745,
+      "step": 34755
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420504116187103e-05,
+      "loss": 0.2817,
+      "step": 34760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4203217263136034e-05,
+      "loss": 0.2904,
+      "step": 34765
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.420139311505991e-05,
+      "loss": 0.3057,
+      "step": 34770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.419956871766636e-05,
+      "loss": 0.2846,
+      "step": 34775
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4197744070979056e-05,
+      "loss": 0.2672,
+      "step": 34780
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.41959191750217e-05,
+      "loss": 0.3128,
+      "step": 34785
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.419409402981799e-05,
+      "loss": 0.2899,
+      "step": 34790
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.41922686353916e-05,
+      "loss": 0.3013,
+      "step": 34795
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4190442991766264e-05,
+      "loss": 0.2865,
+      "step": 34800
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4188617098965675e-05,
+      "loss": 0.2979,
+      "step": 34805
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.4186790957013536e-05,
+      "loss": 0.2925,
+      "step": 34810
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.418496456593356e-05,
+      "loss": 0.2984,
+      "step": 34815
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.418313792574946e-05,
+      "loss": 0.2933,
+      "step": 34820
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4181311036484955e-05,
+      "loss": 0.2885,
+      "step": 34825
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.417948389816377e-05,
+      "loss": 0.3076,
+      "step": 34830
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.417765651080963e-05,
+      "loss": 0.271,
+      "step": 34835
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.417582887444625e-05,
+      "loss": 0.3028,
+      "step": 34840
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4174000989097375e-05,
+      "loss": 0.3184,
+      "step": 34845
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.417217285478673e-05,
+      "loss": 0.2935,
+      "step": 34850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.417034447153806e-05,
+      "loss": 0.2841,
+      "step": 34855
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41685158393751e-05,
+      "loss": 0.2943,
+      "step": 34860
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41666869583216e-05,
+      "loss": 0.2861,
+      "step": 34865
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.416485782840129e-05,
+      "loss": 0.2744,
+      "step": 34870
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4163028449637933e-05,
+      "loss": 0.2977,
+      "step": 34875
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.416119882205528e-05,
+      "loss": 0.3126,
+      "step": 34880
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41593689456771e-05,
+      "loss": 0.2921,
+      "step": 34885
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.415753882052713e-05,
+      "loss": 0.3003,
+      "step": 34890
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.415570844662914e-05,
+      "loss": 0.27,
+      "step": 34895
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.415387782400691e-05,
+      "loss": 0.2767,
+      "step": 34900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.415204695268419e-05,
+      "loss": 0.2865,
+      "step": 34905
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.415021583268477e-05,
+      "loss": 0.3064,
+      "step": 34910
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.414838446403242e-05,
+      "loss": 0.2653,
+      "step": 34915
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.414655284675091e-05,
+      "loss": 0.2878,
+      "step": 34920
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.414472098086403e-05,
+      "loss": 0.2892,
+      "step": 34925
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4142888866395566e-05,
+      "loss": 0.2682,
+      "step": 34930
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41410565033693e-05,
+      "loss": 0.3151,
+      "step": 34935
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.413922389180904e-05,
+      "loss": 0.2662,
+      "step": 34940
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.413739103173856e-05,
+      "loss": 0.2859,
+      "step": 34945
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4135557923181684e-05,
+      "loss": 0.302,
+      "step": 34950
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4133724566162195e-05,
+      "loss": 0.2936,
+      "step": 34955
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.413189096070389e-05,
+      "loss": 0.2965,
+      "step": 34960
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41300571068306e-05,
+      "loss": 0.2763,
+      "step": 34965
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.412822300456613e-05,
+      "loss": 0.2747,
+      "step": 34970
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4126388653934276e-05,
+      "loss": 0.2767,
+      "step": 34975
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.412455405495888e-05,
+      "loss": 0.295,
+      "step": 34980
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.412271920766375e-05,
+      "loss": 0.2844,
+      "step": 34985
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4120884112072714e-05,
+      "loss": 0.2889,
+      "step": 34990
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41190487682096e-05,
+      "loss": 0.303,
+      "step": 34995
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.411721317609824e-05,
+      "loss": 0.2971,
+      "step": 35000
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 0.2867205739021301,
+      "eval_runtime": 10.5517,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 35000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4115377335762456e-05,
+      "loss": 0.3064,
+      "step": 35005
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.41135412472261e-05,
+      "loss": 0.2653,
+      "step": 35010
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.411170491051301e-05,
+      "loss": 0.295,
+      "step": 35015
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410986832564702e-05,
+      "loss": 0.2904,
+      "step": 35020
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410803149265199e-05,
+      "loss": 0.256,
+      "step": 35025
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4106194411551766e-05,
+      "loss": 0.2899,
+      "step": 35030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410435708237019e-05,
+      "loss": 0.2897,
+      "step": 35035
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410251950513113e-05,
+      "loss": 0.3124,
+      "step": 35040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.410068167985845e-05,
+      "loss": 0.2721,
+      "step": 35045
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4098843606576004e-05,
+      "loss": 0.2894,
+      "step": 35050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4097005285307654e-05,
+      "loss": 0.2857,
+      "step": 35055
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.409516671607729e-05,
+      "loss": 0.277,
+      "step": 35060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.409332789890876e-05,
+      "loss": 0.3018,
+      "step": 35065
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.409148883382595e-05,
+      "loss": 0.2893,
+      "step": 35070
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4089649520852736e-05,
+      "loss": 0.285,
+      "step": 35075
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4087809960013013e-05,
+      "loss": 0.3,
+      "step": 35080
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.408597015133065e-05,
+      "loss": 0.2834,
+      "step": 35085
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.408413009482954e-05,
+      "loss": 0.3,
+      "step": 35090
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.4082289790533586e-05,
+      "loss": 0.2805,
+      "step": 35095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.408044923846667e-05,
+      "loss": 0.3079,
+      "step": 35100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4078608438652705e-05,
+      "loss": 0.3199,
+      "step": 35105
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.407676739111557e-05,
+      "loss": 0.2876,
+      "step": 35110
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.407492609587919e-05,
+      "loss": 0.2647,
+      "step": 35115
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.407308455296746e-05,
+      "loss": 0.3269,
+      "step": 35120
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4071242762404294e-05,
+      "loss": 0.2996,
+      "step": 35125
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406940072421362e-05,
+      "loss": 0.2991,
+      "step": 35130
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406755843841933e-05,
+      "loss": 0.2908,
+      "step": 35135
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406571590504537e-05,
+      "loss": 0.2929,
+      "step": 35140
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406387312411565e-05,
+      "loss": 0.3158,
+      "step": 35145
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4062030095654104e-05,
+      "loss": 0.2799,
+      "step": 35150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.406018681968466e-05,
+      "loss": 0.2741,
+      "step": 35155
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.405834329623125e-05,
+      "loss": 0.3071,
+      "step": 35160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.405649952531781e-05,
+      "loss": 0.2802,
+      "step": 35165
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.405465550696829e-05,
+      "loss": 0.2824,
+      "step": 35170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4052811241206614e-05,
+      "loss": 0.2678,
+      "step": 35175
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.405096672805675e-05,
+      "loss": 0.2921,
+      "step": 35180
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.404912196754264e-05,
+      "loss": 0.2858,
+      "step": 35185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.404727695968823e-05,
+      "loss": 0.306,
+      "step": 35190
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4045431704517474e-05,
+      "loss": 0.3048,
+      "step": 35195
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.404358620205435e-05,
+      "loss": 0.2906,
+      "step": 35200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.40417404523228e-05,
+      "loss": 0.2993,
+      "step": 35205
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403989445534681e-05,
+      "loss": 0.2896,
+      "step": 35210
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403804821115033e-05,
+      "loss": 0.3059,
+      "step": 35215
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403620171975734e-05,
+      "loss": 0.286,
+      "step": 35220
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403435498119182e-05,
+      "loss": 0.324,
+      "step": 35225
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.403250799547774e-05,
+      "loss": 0.2888,
+      "step": 35230
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.40306607626391e-05,
+      "loss": 0.2763,
+      "step": 35235
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.402881328269985e-05,
+      "loss": 0.3169,
+      "step": 35240
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.402696555568402e-05,
+      "loss": 0.3118,
+      "step": 35245
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.402511758161557e-05,
+      "loss": 0.2793,
+      "step": 35250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.402326936051852e-05,
+      "loss": 0.3025,
+      "step": 35255
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.402142089241684e-05,
+      "loss": 0.261,
+      "step": 35260
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4019572177334545e-05,
+      "loss": 0.2901,
+      "step": 35265
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.401772321529565e-05,
+      "loss": 0.2884,
+      "step": 35270
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.401587400632414e-05,
+      "loss": 0.3042,
+      "step": 35275
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4014024550444044e-05,
+      "loss": 0.2762,
+      "step": 35280
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.401217484767937e-05,
+      "loss": 0.2782,
+      "step": 35285
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.401032489805413e-05,
+      "loss": 0.2859,
+      "step": 35290
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.400847470159236e-05,
+      "loss": 0.2922,
+      "step": 35295
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.400662425831807e-05,
+      "loss": 0.2677,
+      "step": 35300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.4004773568255284e-05,
+      "loss": 0.2656,
+      "step": 35305
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.400292263142805e-05,
+      "loss": 0.2953,
+      "step": 35310
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.400107144786038e-05,
+      "loss": 0.2876,
+      "step": 35315
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.399922001757631e-05,
+      "loss": 0.2982,
+      "step": 35320
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3997368340599896e-05,
+      "loss": 0.2891,
+      "step": 35325
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3995516416955187e-05,
+      "loss": 0.2719,
+      "step": 35330
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.39936642466662e-05,
+      "loss": 0.3066,
+      "step": 35335
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.399181182975701e-05,
+      "loss": 0.2932,
+      "step": 35340
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398995916625165e-05,
+      "loss": 0.3074,
+      "step": 35345
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398810625617419e-05,
+      "loss": 0.294,
+      "step": 35350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398625309954868e-05,
+      "loss": 0.2983,
+      "step": 35355
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398439969639919e-05,
+      "loss": 0.2994,
+      "step": 35360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.398254604674978e-05,
+      "loss": 0.265,
+      "step": 35365
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3980692150624524e-05,
+      "loss": 0.2759,
+      "step": 35370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.397883800804749e-05,
+      "loss": 0.2845,
+      "step": 35375
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.397698361904274e-05,
+      "loss": 0.2766,
+      "step": 35380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.397512898363437e-05,
+      "loss": 0.2817,
+      "step": 35385
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.397327410184646e-05,
+      "loss": 0.293,
+      "step": 35390
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3971418973703085e-05,
+      "loss": 0.3059,
+      "step": 35395
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.396956359922834e-05,
+      "loss": 0.2993,
+      "step": 35400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3967707978446316e-05,
+      "loss": 0.2904,
+      "step": 35405
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.39658521113811e-05,
+      "loss": 0.2725,
+      "step": 35410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3963995998056795e-05,
+      "loss": 0.2539,
+      "step": 35415
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.396213963849749e-05,
+      "loss": 0.3026,
+      "step": 35420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3960283032727313e-05,
+      "loss": 0.3028,
+      "step": 35425
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.395842618077034e-05,
+      "loss": 0.2885,
+      "step": 35430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.395656908265071e-05,
+      "loss": 0.2749,
+      "step": 35435
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3954711738392526e-05,
+      "loss": 0.3019,
+      "step": 35440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.395285414801989e-05,
+      "loss": 0.3046,
+      "step": 35445
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.395099631155694e-05,
+      "loss": 0.2637,
+      "step": 35450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3949138229027795e-05,
+      "loss": 0.2905,
+      "step": 35455
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.394727990045657e-05,
+      "loss": 0.2931,
+      "step": 35460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.394542132586741e-05,
+      "loss": 0.3078,
+      "step": 35465
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.394356250528443e-05,
+      "loss": 0.2762,
+      "step": 35470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.394170343873178e-05,
+      "loss": 0.2756,
+      "step": 35475
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.39398441262336e-05,
+      "loss": 0.2865,
+      "step": 35480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393798456781402e-05,
+      "loss": 0.2863,
+      "step": 35485
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393612476349719e-05,
+      "loss": 0.2765,
+      "step": 35490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393426471330726e-05,
+      "loss": 0.2794,
+      "step": 35495
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393240441726838e-05,
+      "loss": 0.3137,
+      "step": 35500
+    },
+    {
+      "epoch": 1.25,
+      "eval_loss": 0.285795658826828,
+      "eval_runtime": 10.5526,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 35500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.393054387540471e-05,
+      "loss": 0.2958,
+      "step": 35505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3928683087740394e-05,
+      "loss": 0.27,
+      "step": 35510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.392682205429962e-05,
+      "loss": 0.296,
+      "step": 35515
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.392496077510652e-05,
+      "loss": 0.3007,
+      "step": 35520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.392309925018528e-05,
+      "loss": 0.3007,
+      "step": 35525
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3921237479560066e-05,
+      "loss": 0.2779,
+      "step": 35530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3919375463255055e-05,
+      "loss": 0.2885,
+      "step": 35535
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391751320129442e-05,
+      "loss": 0.2823,
+      "step": 35540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3915650693702345e-05,
+      "loss": 0.2783,
+      "step": 35545
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391378794050302e-05,
+      "loss": 0.29,
+      "step": 35550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391192494172061e-05,
+      "loss": 0.2846,
+      "step": 35555
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.391006169737932e-05,
+      "loss": 0.2778,
+      "step": 35560
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390819820750334e-05,
+      "loss": 0.2847,
+      "step": 35565
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390633447211687e-05,
+      "loss": 0.3156,
+      "step": 35570
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390447049124411e-05,
+      "loss": 0.3048,
+      "step": 35575
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390260626490926e-05,
+      "loss": 0.2952,
+      "step": 35580
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.390074179313652e-05,
+      "loss": 0.2893,
+      "step": 35585
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.389887707595011e-05,
+      "loss": 0.3191,
+      "step": 35590
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3897012113374224e-05,
+      "loss": 0.2988,
+      "step": 35595
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3895146905433096e-05,
+      "loss": 0.3018,
+      "step": 35600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.389328145215094e-05,
+      "loss": 0.2972,
+      "step": 35605
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.389141575355198e-05,
+      "loss": 0.2698,
+      "step": 35610
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3889549809660425e-05,
+      "loss": 0.2763,
+      "step": 35615
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.388768362050052e-05,
+      "loss": 0.278,
+      "step": 35620
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.388581718609649e-05,
+      "loss": 0.2908,
+      "step": 35625
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.388395050647257e-05,
+      "loss": 0.2866,
+      "step": 35630
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3882083581652996e-05,
+      "loss": 0.3039,
+      "step": 35635
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.3880216411662015e-05,
+      "loss": 0.2785,
+      "step": 35640
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387834899652387e-05,
+      "loss": 0.2908,
+      "step": 35645
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.38764813362628e-05,
+      "loss": 0.3119,
+      "step": 35650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387461343090305e-05,
+      "loss": 0.3132,
+      "step": 35655
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.38727452804689e-05,
+      "loss": 0.2776,
+      "step": 35660
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.387087688498458e-05,
+      "loss": 0.2854,
+      "step": 35665
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.386900824447436e-05,
+      "loss": 0.2701,
+      "step": 35670
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3867139358962506e-05,
+      "loss": 0.2914,
+      "step": 35675
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.386527022847328e-05,
+      "loss": 0.2743,
+      "step": 35680
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3863400853030954e-05,
+      "loss": 0.3009,
+      "step": 35685
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3861531232659796e-05,
+      "loss": 0.2845,
+      "step": 35690
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3859661367384084e-05,
+      "loss": 0.2956,
+      "step": 35695
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.385779125722811e-05,
+      "loss": 0.2896,
+      "step": 35700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.385592090221613e-05,
+      "loss": 0.2897,
+      "step": 35705
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3854050302372453e-05,
+      "loss": 0.299,
+      "step": 35710
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3852179457721354e-05,
+      "loss": 0.2715,
+      "step": 35715
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.385030836828713e-05,
+      "loss": 0.3069,
+      "step": 35720
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.384843703409407e-05,
+      "loss": 0.2927,
+      "step": 35725
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.384656545516649e-05,
+      "loss": 0.3173,
+      "step": 35730
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.384469363152867e-05,
+      "loss": 0.2901,
+      "step": 35735
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3842821563204916e-05,
+      "loss": 0.2724,
+      "step": 35740
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3840949250219546e-05,
+      "loss": 0.2853,
+      "step": 35745
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.383907669259687e-05,
+      "loss": 0.2798,
+      "step": 35750
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.383720389036119e-05,
+      "loss": 0.2963,
+      "step": 35755
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.383533084353684e-05,
+      "loss": 0.2845,
+      "step": 35760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.383345755214813e-05,
+      "loss": 0.2914,
+      "step": 35765
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3831584016219385e-05,
+      "loss": 0.2825,
+      "step": 35770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382971023577493e-05,
+      "loss": 0.3114,
+      "step": 35775
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382783621083911e-05,
+      "loss": 0.3057,
+      "step": 35780
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3825961941436236e-05,
+      "loss": 0.27,
+      "step": 35785
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3824087427590645e-05,
+      "loss": 0.2814,
+      "step": 35790
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382221266932669e-05,
+      "loss": 0.2841,
+      "step": 35795
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.382033766666871e-05,
+      "loss": 0.2814,
+      "step": 35800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381846241964105e-05,
+      "loss": 0.2871,
+      "step": 35805
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381658692826805e-05,
+      "loss": 0.2807,
+      "step": 35810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3814711192574075e-05,
+      "loss": 0.2895,
+      "step": 35815
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.381283521258347e-05,
+      "loss": 0.3185,
+      "step": 35820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.38109589883206e-05,
+      "loss": 0.2811,
+      "step": 35825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3809082519809826e-05,
+      "loss": 0.291,
+      "step": 35830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380720580707551e-05,
+      "loss": 0.2967,
+      "step": 35835
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380532885014202e-05,
+      "loss": 0.2848,
+      "step": 35840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380345164903373e-05,
+      "loss": 0.2688,
+      "step": 35845
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.380157420377501e-05,
+      "loss": 0.2852,
+      "step": 35850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379969651439024e-05,
+      "loss": 0.297,
+      "step": 35855
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3797818580903806e-05,
+      "loss": 0.2875,
+      "step": 35860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3795940403340076e-05,
+      "loss": 0.2677,
+      "step": 35865
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379406198172345e-05,
+      "loss": 0.2658,
+      "step": 35870
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3792183316078324e-05,
+      "loss": 0.2732,
+      "step": 35875
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.379030440642907e-05,
+      "loss": 0.3065,
+      "step": 35880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3788425252800104e-05,
+      "loss": 0.294,
+      "step": 35885
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378654585521582e-05,
+      "loss": 0.2804,
+      "step": 35890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378466621370061e-05,
+      "loss": 0.2957,
+      "step": 35895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378278632827889e-05,
+      "loss": 0.2936,
+      "step": 35900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.378090619897508e-05,
+      "loss": 0.2891,
+      "step": 35905
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.377902582581357e-05,
+      "loss": 0.267,
+      "step": 35910
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.377714520881879e-05,
+      "loss": 0.3045,
+      "step": 35915
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.377526434801515e-05,
+      "loss": 0.3042,
+      "step": 35920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.377338324342708e-05,
+      "loss": 0.3083,
+      "step": 35925
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.377150189507899e-05,
+      "loss": 0.292,
+      "step": 35930
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376962030299533e-05,
+      "loss": 0.311,
+      "step": 35935
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376773846720051e-05,
+      "loss": 0.2774,
+      "step": 35940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3765856387718984e-05,
+      "loss": 0.2974,
+      "step": 35945
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.3763974064575175e-05,
+      "loss": 0.2974,
+      "step": 35950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.376209149779353e-05,
+      "loss": 0.2679,
+      "step": 35955
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.37602086873985e-05,
+      "loss": 0.294,
+      "step": 35960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375832563341451e-05,
+      "loss": 0.3024,
+      "step": 35965
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3756442335866034e-05,
+      "loss": 0.267,
+      "step": 35970
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3754558794777514e-05,
+      "loss": 0.291,
+      "step": 35975
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.37526750101734e-05,
+      "loss": 0.2738,
+      "step": 35980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.375079098207816e-05,
+      "loss": 0.2847,
+      "step": 35985
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374890671051626e-05,
+      "loss": 0.2949,
+      "step": 35990
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3747022195512165e-05,
+      "loss": 0.3078,
+      "step": 35995
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374513743709034e-05,
+      "loss": 0.2833,
+      "step": 36000
+    },
+    {
+      "epoch": 1.27,
+      "eval_loss": 0.2862606644630432,
+      "eval_runtime": 10.5286,
+      "eval_samples_per_second": 9.498,
+      "eval_steps_per_second": 9.498,
+      "step": 36000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.374325243527526e-05,
+      "loss": 0.2925,
+      "step": 36005
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3741367190091396e-05,
+      "loss": 0.2695,
+      "step": 36010
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3739481701563236e-05,
+      "loss": 0.2741,
+      "step": 36015
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3737595969715254e-05,
+      "loss": 0.3076,
+      "step": 36020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3735709994571924e-05,
+      "loss": 0.2953,
+      "step": 36025
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3733823776157764e-05,
+      "loss": 0.2976,
+      "step": 36030
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373193731449724e-05,
+      "loss": 0.3005,
+      "step": 36035
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.373005060961486e-05,
+      "loss": 0.3054,
+      "step": 36040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372816366153511e-05,
+      "loss": 0.288,
+      "step": 36045
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.37262764702825e-05,
+      "loss": 0.2644,
+      "step": 36050
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372438903588153e-05,
+      "loss": 0.3041,
+      "step": 36055
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3722501358356705e-05,
+      "loss": 0.2938,
+      "step": 36060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.372061343773255e-05,
+      "loss": 0.2867,
+      "step": 36065
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3718725274033556e-05,
+      "loss": 0.2751,
+      "step": 36070
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371683686728426e-05,
+      "loss": 0.3224,
+      "step": 36075
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371494821750916e-05,
+      "loss": 0.2743,
+      "step": 36080
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3713059324732796e-05,
+      "loss": 0.3075,
+      "step": 36085
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.371117018897969e-05,
+      "loss": 0.2818,
+      "step": 36090
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370928081027437e-05,
+      "loss": 0.2828,
+      "step": 36095
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370739118864137e-05,
+      "loss": 0.3118,
+      "step": 36100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3705501324105224e-05,
+      "loss": 0.2779,
+      "step": 36105
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370361121669047e-05,
+      "loss": 0.2905,
+      "step": 36110
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.370172086642165e-05,
+      "loss": 0.3131,
+      "step": 36115
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369983027332331e-05,
+      "loss": 0.2999,
+      "step": 36120
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369793943742e-05,
+      "loss": 0.3009,
+      "step": 36125
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.369604835873626e-05,
+      "loss": 0.281,
+      "step": 36130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3694157037296665e-05,
+      "loss": 0.2928,
+      "step": 36135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3692265473125756e-05,
+      "loss": 0.3069,
+      "step": 36140
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3690373666248094e-05,
+      "loss": 0.3013,
+      "step": 36145
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3688481616688255e-05,
+      "loss": 0.2989,
+      "step": 36150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3686589324470796e-05,
+      "loss": 0.2663,
+      "step": 36155
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368469678962028e-05,
+      "loss": 0.2916,
+      "step": 36160
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3682804012161305e-05,
+      "loss": 0.2982,
+      "step": 36165
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.368091099211843e-05,
+      "loss": 0.3024,
+      "step": 36170
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3679017729516226e-05,
+      "loss": 0.2766,
+      "step": 36175
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367712422437929e-05,
+      "loss": 0.3051,
+      "step": 36180
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3675230476732206e-05,
+      "loss": 0.2863,
+      "step": 36185
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367333648659956e-05,
+      "loss": 0.3048,
+      "step": 36190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3671442254005955e-05,
+      "loss": 0.2668,
+      "step": 36195
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366954777897597e-05,
+      "loss": 0.2889,
+      "step": 36200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366765306153421e-05,
+      "loss": 0.2853,
+      "step": 36205
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366575810170528e-05,
+      "loss": 0.2799,
+      "step": 36210
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.3663862899513784e-05,
+      "loss": 0.312,
+      "step": 36215
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366196745498432e-05,
+      "loss": 0.2744,
+      "step": 36220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.366007176814152e-05,
+      "loss": 0.3083,
+      "step": 36225
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.365817583900998e-05,
+      "loss": 0.3109,
+      "step": 36230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.365627966761432e-05,
+      "loss": 0.2955,
+      "step": 36235
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3654383253979163e-05,
+      "loss": 0.2991,
+      "step": 36240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365248659812914e-05,
+      "loss": 0.2743,
+      "step": 36245
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.365058970008886e-05,
+      "loss": 0.2983,
+      "step": 36250
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364869255988298e-05,
+      "loss": 0.2837,
+      "step": 36255
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36467951775361e-05,
+      "loss": 0.2954,
+      "step": 36260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3644897553072885e-05,
+      "loss": 0.3065,
+      "step": 36265
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364299968651795e-05,
+      "loss": 0.2985,
+      "step": 36270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.364110157789596e-05,
+      "loss": 0.2885,
+      "step": 36275
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363920322723155e-05,
+      "loss": 0.2916,
+      "step": 36280
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363730463454937e-05,
+      "loss": 0.301,
+      "step": 36285
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3635405799874076e-05,
+      "loss": 0.2864,
+      "step": 36290
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.363350672323031e-05,
+      "loss": 0.303,
+      "step": 36295
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3631607404642747e-05,
+      "loss": 0.2987,
+      "step": 36300
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362970784413604e-05,
+      "loss": 0.2927,
+      "step": 36305
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362780804173484e-05,
+      "loss": 0.2928,
+      "step": 36310
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.362590799746384e-05,
+      "loss": 0.2916,
+      "step": 36315
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36240077113477e-05,
+      "loss": 0.2699,
+      "step": 36320
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3622107183411096e-05,
+      "loss": 0.2861,
+      "step": 36325
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36202064136787e-05,
+      "loss": 0.2675,
+      "step": 36330
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3618305402175196e-05,
+      "loss": 0.2824,
+      "step": 36335
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.361640414892526e-05,
+      "loss": 0.2912,
+      "step": 36340
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3614502653953596e-05,
+      "loss": 0.2873,
+      "step": 36345
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3612600917284874e-05,
+      "loss": 0.2911,
+      "step": 36350
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3610698938943795e-05,
+      "loss": 0.2893,
+      "step": 36355
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.360879671895506e-05,
+      "loss": 0.3058,
+      "step": 36360
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3606894257343356e-05,
+      "loss": 0.2766,
+      "step": 36365
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.36049915541334e-05,
+      "loss": 0.2626,
+      "step": 36370
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.360308860934989e-05,
+      "loss": 0.2879,
+      "step": 36375
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.360118542301753e-05,
+      "loss": 0.2785,
+      "step": 36380
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3599281995161035e-05,
+      "loss": 0.2919,
+      "step": 36385
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.359737832580512e-05,
+      "loss": 0.2894,
+      "step": 36390
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.359547441497451e-05,
+      "loss": 0.2673,
+      "step": 36395
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.359357026269392e-05,
+      "loss": 0.2778,
+      "step": 36400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3591665868988064e-05,
+      "loss": 0.3002,
+      "step": 36405
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358976123388168e-05,
+      "loss": 0.3183,
+      "step": 36410
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3587856357399504e-05,
+      "loss": 0.276,
+      "step": 36415
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358595123956626e-05,
+      "loss": 0.3022,
+      "step": 36420
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358404588040669e-05,
+      "loss": 0.2946,
+      "step": 36425
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358214027994553e-05,
+      "loss": 0.2858,
+      "step": 36430
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.358023443820752e-05,
+      "loss": 0.3093,
+      "step": 36435
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.357832835521742e-05,
+      "loss": 0.2731,
+      "step": 36440
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.357642203099996e-05,
+      "loss": 0.2918,
+      "step": 36445
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3574515465579915e-05,
+      "loss": 0.3219,
+      "step": 36450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3572608658982016e-05,
+      "loss": 0.2716,
+      "step": 36455
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3570701611231045e-05,
+      "loss": 0.2929,
+      "step": 36460
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356879432235175e-05,
+      "loss": 0.3035,
+      "step": 36465
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356688679236889e-05,
+      "loss": 0.2904,
+      "step": 36470
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356497902130724e-05,
+      "loss": 0.3162,
+      "step": 36475
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356307100919158e-05,
+      "loss": 0.3035,
+      "step": 36480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.356116275604667e-05,
+      "loss": 0.3243,
+      "step": 36485
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.35592542618973e-05,
+      "loss": 0.3113,
+      "step": 36490
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.355734552676823e-05,
+      "loss": 0.2986,
+      "step": 36495
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.355543655068428e-05,
+      "loss": 0.2644,
+      "step": 36500
+    },
+    {
+      "epoch": 1.28,
+      "eval_loss": 0.28553855419158936,
+      "eval_runtime": 10.5423,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 36500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.35535273336702e-05,
+      "loss": 0.2938,
+      "step": 36505
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.35516178757508e-05,
+      "loss": 0.2965,
+      "step": 36510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3549708176950864e-05,
+      "loss": 0.2701,
+      "step": 36515
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.35477982372952e-05,
+      "loss": 0.2727,
+      "step": 36520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3545888056808595e-05,
+      "loss": 0.3053,
+      "step": 36525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3543977635515856e-05,
+      "loss": 0.2859,
+      "step": 36530
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3542066973441784e-05,
+      "loss": 0.2769,
+      "step": 36535
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3540156070611194e-05,
+      "loss": 0.2967,
+      "step": 36540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35382449270489e-05,
+      "loss": 0.2793,
+      "step": 36545
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3536333542779704e-05,
+      "loss": 0.3308,
+      "step": 36550
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.353442191782844e-05,
+      "loss": 0.2875,
+      "step": 36555
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3532510052219924e-05,
+      "loss": 0.2916,
+      "step": 36560
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.353059794597898e-05,
+      "loss": 0.3129,
+      "step": 36565
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3528685599130427e-05,
+      "loss": 0.2945,
+      "step": 36570
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35267730116991e-05,
+      "loss": 0.3076,
+      "step": 36575
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3524860183709836e-05,
+      "loss": 0.2925,
+      "step": 36580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.352294711518748e-05,
+      "loss": 0.3183,
+      "step": 36585
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.352103380615686e-05,
+      "loss": 0.3042,
+      "step": 36590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3519120256642814e-05,
+      "loss": 0.2945,
+      "step": 36595
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.35172064666702e-05,
+      "loss": 0.3112,
+      "step": 36600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3515292436263863e-05,
+      "loss": 0.2821,
+      "step": 36605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.351337816544866e-05,
+      "loss": 0.2871,
+      "step": 36610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.351146365424944e-05,
+      "loss": 0.3181,
+      "step": 36615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.350954890269106e-05,
+      "loss": 0.2962,
+      "step": 36620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3507633910798394e-05,
+      "loss": 0.2991,
+      "step": 36625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3505718678596294e-05,
+      "loss": 0.3041,
+      "step": 36630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.350380320610963e-05,
+      "loss": 0.301,
+      "step": 36635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.350188749336328e-05,
+      "loss": 0.2916,
+      "step": 36640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.349997154038211e-05,
+      "loss": 0.2933,
+      "step": 36645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3498055347191005e-05,
+      "loss": 0.2878,
+      "step": 36650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3496138913814837e-05,
+      "loss": 0.3054,
+      "step": 36655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.34942222402785e-05,
+      "loss": 0.2858,
+      "step": 36660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3492305326606876e-05,
+      "loss": 0.2823,
+      "step": 36665
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.349038817282485e-05,
+      "loss": 0.2597,
+      "step": 36670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3488470778957316e-05,
+      "loss": 0.2985,
+      "step": 36675
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.348655314502919e-05,
+      "loss": 0.2786,
+      "step": 36680
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.348463527106533e-05,
+      "loss": 0.2555,
+      "step": 36685
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.348271715709068e-05,
+      "loss": 0.2963,
+      "step": 36690
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.348079880313013e-05,
+      "loss": 0.2941,
+      "step": 36695
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.347888020920858e-05,
+      "loss": 0.2925,
+      "step": 36700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3476961375350945e-05,
+      "loss": 0.2928,
+      "step": 36705
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3475042301582145e-05,
+      "loss": 0.28,
+      "step": 36710
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.347312298792711e-05,
+      "loss": 0.301,
+      "step": 36715
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3471203434410726e-05,
+      "loss": 0.2977,
+      "step": 36720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3469283641057954e-05,
+      "loss": 0.2834,
+      "step": 36725
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3467363607893695e-05,
+      "loss": 0.2851,
+      "step": 36730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.34654433349429e-05,
+      "loss": 0.2715,
+      "step": 36735
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.346352282223048e-05,
+      "loss": 0.2796,
+      "step": 36740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3461602069781396e-05,
+      "loss": 0.3012,
+      "step": 36745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3459681077620576e-05,
+      "loss": 0.2873,
+      "step": 36750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.345775984577295e-05,
+      "loss": 0.2934,
+      "step": 36755
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.345583837426349e-05,
+      "loss": 0.2894,
+      "step": 36760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.345391666311712e-05,
+      "loss": 0.2864,
+      "step": 36765
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3451994712358815e-05,
+      "loss": 0.2974,
+      "step": 36770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.345007252201351e-05,
+      "loss": 0.2784,
+      "step": 36775
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3448150092106185e-05,
+      "loss": 0.3068,
+      "step": 36780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.344622742266178e-05,
+      "loss": 0.2844,
+      "step": 36785
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3444689114656336e-05,
+      "loss": 0.2789,
+      "step": 36790
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.344276601410812e-05,
+      "loss": 0.2819,
+      "step": 36795
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.3440842674092734e-05,
+      "loss": 0.2916,
+      "step": 36800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.343891909463517e-05,
+      "loss": 0.3131,
+      "step": 36805
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3436995275760376e-05,
+      "loss": 0.2675,
+      "step": 36810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.343507121749336e-05,
+      "loss": 0.2875,
+      "step": 36815
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3433146919859094e-05,
+      "loss": 0.2685,
+      "step": 36820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.343122238288257e-05,
+      "loss": 0.2749,
+      "step": 36825
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3429297606588747e-05,
+      "loss": 0.2847,
+      "step": 36830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.342737259100266e-05,
+      "loss": 0.2959,
+      "step": 36835
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3425447336149275e-05,
+      "loss": 0.2964,
+      "step": 36840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3423521842053606e-05,
+      "loss": 0.2817,
+      "step": 36845
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.342159610874063e-05,
+      "loss": 0.2893,
+      "step": 36850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.341967013623539e-05,
+      "loss": 0.2597,
+      "step": 36855
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.341774392456286e-05,
+      "loss": 0.3009,
+      "step": 36860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3415817473748074e-05,
+      "loss": 0.2799,
+      "step": 36865
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3413890783816026e-05,
+      "loss": 0.2736,
+      "step": 36870
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3411963854791745e-05,
+      "loss": 0.3029,
+      "step": 36875
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.341003668670025e-05,
+      "loss": 0.2777,
+      "step": 36880
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.340810927956656e-05,
+      "loss": 0.2807,
+      "step": 36885
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.340618163341569e-05,
+      "loss": 0.2852,
+      "step": 36890
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.340425374827269e-05,
+      "loss": 0.2986,
+      "step": 36895
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3402325624162585e-05,
+      "loss": 0.2926,
+      "step": 36900
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.340039726111041e-05,
+      "loss": 0.273,
+      "step": 36905
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3398468659141206e-05,
+      "loss": 0.2762,
+      "step": 36910
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3396539818280004e-05,
+      "loss": 0.2821,
+      "step": 36915
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.339461073855186e-05,
+      "loss": 0.3022,
+      "step": 36920
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3392681419981815e-05,
+      "loss": 0.3041,
+      "step": 36925
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.339075186259493e-05,
+      "loss": 0.2892,
+      "step": 36930
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.338882206641625e-05,
+      "loss": 0.2866,
+      "step": 36935
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3386892031470826e-05,
+      "loss": 0.2874,
+      "step": 36940
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.338496175778374e-05,
+      "loss": 0.2866,
+      "step": 36945
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.338303124538004e-05,
+      "loss": 0.2786,
+      "step": 36950
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.338110049428478e-05,
+      "loss": 0.264,
+      "step": 36955
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3379169504523054e-05,
+      "loss": 0.2933,
+      "step": 36960
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.337723827611992e-05,
+      "loss": 0.2882,
+      "step": 36965
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3375306809100455e-05,
+      "loss": 0.276,
+      "step": 36970
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.337337510348975e-05,
+      "loss": 0.2903,
+      "step": 36975
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.337144315931288e-05,
+      "loss": 0.2877,
+      "step": 36980
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3369510976594917e-05,
+      "loss": 0.2836,
+      "step": 36985
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336757855536097e-05,
+      "loss": 0.3013,
+      "step": 36990
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336564589563611e-05,
+      "loss": 0.2935,
+      "step": 36995
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336371299744545e-05,
+      "loss": 0.295,
+      "step": 37000
+    },
+    {
+      "epoch": 1.3,
+      "eval_loss": 0.2844620645046234,
+      "eval_runtime": 10.5304,
+      "eval_samples_per_second": 9.496,
+      "eval_steps_per_second": 9.496,
+      "step": 37000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.336177986081407e-05,
+      "loss": 0.3063,
+      "step": 37005
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3359846485767083e-05,
+      "loss": 0.2739,
+      "step": 37010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.33579128723296e-05,
+      "loss": 0.2798,
+      "step": 37015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3355979020526704e-05,
+      "loss": 0.3129,
+      "step": 37020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.335404493038352e-05,
+      "loss": 0.2926,
+      "step": 37025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3352110601925155e-05,
+      "loss": 0.3111,
+      "step": 37030
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3350176035176734e-05,
+      "loss": 0.3063,
+      "step": 37035
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3348241230163367e-05,
+      "loss": 0.2934,
+      "step": 37040
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3346306186910185e-05,
+      "loss": 0.3079,
+      "step": 37045
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.33443709054423e-05,
+      "loss": 0.3001,
+      "step": 37050
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.334243538578486e-05,
+      "loss": 0.303,
+      "step": 37055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.334049962796297e-05,
+      "loss": 0.3096,
+      "step": 37060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.333856363200179e-05,
+      "loss": 0.2888,
+      "step": 37065
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3336627397926444e-05,
+      "loss": 0.294,
+      "step": 37070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.3334690925762074e-05,
+      "loss": 0.3129,
+      "step": 37075
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.333275421553382e-05,
+      "loss": 0.2793,
+      "step": 37080
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.333081726726684e-05,
+      "loss": 0.275,
+      "step": 37085
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.332888008098629e-05,
+      "loss": 0.2776,
+      "step": 37090
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.33269426567173e-05,
+      "loss": 0.286,
+      "step": 37095
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.332500499448503e-05,
+      "loss": 0.2853,
+      "step": 37100
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3323067094314654e-05,
+      "loss": 0.2795,
+      "step": 37105
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.332112895623133e-05,
+      "loss": 0.2884,
+      "step": 37110
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3319190580260215e-05,
+      "loss": 0.2788,
+      "step": 37115
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.331725196642649e-05,
+      "loss": 0.2847,
+      "step": 37120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.331531311475531e-05,
+      "loss": 0.3001,
+      "step": 37125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.331337402527187e-05,
+      "loss": 0.2949,
+      "step": 37130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.331143469800133e-05,
+      "loss": 0.3044,
+      "step": 37135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.330949513296888e-05,
+      "loss": 0.2916,
+      "step": 37140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.33075553301997e-05,
+      "loss": 0.2822,
+      "step": 37145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.330561528971898e-05,
+      "loss": 0.2897,
+      "step": 37150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.330367501155191e-05,
+      "loss": 0.2808,
+      "step": 37155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3301734495723687e-05,
+      "loss": 0.2806,
+      "step": 37160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.32997937422595e-05,
+      "loss": 0.2932,
+      "step": 37165
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3297852751184555e-05,
+      "loss": 0.2838,
+      "step": 37170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3295911522524044e-05,
+      "loss": 0.2988,
+      "step": 37175
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.329397005630318e-05,
+      "loss": 0.2963,
+      "step": 37180
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3292028352547176e-05,
+      "loss": 0.2902,
+      "step": 37185
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3290086411281234e-05,
+      "loss": 0.2953,
+      "step": 37190
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328814423253058e-05,
+      "loss": 0.2826,
+      "step": 37195
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3286201816320415e-05,
+      "loss": 0.2786,
+      "step": 37200
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328425916267598e-05,
+      "loss": 0.288,
+      "step": 37205
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328231627162248e-05,
+      "loss": 0.285,
+      "step": 37210
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.328037314318516e-05,
+      "loss": 0.2882,
+      "step": 37215
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327842977738924e-05,
+      "loss": 0.287,
+      "step": 37220
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327648617425995e-05,
+      "loss": 0.2864,
+      "step": 37225
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3274542333822544e-05,
+      "loss": 0.3049,
+      "step": 37230
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327259825610224e-05,
+      "loss": 0.2877,
+      "step": 37235
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.327065394112429e-05,
+      "loss": 0.2963,
+      "step": 37240
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3268709388913944e-05,
+      "loss": 0.2902,
+      "step": 37245
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326676459949644e-05,
+      "loss": 0.293,
+      "step": 37250
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326481957289704e-05,
+      "loss": 0.2932,
+      "step": 37255
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326287430914099e-05,
+      "loss": 0.3053,
+      "step": 37260
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.326092880825355e-05,
+      "loss": 0.2989,
+      "step": 37265
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3258983070259994e-05,
+      "loss": 0.2813,
+      "step": 37270
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.325703709518557e-05,
+      "loss": 0.2908,
+      "step": 37275
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.325509088305556e-05,
+      "loss": 0.2831,
+      "step": 37280
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.325314443389522e-05,
+      "loss": 0.2986,
+      "step": 37285
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.325119774772982e-05,
+      "loss": 0.2859,
+      "step": 37290
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.324925082458465e-05,
+      "loss": 0.2937,
+      "step": 37295
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3247303664485e-05,
+      "loss": 0.2865,
+      "step": 37300
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.324535626745612e-05,
+      "loss": 0.2826,
+      "step": 37305
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.324340863352332e-05,
+      "loss": 0.2826,
+      "step": 37310
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3241460762711874e-05,
+      "loss": 0.2433,
+      "step": 37315
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3239512655047086e-05,
+      "loss": 0.2867,
+      "step": 37320
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.323756431055425e-05,
+      "loss": 0.2893,
+      "step": 37325
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.323561572925866e-05,
+      "loss": 0.3029,
+      "step": 37330
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.323366691118561e-05,
+      "loss": 0.2687,
+      "step": 37335
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.323171785636042e-05,
+      "loss": 0.2693,
+      "step": 37340
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322976856480839e-05,
+      "loss": 0.2726,
+      "step": 37345
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322781903655482e-05,
+      "loss": 0.2934,
+      "step": 37350
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322586927162504e-05,
+      "loss": 0.2632,
+      "step": 37355
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322391927004435e-05,
+      "loss": 0.3015,
+      "step": 37360
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322196903183809e-05,
+      "loss": 0.2796,
+      "step": 37365
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.322001855703156e-05,
+      "loss": 0.292,
+      "step": 37370
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.3218067845650103e-05,
+      "loss": 0.2613,
+      "step": 37375
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.321611689771904e-05,
+      "loss": 0.2833,
+      "step": 37380
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.321416571326371e-05,
+      "loss": 0.3139,
+      "step": 37385
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3212214292309435e-05,
+      "loss": 0.2846,
+      "step": 37390
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.321026263488156e-05,
+      "loss": 0.2884,
+      "step": 37395
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.320831074100542e-05,
+      "loss": 0.2962,
+      "step": 37400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3206358610706376e-05,
+      "loss": 0.3143,
+      "step": 37405
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3204406244009754e-05,
+      "loss": 0.2929,
+      "step": 37410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3202453640940924e-05,
+      "loss": 0.2903,
+      "step": 37415
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.320050080152522e-05,
+      "loss": 0.2909,
+      "step": 37420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.319854772578801e-05,
+      "loss": 0.2865,
+      "step": 37425
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.319659441375465e-05,
+      "loss": 0.3007,
+      "step": 37430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.319464086545052e-05,
+      "loss": 0.2722,
+      "step": 37435
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.319268708090095e-05,
+      "loss": 0.3013,
+      "step": 37440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3190733060131326e-05,
+      "loss": 0.2961,
+      "step": 37445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.318877880316703e-05,
+      "loss": 0.3062,
+      "step": 37450
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.318682431003343e-05,
+      "loss": 0.2742,
+      "step": 37455
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.31848695807559e-05,
+      "loss": 0.2973,
+      "step": 37460
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.318291461535982e-05,
+      "loss": 0.2853,
+      "step": 37465
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.318095941387058e-05,
+      "loss": 0.2627,
+      "step": 37470
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3179003976313556e-05,
+      "loss": 0.2873,
+      "step": 37475
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3177048302714154e-05,
+      "loss": 0.2737,
+      "step": 37480
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.317509239309776e-05,
+      "loss": 0.2906,
+      "step": 37485
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3173136247489766e-05,
+      "loss": 0.2583,
+      "step": 37490
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.317117986591557e-05,
+      "loss": 0.3009,
+      "step": 37495
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.316922324840058e-05,
+      "loss": 0.2795,
+      "step": 37500
+    },
+    {
+      "epoch": 1.32,
+      "eval_loss": 0.28416377305984497,
+      "eval_runtime": 10.5464,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 37500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3167266394970206e-05,
+      "loss": 0.2812,
+      "step": 37505
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.316530930564985e-05,
+      "loss": 0.2841,
+      "step": 37510
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3163351980464916e-05,
+      "loss": 0.2898,
+      "step": 37515
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3161394419440836e-05,
+      "loss": 0.2741,
+      "step": 37520
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3159436622603013e-05,
+      "loss": 0.3133,
+      "step": 37525
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3157478589976874e-05,
+      "loss": 0.3004,
+      "step": 37530
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.315552032158784e-05,
+      "loss": 0.3081,
+      "step": 37535
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.315356181746134e-05,
+      "loss": 0.2947,
+      "step": 37540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.31516030776228e-05,
+      "loss": 0.3216,
+      "step": 37545
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.314964410209767e-05,
+      "loss": 0.2855,
+      "step": 37550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3147684890911365e-05,
+      "loss": 0.308,
+      "step": 37555
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.314572544408932e-05,
+      "loss": 0.3049,
+      "step": 37560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.314376576165701e-05,
+      "loss": 0.2787,
+      "step": 37565
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.314180584363984e-05,
+      "loss": 0.2797,
+      "step": 37570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.313984569006329e-05,
+      "loss": 0.3024,
+      "step": 37575
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.313788530095279e-05,
+      "loss": 0.2949,
+      "step": 37580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3135924676333806e-05,
+      "loss": 0.2951,
+      "step": 37585
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.313396381623179e-05,
+      "loss": 0.2992,
+      "step": 37590
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.313200272067221e-05,
+      "loss": 0.2875,
+      "step": 37595
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.313004138968052e-05,
+      "loss": 0.2947,
+      "step": 37600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3128079823282185e-05,
+      "loss": 0.2795,
+      "step": 37605
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.312611802150269e-05,
+      "loss": 0.29,
+      "step": 37610
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.31241559843675e-05,
+      "loss": 0.3024,
+      "step": 37615
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.312219371190208e-05,
+      "loss": 0.2777,
+      "step": 37620
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.3120231204131924e-05,
+      "loss": 0.2749,
+      "step": 37625
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.31182684610825e-05,
+      "loss": 0.2936,
+      "step": 37630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.311630548277931e-05,
+      "loss": 0.293,
+      "step": 37635
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.311434226924782e-05,
+      "loss": 0.305,
+      "step": 37640
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.311237882051354e-05,
+      "loss": 0.2852,
+      "step": 37645
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.311041513660195e-05,
+      "loss": 0.2913,
+      "step": 37650
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.310845121753857e-05,
+      "loss": 0.2762,
+      "step": 37655
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.310648706334888e-05,
+      "loss": 0.3097,
+      "step": 37660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.310452267405839e-05,
+      "loss": 0.2954,
+      "step": 37665
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.31025580496926e-05,
+      "loss": 0.2794,
+      "step": 37670
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.310059319027702e-05,
+      "loss": 0.3138,
+      "step": 37675
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.309862809583717e-05,
+      "loss": 0.27,
+      "step": 37680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.309666276639856e-05,
+      "loss": 0.3006,
+      "step": 37685
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.309469720198671e-05,
+      "loss": 0.2863,
+      "step": 37690
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.309273140262715e-05,
+      "loss": 0.2867,
+      "step": 37695
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3090765368345384e-05,
+      "loss": 0.2901,
+      "step": 37700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.308879909916696e-05,
+      "loss": 0.2751,
+      "step": 37705
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3086832595117397e-05,
+      "loss": 0.3008,
+      "step": 37710
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3084865856222234e-05,
+      "loss": 0.2783,
+      "step": 37715
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.308289888250701e-05,
+      "loss": 0.2733,
+      "step": 37720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3080931673997256e-05,
+      "loss": 0.2934,
+      "step": 37725
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.307896423071852e-05,
+      "loss": 0.2764,
+      "step": 37730
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.307699655269635e-05,
+      "loss": 0.2962,
+      "step": 37735
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3075028639956295e-05,
+      "loss": 0.2987,
+      "step": 37740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.30730604925239e-05,
+      "loss": 0.298,
+      "step": 37745
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3071092110424724e-05,
+      "loss": 0.2706,
+      "step": 37750
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.306912349368433e-05,
+      "loss": 0.2879,
+      "step": 37755
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.306715464232828e-05,
+      "loss": 0.2762,
+      "step": 37760
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3065185556382124e-05,
+      "loss": 0.2766,
+      "step": 37765
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.306321623587144e-05,
+      "loss": 0.305,
+      "step": 37770
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3061246680821806e-05,
+      "loss": 0.2775,
+      "step": 37775
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3059276891258784e-05,
+      "loss": 0.2644,
+      "step": 37780
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3057306867207946e-05,
+      "loss": 0.2929,
+      "step": 37785
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3055336608694884e-05,
+      "loss": 0.3116,
+      "step": 37790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.305336611574517e-05,
+      "loss": 0.2964,
+      "step": 37795
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3051395388384405e-05,
+      "loss": 0.2832,
+      "step": 37800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3049424426638156e-05,
+      "loss": 0.2944,
+      "step": 37805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.304745323053203e-05,
+      "loss": 0.2928,
+      "step": 37810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.304548180009162e-05,
+      "loss": 0.2924,
+      "step": 37815
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.304351013534252e-05,
+      "loss": 0.2803,
+      "step": 37820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.304153823631033e-05,
+      "loss": 0.3049,
+      "step": 37825
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3039566103020654e-05,
+      "loss": 0.2881,
+      "step": 37830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.30375937354991e-05,
+      "loss": 0.2739,
+      "step": 37835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.303562113377129e-05,
+      "loss": 0.2623,
+      "step": 37840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3033648297862814e-05,
+      "loss": 0.2819,
+      "step": 37845
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.303167522779931e-05,
+      "loss": 0.2592,
+      "step": 37850
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302970192360638e-05,
+      "loss": 0.3023,
+      "step": 37855
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3027728385309655e-05,
+      "loss": 0.2909,
+      "step": 37860
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302575461293476e-05,
+      "loss": 0.3025,
+      "step": 37865
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302378060650731e-05,
+      "loss": 0.2759,
+      "step": 37870
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.302180636605295e-05,
+      "loss": 0.2955,
+      "step": 37875
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.301983189159731e-05,
+      "loss": 0.2767,
+      "step": 37880
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.301785718316603e-05,
+      "loss": 0.3205,
+      "step": 37885
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3015882240784755e-05,
+      "loss": 0.3078,
+      "step": 37890
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.301390706447912e-05,
+      "loss": 0.2932,
+      "step": 37895
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3011931654274776e-05,
+      "loss": 0.292,
+      "step": 37900
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300995601019736e-05,
+      "loss": 0.2999,
+      "step": 37905
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300798013227254e-05,
+      "loss": 0.2601,
+      "step": 37910
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300600402052597e-05,
+      "loss": 0.2811,
+      "step": 37915
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3004027674983294e-05,
+      "loss": 0.2905,
+      "step": 37920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.3002051095670195e-05,
+      "loss": 0.2913,
+      "step": 37925
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.300007428261231e-05,
+      "loss": 0.2848,
+      "step": 37930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.299809723583534e-05,
+      "loss": 0.2817,
+      "step": 37935
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.2996119955364936e-05,
+      "loss": 0.2777,
+      "step": 37940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2994142441226764e-05,
+      "loss": 0.2785,
+      "step": 37945
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.299216469344652e-05,
+      "loss": 0.2951,
+      "step": 37950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2990186712049874e-05,
+      "loss": 0.3124,
+      "step": 37955
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2988208497062506e-05,
+      "loss": 0.3078,
+      "step": 37960
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.29862300485101e-05,
+      "loss": 0.2748,
+      "step": 37965
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.298425136641836e-05,
+      "loss": 0.2856,
+      "step": 37970
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2982272450812964e-05,
+      "loss": 0.2934,
+      "step": 37975
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2980293301719614e-05,
+      "loss": 0.293,
+      "step": 37980
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.297831391916401e-05,
+      "loss": 0.272,
+      "step": 37985
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.297633430317183e-05,
+      "loss": 0.2812,
+      "step": 37990
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.297435445376881e-05,
+      "loss": 0.2859,
+      "step": 37995
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.297237437098064e-05,
+      "loss": 0.3075,
+      "step": 38000
+    },
+    {
+      "epoch": 1.34,
+      "eval_loss": 0.28374922275543213,
+      "eval_runtime": 10.5368,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 38000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.297039405483303e-05,
+      "loss": 0.2702,
+      "step": 38005
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2968413505351704e-05,
+      "loss": 0.2736,
+      "step": 38010
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.296643272256237e-05,
+      "loss": 0.2756,
+      "step": 38015
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.296445170649074e-05,
+      "loss": 0.2682,
+      "step": 38020
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.296247045716254e-05,
+      "loss": 0.2696,
+      "step": 38025
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2960488974603506e-05,
+      "loss": 0.2754,
+      "step": 38030
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.295850725883936e-05,
+      "loss": 0.2919,
+      "step": 38035
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.295652530989583e-05,
+      "loss": 0.2839,
+      "step": 38040
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.295454312779865e-05,
+      "loss": 0.2828,
+      "step": 38045
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2952560712573565e-05,
+      "loss": 0.2947,
+      "step": 38050
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2950578064246305e-05,
+      "loss": 0.2927,
+      "step": 38055
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.294859518284263e-05,
+      "loss": 0.3115,
+      "step": 38060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.294661206838827e-05,
+      "loss": 0.3096,
+      "step": 38065
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.294462872090897e-05,
+      "loss": 0.2757,
+      "step": 38070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.29426451404305e-05,
+      "loss": 0.2621,
+      "step": 38075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2940661326978604e-05,
+      "loss": 0.2941,
+      "step": 38080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293867728057905e-05,
+      "loss": 0.288,
+      "step": 38085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2936693001257586e-05,
+      "loss": 0.295,
+      "step": 38090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293470848903999e-05,
+      "loss": 0.3041,
+      "step": 38095
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293272374395202e-05,
+      "loss": 0.3014,
+      "step": 38100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.293073876601945e-05,
+      "loss": 0.245,
+      "step": 38105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.292875355526806e-05,
+      "loss": 0.2874,
+      "step": 38110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2926768111723615e-05,
+      "loss": 0.2582,
+      "step": 38115
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.29247824354119e-05,
+      "loss": 0.307,
+      "step": 38120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.29227965263587e-05,
+      "loss": 0.2779,
+      "step": 38125
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.29208103845898e-05,
+      "loss": 0.3095,
+      "step": 38130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291882401013099e-05,
+      "loss": 0.2944,
+      "step": 38135
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291683740300805e-05,
+      "loss": 0.2929,
+      "step": 38140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291485056324679e-05,
+      "loss": 0.2837,
+      "step": 38145
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2912863490873e-05,
+      "loss": 0.2728,
+      "step": 38150
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.291087618591249e-05,
+      "loss": 0.2735,
+      "step": 38155
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.290888864839104e-05,
+      "loss": 0.2843,
+      "step": 38160
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.290690087833449e-05,
+      "loss": 0.267,
+      "step": 38165
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2904912875768624e-05,
+      "loss": 0.3167,
+      "step": 38170
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2902924640719267e-05,
+      "loss": 0.2905,
+      "step": 38175
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.290093617321223e-05,
+      "loss": 0.2805,
+      "step": 38180
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.289894747327333e-05,
+      "loss": 0.2999,
+      "step": 38185
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.28969585409284e-05,
+      "loss": 0.2963,
+      "step": 38190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.2894969376203257e-05,
+      "loss": 0.2979,
+      "step": 38195
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.289297997912373e-05,
+      "loss": 0.2835,
+      "step": 38200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.289099034971565e-05,
+      "loss": 0.2844,
+      "step": 38205
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288900048800485e-05,
+      "loss": 0.303,
+      "step": 38210
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288701039401717e-05,
+      "loss": 0.291,
+      "step": 38215
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288502006777844e-05,
+      "loss": 0.2979,
+      "step": 38220
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.288302950931452e-05,
+      "loss": 0.2875,
+      "step": 38225
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.288103871865124e-05,
+      "loss": 0.286,
+      "step": 38230
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.287904769581446e-05,
+      "loss": 0.2736,
+      "step": 38235
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.287705644083003e-05,
+      "loss": 0.277,
+      "step": 38240
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.28750649537238e-05,
+      "loss": 0.2884,
+      "step": 38245
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.287307323452163e-05,
+      "loss": 0.2813,
+      "step": 38250
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2871081283249395e-05,
+      "loss": 0.2952,
+      "step": 38255
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2869089099932936e-05,
+      "loss": 0.288,
+      "step": 38260
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2867096684598134e-05,
+      "loss": 0.2726,
+      "step": 38265
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.286510403727086e-05,
+      "loss": 0.3139,
+      "step": 38270
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2863111157976974e-05,
+      "loss": 0.3018,
+      "step": 38275
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.286111804674237e-05,
+      "loss": 0.2958,
+      "step": 38280
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2859124703592915e-05,
+      "loss": 0.2771,
+      "step": 38285
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.28571311285545e-05,
+      "loss": 0.3014,
+      "step": 38290
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2855137321653007e-05,
+      "loss": 0.2514,
+      "step": 38295
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.285314328291431e-05,
+      "loss": 0.3112,
+      "step": 38300
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.285114901236433e-05,
+      "loss": 0.2921,
+      "step": 38305
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.284915451002893e-05,
+      "loss": 0.2843,
+      "step": 38310
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.284715977593403e-05,
+      "loss": 0.282,
+      "step": 38315
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.284516481010552e-05,
+      "loss": 0.2915,
+      "step": 38320
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2843169612569306e-05,
+      "loss": 0.2836,
+      "step": 38325
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2841174183351294e-05,
+      "loss": 0.2788,
+      "step": 38330
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2839178522477394e-05,
+      "loss": 0.295,
+      "step": 38335
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.283718262997352e-05,
+      "loss": 0.2832,
+      "step": 38340
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2835186505865574e-05,
+      "loss": 0.3233,
+      "step": 38345
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.283319015017949e-05,
+      "loss": 0.2912,
+      "step": 38350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2831193562941185e-05,
+      "loss": 0.2738,
+      "step": 38355
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.282919674417658e-05,
+      "loss": 0.2797,
+      "step": 38360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2827199693911616e-05,
+      "loss": 0.3054,
+      "step": 38365
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.28252024121722e-05,
+      "loss": 0.2882,
+      "step": 38370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.282320489898428e-05,
+      "loss": 0.2655,
+      "step": 38375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2821207154373796e-05,
+      "loss": 0.2844,
+      "step": 38380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.281920917836668e-05,
+      "loss": 0.2826,
+      "step": 38385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2817210970988875e-05,
+      "loss": 0.2975,
+      "step": 38390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2815212532266325e-05,
+      "loss": 0.2714,
+      "step": 38395
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2813213862224986e-05,
+      "loss": 0.2979,
+      "step": 38400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.28112149608908e-05,
+      "loss": 0.2812,
+      "step": 38405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2809215828289725e-05,
+      "loss": 0.2823,
+      "step": 38410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2807216464447724e-05,
+      "loss": 0.282,
+      "step": 38415
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280521686939075e-05,
+      "loss": 0.3143,
+      "step": 38420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280321704314477e-05,
+      "loss": 0.3048,
+      "step": 38425
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.280121698573575e-05,
+      "loss": 0.3021,
+      "step": 38430
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.279921669718966e-05,
+      "loss": 0.3082,
+      "step": 38435
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.279721617753247e-05,
+      "loss": 0.2914,
+      "step": 38440
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2795215426790155e-05,
+      "loss": 0.2763,
+      "step": 38445
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.279321444498869e-05,
+      "loss": 0.2776,
+      "step": 38450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.279121323215407e-05,
+      "loss": 0.2933,
+      "step": 38455
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278921178831227e-05,
+      "loss": 0.2893,
+      "step": 38460
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278721011348927e-05,
+      "loss": 0.2816,
+      "step": 38465
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278520820771108e-05,
+      "loss": 0.2731,
+      "step": 38470
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278320607100368e-05,
+      "loss": 0.2938,
+      "step": 38475
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.278120370339307e-05,
+      "loss": 0.2871,
+      "step": 38480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.277920110490524e-05,
+      "loss": 0.2997,
+      "step": 38485
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2777198275566214e-05,
+      "loss": 0.2904,
+      "step": 38490
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.277519521540197e-05,
+      "loss": 0.2833,
+      "step": 38495
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.277319192443854e-05,
+      "loss": 0.2809,
+      "step": 38500
+    },
+    {
+      "epoch": 1.35,
+      "eval_loss": 0.28302499651908875,
+      "eval_runtime": 10.5824,
+      "eval_samples_per_second": 9.45,
+      "eval_steps_per_second": 9.45,
+      "step": 38500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.277118840270192e-05,
+      "loss": 0.3101,
+      "step": 38505
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.2769184650218133e-05,
+      "loss": 0.2771,
+      "step": 38510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.276718066701319e-05,
+      "loss": 0.278,
+      "step": 38515
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.276517645311312e-05,
+      "loss": 0.2896,
+      "step": 38520
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.276317200854394e-05,
+      "loss": 0.288,
+      "step": 38525
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.276116733333167e-05,
+      "loss": 0.2893,
+      "step": 38530
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.275916242750236e-05,
+      "loss": 0.294,
+      "step": 38535
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2757157291082026e-05,
+      "loss": 0.3026,
+      "step": 38540
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.27551519240967e-05,
+      "loss": 0.2939,
+      "step": 38545
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2753146326572436e-05,
+      "loss": 0.2771,
+      "step": 38550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.275114049853526e-05,
+      "loss": 0.3041,
+      "step": 38555
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.274913444001123e-05,
+      "loss": 0.2886,
+      "step": 38560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.274712815102638e-05,
+      "loss": 0.2751,
+      "step": 38565
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.274512163160677e-05,
+      "loss": 0.2692,
+      "step": 38570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.274311488177845e-05,
+      "loss": 0.3038,
+      "step": 38575
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.274110790156747e-05,
+      "loss": 0.281,
+      "step": 38580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.273910069099991e-05,
+      "loss": 0.2563,
+      "step": 38585
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.273709325010181e-05,
+      "loss": 0.286,
+      "step": 38590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2735085578899236e-05,
+      "loss": 0.2666,
+      "step": 38595
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2733077677418275e-05,
+      "loss": 0.2747,
+      "step": 38600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2731069545684974e-05,
+      "loss": 0.2701,
+      "step": 38605
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.272906118372544e-05,
+      "loss": 0.3037,
+      "step": 38610
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.272705259156572e-05,
+      "loss": 0.2786,
+      "step": 38615
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.27250437692319e-05,
+      "loss": 0.2938,
+      "step": 38620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.272303471675007e-05,
+      "loss": 0.2668,
+      "step": 38625
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.272102543414631e-05,
+      "loss": 0.2861,
+      "step": 38630
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2719015921446726e-05,
+      "loss": 0.2727,
+      "step": 38635
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2717006178677386e-05,
+      "loss": 0.2813,
+      "step": 38640
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.271499620586441e-05,
+      "loss": 0.2772,
+      "step": 38645
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.271298600303387e-05,
+      "loss": 0.3153,
+      "step": 38650
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2710975570211884e-05,
+      "loss": 0.3017,
+      "step": 38655
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2708964907424555e-05,
+      "loss": 0.3075,
+      "step": 38660
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2706954014697984e-05,
+      "loss": 0.2937,
+      "step": 38665
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2704942892058285e-05,
+      "loss": 0.2905,
+      "step": 38670
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2702931539531567e-05,
+      "loss": 0.2858,
+      "step": 38675
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.270091995714396e-05,
+      "loss": 0.2628,
+      "step": 38680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.269890814492156e-05,
+      "loss": 0.3051,
+      "step": 38685
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.269689610289051e-05,
+      "loss": 0.3129,
+      "step": 38690
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2694883831076924e-05,
+      "loss": 0.2792,
+      "step": 38695
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2692871329506936e-05,
+      "loss": 0.3053,
+      "step": 38700
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2690858598206665e-05,
+      "loss": 0.2971,
+      "step": 38705
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.268884563720226e-05,
+      "loss": 0.2868,
+      "step": 38710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2686832446519856e-05,
+      "loss": 0.2846,
+      "step": 38715
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2684819026185583e-05,
+      "loss": 0.2756,
+      "step": 38720
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.268280537622559e-05,
+      "loss": 0.291,
+      "step": 38725
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.268079149666603e-05,
+      "loss": 0.3187,
+      "step": 38730
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.267877738753303e-05,
+      "loss": 0.2888,
+      "step": 38735
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2676763048852764e-05,
+      "loss": 0.2993,
+      "step": 38740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2674748480651385e-05,
+      "loss": 0.2921,
+      "step": 38745
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.267273368295504e-05,
+      "loss": 0.2735,
+      "step": 38750
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2670718655789896e-05,
+      "loss": 0.296,
+      "step": 38755
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.266870339918211e-05,
+      "loss": 0.3009,
+      "step": 38760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.266668791315786e-05,
+      "loss": 0.2583,
+      "step": 38765
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.266467219774331e-05,
+      "loss": 0.28,
+      "step": 38770
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2662656252964635e-05,
+      "loss": 0.2812,
+      "step": 38775
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2660640078848e-05,
+      "loss": 0.3001,
+      "step": 38780
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.26586236754196e-05,
+      "loss": 0.2902,
+      "step": 38785
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.265660704270561e-05,
+      "loss": 0.2587,
+      "step": 38790
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.2654590180732215e-05,
+      "loss": 0.3016,
+      "step": 38795
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2652573089525594e-05,
+      "loss": 0.2793,
+      "step": 38800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2650555769111945e-05,
+      "loss": 0.2888,
+      "step": 38805
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2648538219517466e-05,
+      "loss": 0.3073,
+      "step": 38810
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2646520440768346e-05,
+      "loss": 0.2823,
+      "step": 38815
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.264450243289079e-05,
+      "loss": 0.3026,
+      "step": 38820
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2642484195911e-05,
+      "loss": 0.2741,
+      "step": 38825
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2640465729855186e-05,
+      "loss": 0.2949,
+      "step": 38830
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.263844703474954e-05,
+      "loss": 0.2495,
+      "step": 38835
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2636428110620274e-05,
+      "loss": 0.2771,
+      "step": 38840
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.263440895749363e-05,
+      "loss": 0.2698,
+      "step": 38845
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2632389575395795e-05,
+      "loss": 0.3005,
+      "step": 38850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.263036996435301e-05,
+      "loss": 0.2822,
+      "step": 38855
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.262835012439148e-05,
+      "loss": 0.2735,
+      "step": 38860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.262633005553745e-05,
+      "loss": 0.2999,
+      "step": 38865
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.262430975781714e-05,
+      "loss": 0.2825,
+      "step": 38870
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2622289231256776e-05,
+      "loss": 0.2627,
+      "step": 38875
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.26202684758826e-05,
+      "loss": 0.2937,
+      "step": 38880
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.261824749172086e-05,
+      "loss": 0.283,
+      "step": 38885
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2616226278797776e-05,
+      "loss": 0.2989,
+      "step": 38890
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2614204837139616e-05,
+      "loss": 0.2897,
+      "step": 38895
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.26121831667726e-05,
+      "loss": 0.3178,
+      "step": 38900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2610161267723005e-05,
+      "loss": 0.281,
+      "step": 38905
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.260813914001706e-05,
+      "loss": 0.2685,
+      "step": 38910
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2606116783681045e-05,
+      "loss": 0.291,
+      "step": 38915
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2604094198741195e-05,
+      "loss": 0.2739,
+      "step": 38920
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.260207138522379e-05,
+      "loss": 0.3218,
+      "step": 38925
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.260004834315509e-05,
+      "loss": 0.2865,
+      "step": 38930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.259802507256135e-05,
+      "loss": 0.3042,
+      "step": 38935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2596001573468864e-05,
+      "loss": 0.2767,
+      "step": 38940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.259397784590389e-05,
+      "loss": 0.2787,
+      "step": 38945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2591953889892706e-05,
+      "loss": 0.2751,
+      "step": 38950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.25899297054616e-05,
+      "loss": 0.2917,
+      "step": 38955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2587905292636845e-05,
+      "loss": 0.279,
+      "step": 38960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.258588065144473e-05,
+      "loss": 0.2934,
+      "step": 38965
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2583855781911544e-05,
+      "loss": 0.2789,
+      "step": 38970
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.258183068406358e-05,
+      "loss": 0.2685,
+      "step": 38975
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.257980535792713e-05,
+      "loss": 0.3244,
+      "step": 38980
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2577779803528495e-05,
+      "loss": 0.3174,
+      "step": 38985
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2575754020893966e-05,
+      "loss": 0.2866,
+      "step": 38990
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.257372801004986e-05,
+      "loss": 0.2718,
+      "step": 38995
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.257170177102248e-05,
+      "loss": 0.2912,
+      "step": 39000
+    },
+    {
+      "epoch": 1.37,
+      "eval_loss": 0.28303414583206177,
+      "eval_runtime": 10.5465,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 39000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.256967530383813e-05,
+      "loss": 0.3285,
+      "step": 39005
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2567648608523127e-05,
+      "loss": 0.2956,
+      "step": 39010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2565621685103784e-05,
+      "loss": 0.3043,
+      "step": 39015
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.256359453360642e-05,
+      "loss": 0.2776,
+      "step": 39020
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2561567154057355e-05,
+      "loss": 0.2852,
+      "step": 39025
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.255953954648291e-05,
+      "loss": 0.2754,
+      "step": 39030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.255751171090942e-05,
+      "loss": 0.3048,
+      "step": 39035
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.255548364736321e-05,
+      "loss": 0.2933,
+      "step": 39040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2553455355870615e-05,
+      "loss": 0.288,
+      "step": 39045
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2551426836457976e-05,
+      "loss": 0.2834,
+      "step": 39050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.254939808915162e-05,
+      "loss": 0.3058,
+      "step": 39055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.254736911397789e-05,
+      "loss": 0.275,
+      "step": 39060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2545339910963146e-05,
+      "loss": 0.2764,
+      "step": 39065
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.2543310480133725e-05,
+      "loss": 0.2942,
+      "step": 39070
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.254128082151597e-05,
+      "loss": 0.2854,
+      "step": 39075
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.253925093513626e-05,
+      "loss": 0.2886,
+      "step": 39080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2537220821020916e-05,
+      "loss": 0.2735,
+      "step": 39085
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.253519047919633e-05,
+      "loss": 0.2643,
+      "step": 39090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2533159909688844e-05,
+      "loss": 0.3042,
+      "step": 39095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.253112911252484e-05,
+      "loss": 0.2939,
+      "step": 39100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2529098087730665e-05,
+      "loss": 0.2816,
+      "step": 39105
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.252706683533272e-05,
+      "loss": 0.2897,
+      "step": 39110
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.252503535535735e-05,
+      "loss": 0.279,
+      "step": 39115
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.252300364783095e-05,
+      "loss": 0.2804,
+      "step": 39120
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.25209717127799e-05,
+      "loss": 0.2836,
+      "step": 39125
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2518939550230574e-05,
+      "loss": 0.2867,
+      "step": 39130
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.251690716020936e-05,
+      "loss": 0.3086,
+      "step": 39135
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2514874542742653e-05,
+      "loss": 0.2838,
+      "step": 39140
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2512841697856855e-05,
+      "loss": 0.2926,
+      "step": 39145
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.251080862557833e-05,
+      "loss": 0.293,
+      "step": 39150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.25087753259335e-05,
+      "loss": 0.2791,
+      "step": 39155
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.250674179894877e-05,
+      "loss": 0.2782,
+      "step": 39160
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.250470804465053e-05,
+      "loss": 0.3029,
+      "step": 39165
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.250267406306519e-05,
+      "loss": 0.2718,
+      "step": 39170
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.250063985421917e-05,
+      "loss": 0.2931,
+      "step": 39175
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2498605418138865e-05,
+      "loss": 0.2912,
+      "step": 39180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2496570754850704e-05,
+      "loss": 0.2946,
+      "step": 39185
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2494535864381104e-05,
+      "loss": 0.2748,
+      "step": 39190
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2492500746756484e-05,
+      "loss": 0.2726,
+      "step": 39195
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2490465402003273e-05,
+      "loss": 0.2799,
+      "step": 39200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248842983014789e-05,
+      "loss": 0.2898,
+      "step": 39205
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2486394031216766e-05,
+      "loss": 0.2735,
+      "step": 39210
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248435800523636e-05,
+      "loss": 0.2879,
+      "step": 39215
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248232175223307e-05,
+      "loss": 0.2592,
+      "step": 39220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.248028527223335e-05,
+      "loss": 0.2795,
+      "step": 39225
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.247824856526366e-05,
+      "loss": 0.2916,
+      "step": 39230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.247621163135043e-05,
+      "loss": 0.2857,
+      "step": 39235
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2474174470520103e-05,
+      "loss": 0.2693,
+      "step": 39240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.247213708279913e-05,
+      "loss": 0.2577,
+      "step": 39245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.247009946821398e-05,
+      "loss": 0.2838,
+      "step": 39250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.24680616267911e-05,
+      "loss": 0.2802,
+      "step": 39255
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.246602355855696e-05,
+      "loss": 0.2706,
+      "step": 39260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2463985263537995e-05,
+      "loss": 0.2977,
+      "step": 39265
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2461946741760704e-05,
+      "loss": 0.2698,
+      "step": 39270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2459907993251545e-05,
+      "loss": 0.2963,
+      "step": 39275
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.245786901803698e-05,
+      "loss": 0.2669,
+      "step": 39280
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.245582981614349e-05,
+      "loss": 0.3193,
+      "step": 39285
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.245379038759756e-05,
+      "loss": 0.2918,
+      "step": 39290
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2451750732425665e-05,
+      "loss": 0.2896,
+      "step": 39295
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.244971085065429e-05,
+      "loss": 0.295,
+      "step": 39300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.244767074230991e-05,
+      "loss": 0.2789,
+      "step": 39305
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.244563040741903e-05,
+      "loss": 0.2951,
+      "step": 39310
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.244358984600813e-05,
+      "loss": 0.2768,
+      "step": 39315
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.244154905810371e-05,
+      "loss": 0.302,
+      "step": 39320
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.243950804373228e-05,
+      "loss": 0.2966,
+      "step": 39325
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.243746680292033e-05,
+      "loss": 0.2965,
+      "step": 39330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2435425335694355e-05,
+      "loss": 0.2845,
+      "step": 39335
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2433383642080884e-05,
+      "loss": 0.3013,
+      "step": 39340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.243134172210641e-05,
+      "loss": 0.2755,
+      "step": 39345
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2429299575797454e-05,
+      "loss": 0.2901,
+      "step": 39350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.2427257203180526e-05,
+      "loss": 0.2814,
+      "step": 39355
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.242521460428215e-05,
+      "loss": 0.2813,
+      "step": 39360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.242317177912885e-05,
+      "loss": 0.2902,
+      "step": 39365
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2421128727747136e-05,
+      "loss": 0.2907,
+      "step": 39370
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2419085450163555e-05,
+      "loss": 0.2991,
+      "step": 39375
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2417041946404626e-05,
+      "loss": 0.2856,
+      "step": 39380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.241499821649689e-05,
+      "loss": 0.2923,
+      "step": 39385
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2412954260466876e-05,
+      "loss": 0.3112,
+      "step": 39390
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.241091007834113e-05,
+      "loss": 0.2896,
+      "step": 39395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.240886567014618e-05,
+      "loss": 0.3001,
+      "step": 39400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.240682103590859e-05,
+      "loss": 0.291,
+      "step": 39405
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2404776175654906e-05,
+      "loss": 0.2911,
+      "step": 39410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2402731089411664e-05,
+      "loss": 0.2845,
+      "step": 39415
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.240068577720543e-05,
+      "loss": 0.3,
+      "step": 39420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239864023906276e-05,
+      "loss": 0.2983,
+      "step": 39425
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239659447501021e-05,
+      "loss": 0.277,
+      "step": 39430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239454848507435e-05,
+      "loss": 0.2754,
+      "step": 39435
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239250226928174e-05,
+      "loss": 0.2756,
+      "step": 39440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.239045582765895e-05,
+      "loss": 0.2923,
+      "step": 39445
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2388409160232545e-05,
+      "loss": 0.2859,
+      "step": 39450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2386362267029116e-05,
+      "loss": 0.2837,
+      "step": 39455
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2384315148075225e-05,
+      "loss": 0.2795,
+      "step": 39460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2382267803397454e-05,
+      "loss": 0.2934,
+      "step": 39465
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.23802202330224e-05,
+      "loss": 0.2652,
+      "step": 39470
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2378172436976644e-05,
+      "loss": 0.3227,
+      "step": 39475
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2376124415286764e-05,
+      "loss": 0.2581,
+      "step": 39480
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.237407616797936e-05,
+      "loss": 0.2916,
+      "step": 39485
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.237202769508103e-05,
+      "loss": 0.2714,
+      "step": 39490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2369978996618365e-05,
+      "loss": 0.2715,
+      "step": 39495
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.236793007261798e-05,
+      "loss": 0.2822,
+      "step": 39500
+    },
+    {
+      "epoch": 1.39,
+      "eval_loss": 0.28314095735549927,
+      "eval_runtime": 10.5583,
+      "eval_samples_per_second": 9.471,
+      "eval_steps_per_second": 9.471,
+      "step": 39500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.236588092310646e-05,
+      "loss": 0.2802,
+      "step": 39505
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.236383154811042e-05,
+      "loss": 0.284,
+      "step": 39510
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.236178194765647e-05,
+      "loss": 0.3279,
+      "step": 39515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2359732121771226e-05,
+      "loss": 0.2795,
+      "step": 39520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.23576820704813e-05,
+      "loss": 0.2839,
+      "step": 39525
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.235563179381331e-05,
+      "loss": 0.271,
+      "step": 39530
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.235358129179389e-05,
+      "loss": 0.2893,
+      "step": 39535
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2351530564449635e-05,
+      "loss": 0.2983,
+      "step": 39540
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.23494796118072e-05,
+      "loss": 0.2856,
+      "step": 39545
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234742843389321e-05,
+      "loss": 0.2899,
+      "step": 39550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234537703073429e-05,
+      "loss": 0.2832,
+      "step": 39555
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234332540235708e-05,
+      "loss": 0.3104,
+      "step": 39560
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.234127354878822e-05,
+      "loss": 0.2856,
+      "step": 39565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2339221470054355e-05,
+      "loss": 0.2836,
+      "step": 39570
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2337169166182125e-05,
+      "loss": 0.2826,
+      "step": 39575
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.233511663719817e-05,
+      "loss": 0.2842,
+      "step": 39580
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2333063883129156e-05,
+      "loss": 0.2884,
+      "step": 39585
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.233101090400173e-05,
+      "loss": 0.2684,
+      "step": 39590
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232895769984255e-05,
+      "loss": 0.2712,
+      "step": 39595
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232690427067827e-05,
+      "loss": 0.2711,
+      "step": 39600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232485061653556e-05,
+      "loss": 0.3023,
+      "step": 39605
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232279673744108e-05,
+      "loss": 0.292,
+      "step": 39610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.232074263342151e-05,
+      "loss": 0.2956,
+      "step": 39615
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2318688304503505e-05,
+      "loss": 0.3008,
+      "step": 39620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.231663375071375e-05,
+      "loss": 0.2826,
+      "step": 39625
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.231457897207891e-05,
+      "loss": 0.2865,
+      "step": 39630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.231252396862568e-05,
+      "loss": 0.2763,
+      "step": 39635
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.2310468740380725e-05,
+      "loss": 0.315,
+      "step": 39640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.230841328737075e-05,
+      "loss": 0.3025,
+      "step": 39645
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.230635760962244e-05,
+      "loss": 0.2954,
+      "step": 39650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2304301707162475e-05,
+      "loss": 0.2741,
+      "step": 39655
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.230224558001755e-05,
+      "loss": 0.2777,
+      "step": 39660
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.230018922821438e-05,
+      "loss": 0.2835,
+      "step": 39665
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.229813265177965e-05,
+      "loss": 0.2953,
+      "step": 39670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2296075850740066e-05,
+      "loss": 0.2712,
+      "step": 39675
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.229401882512234e-05,
+      "loss": 0.3011,
+      "step": 39680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.229196157495318e-05,
+      "loss": 0.2996,
+      "step": 39685
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.228990410025928e-05,
+      "loss": 0.2695,
+      "step": 39690
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2287846401067374e-05,
+      "loss": 0.2822,
+      "step": 39695
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.228578847740419e-05,
+      "loss": 0.2804,
+      "step": 39700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2283730329296417e-05,
+      "loss": 0.2755,
+      "step": 39705
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.228167195677081e-05,
+      "loss": 0.3029,
+      "step": 39710
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.227961335985407e-05,
+      "loss": 0.2772,
+      "step": 39715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.227755453857295e-05,
+      "loss": 0.2997,
+      "step": 39720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.227549549295415e-05,
+      "loss": 0.2893,
+      "step": 39725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2273436223024445e-05,
+      "loss": 0.2749,
+      "step": 39730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.227137672881055e-05,
+      "loss": 0.2954,
+      "step": 39735
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2269317010339215e-05,
+      "loss": 0.2996,
+      "step": 39740
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.226725706763717e-05,
+      "loss": 0.2949,
+      "step": 39745
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2265196900731176e-05,
+      "loss": 0.296,
+      "step": 39750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.226313650964798e-05,
+      "loss": 0.288,
+      "step": 39755
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.226107589441432e-05,
+      "loss": 0.2819,
+      "step": 39760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.225901505505698e-05,
+      "loss": 0.2796,
+      "step": 39765
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2256953991602705e-05,
+      "loss": 0.262,
+      "step": 39770
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.225489270407825e-05,
+      "loss": 0.2852,
+      "step": 39775
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.225283119251039e-05,
+      "loss": 0.2634,
+      "step": 39780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.225076945692587e-05,
+      "loss": 0.2904,
+      "step": 39785
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.224870749735149e-05,
+      "loss": 0.2756,
+      "step": 39790
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.224664531381402e-05,
+      "loss": 0.2935,
+      "step": 39795
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.224458290634021e-05,
+      "loss": 0.2877,
+      "step": 39800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.224252027495686e-05,
+      "loss": 0.2858,
+      "step": 39805
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.224045741969075e-05,
+      "loss": 0.2794,
+      "step": 39810
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.223839434056866e-05,
+      "loss": 0.3027,
+      "step": 39815
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.223633103761738e-05,
+      "loss": 0.2658,
+      "step": 39820
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.22342675108637e-05,
+      "loss": 0.2767,
+      "step": 39825
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.223220376033442e-05,
+      "loss": 0.2956,
+      "step": 39830
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.223013978605633e-05,
+      "loss": 0.2936,
+      "step": 39835
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.222807558805622e-05,
+      "loss": 0.3004,
+      "step": 39840
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.222601116636091e-05,
+      "loss": 0.2876,
+      "step": 39845
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.22239465209972e-05,
+      "loss": 0.2749,
+      "step": 39850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.222188165199189e-05,
+      "loss": 0.293,
+      "step": 39855
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.221981655937179e-05,
+      "loss": 0.3175,
+      "step": 39860
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.221775124316374e-05,
+      "loss": 0.2766,
+      "step": 39865
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2215685703394525e-05,
+      "loss": 0.2864,
+      "step": 39870
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.221361994009098e-05,
+      "loss": 0.2821,
+      "step": 39875
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2211553953279914e-05,
+      "loss": 0.2932,
+      "step": 39880
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.220948774298817e-05,
+      "loss": 0.2625,
+      "step": 39885
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.220742130924257e-05,
+      "loss": 0.2784,
+      "step": 39890
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2205354652069946e-05,
+      "loss": 0.2753,
+      "step": 39895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2203287771497125e-05,
+      "loss": 0.3228,
+      "step": 39900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2201220667550955e-05,
+      "loss": 0.2661,
+      "step": 39905
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.219915334025827e-05,
+      "loss": 0.2924,
+      "step": 39910
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.219708578964591e-05,
+      "loss": 0.2912,
+      "step": 39915
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.2195018015740726e-05,
+      "loss": 0.2761,
+      "step": 39920
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.219295001856957e-05,
+      "loss": 0.2652,
+      "step": 39925
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.219088179815928e-05,
+      "loss": 0.2966,
+      "step": 39930
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.218881335453673e-05,
+      "loss": 0.2974,
+      "step": 39935
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.218674468772875e-05,
+      "loss": 0.2852,
+      "step": 39940
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.218467579776223e-05,
+      "loss": 0.2864,
+      "step": 39945
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.218260668466402e-05,
+      "loss": 0.2779,
+      "step": 39950
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.218053734846098e-05,
+      "loss": 0.2762,
+      "step": 39955
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2178467789179986e-05,
+      "loss": 0.2944,
+      "step": 39960
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217639800684791e-05,
+      "loss": 0.2905,
+      "step": 39965
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217432800149162e-05,
+      "loss": 0.2807,
+      "step": 39970
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.217267183664723e-05,
+      "loss": 0.2647,
+      "step": 39975
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2170601429915105e-05,
+      "loss": 0.3072,
+      "step": 39980
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2168530800234035e-05,
+      "loss": 0.2968,
+      "step": 39985
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216645994763091e-05,
+      "loss": 0.272,
+      "step": 39990
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2164388872132607e-05,
+      "loss": 0.2814,
+      "step": 39995
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216231757376603e-05,
+      "loss": 0.3099,
+      "step": 40000
+    },
+    {
+      "epoch": 1.41,
+      "eval_loss": 0.2824369966983795,
+      "eval_runtime": 10.5321,
+      "eval_samples_per_second": 9.495,
+      "eval_steps_per_second": 9.495,
+      "step": 40000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.216024605255807e-05,
+      "loss": 0.2774,
+      "step": 40005
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.215817430853563e-05,
+      "loss": 0.2988,
+      "step": 40010
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.21561023417256e-05,
+      "loss": 0.3029,
+      "step": 40015
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.215403015215489e-05,
+      "loss": 0.2871,
+      "step": 40020
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.21519577398504e-05,
+      "loss": 0.291,
+      "step": 40025
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214988510483904e-05,
+      "loss": 0.2806,
+      "step": 40030
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214781224714773e-05,
+      "loss": 0.286,
+      "step": 40035
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214573916680338e-05,
+      "loss": 0.2788,
+      "step": 40040
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2143665863832896e-05,
+      "loss": 0.2832,
+      "step": 40045
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.214159233826322e-05,
+      "loss": 0.2754,
+      "step": 40050
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2139518590121256e-05,
+      "loss": 0.2872,
+      "step": 40055
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213744461943395e-05,
+      "loss": 0.2831,
+      "step": 40060
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213537042622821e-05,
+      "loss": 0.2739,
+      "step": 40065
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213329601053098e-05,
+      "loss": 0.3059,
+      "step": 40070
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.213122137236919e-05,
+      "loss": 0.2987,
+      "step": 40075
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.212914651176978e-05,
+      "loss": 0.2718,
+      "step": 40080
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.212707142875969e-05,
+      "loss": 0.2723,
+      "step": 40085
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.212499612336587e-05,
+      "loss": 0.2886,
+      "step": 40090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.212292059561526e-05,
+      "loss": 0.3042,
+      "step": 40095
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2120844845534816e-05,
+      "loss": 0.3038,
+      "step": 40100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211876887315147e-05,
+      "loss": 0.293,
+      "step": 40105
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211669267849221e-05,
+      "loss": 0.2877,
+      "step": 40110
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211461626158396e-05,
+      "loss": 0.2785,
+      "step": 40115
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.211253962245371e-05,
+      "loss": 0.297,
+      "step": 40120
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.21104627611284e-05,
+      "loss": 0.2807,
+      "step": 40125
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.210838567763501e-05,
+      "loss": 0.3068,
+      "step": 40130
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.21063083720005e-05,
+      "loss": 0.2605,
+      "step": 40135
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.210423084425186e-05,
+      "loss": 0.2947,
+      "step": 40140
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2102153094416055e-05,
+      "loss": 0.3167,
+      "step": 40145
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2100075122520055e-05,
+      "loss": 0.2757,
+      "step": 40150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2097996928590854e-05,
+      "loss": 0.2864,
+      "step": 40155
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209591851265543e-05,
+      "loss": 0.308,
+      "step": 40160
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209383987474077e-05,
+      "loss": 0.3065,
+      "step": 40165
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.209176101487387e-05,
+      "loss": 0.3025,
+      "step": 40170
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2089681933081716e-05,
+      "loss": 0.2801,
+      "step": 40175
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.20876026293913e-05,
+      "loss": 0.308,
+      "step": 40180
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.208552310382963e-05,
+      "loss": 0.3071,
+      "step": 40185
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2083443356423694e-05,
+      "loss": 0.2772,
+      "step": 40190
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2081363387200504e-05,
+      "loss": 0.2632,
+      "step": 40195
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2079283196187077e-05,
+      "loss": 0.2998,
+      "step": 40200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207720278341041e-05,
+      "loss": 0.2964,
+      "step": 40205
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.2075122148897516e-05,
+      "loss": 0.3021,
+      "step": 40210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.207304129267542e-05,
+      "loss": 0.295,
+      "step": 40215
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.207096021477113e-05,
+      "loss": 0.2857,
+      "step": 40220
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2068878915211665e-05,
+      "loss": 0.2762,
+      "step": 40225
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.206679739402406e-05,
+      "loss": 0.2891,
+      "step": 40230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2064715651235344e-05,
+      "loss": 0.2594,
+      "step": 40235
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2062633686872545e-05,
+      "loss": 0.2997,
+      "step": 40240
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2060551500962676e-05,
+      "loss": 0.2933,
+      "step": 40245
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2058469093532806e-05,
+      "loss": 0.2704,
+      "step": 40250
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.205638646460994e-05,
+      "loss": 0.2835,
+      "step": 40255
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2054303614221154e-05,
+      "loss": 0.2882,
+      "step": 40260
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.205222054239346e-05,
+      "loss": 0.2911,
+      "step": 40265
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.205013724915393e-05,
+      "loss": 0.2964,
+      "step": 40270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2048053734529604e-05,
+      "loss": 0.2897,
+      "step": 40275
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.204596999854753e-05,
+      "loss": 0.2759,
+      "step": 40280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.204388604123477e-05,
+      "loss": 0.2827,
+      "step": 40285
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.204180186261839e-05,
+      "loss": 0.2804,
+      "step": 40290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2039717462725435e-05,
+      "loss": 0.2762,
+      "step": 40295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.203763284158299e-05,
+      "loss": 0.3067,
+      "step": 40300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.203554799921809e-05,
+      "loss": 0.3022,
+      "step": 40305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.203346293565784e-05,
+      "loss": 0.3145,
+      "step": 40310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.203137765092929e-05,
+      "loss": 0.2803,
+      "step": 40315
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.202929214505954e-05,
+      "loss": 0.306,
+      "step": 40320
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.202720641807565e-05,
+      "loss": 0.2882,
+      "step": 40325
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.20251204700047e-05,
+      "loss": 0.2888,
+      "step": 40330
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.202303430087379e-05,
+      "loss": 0.286,
+      "step": 40335
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.202094791070998e-05,
+      "loss": 0.2982,
+      "step": 40340
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.201886129954039e-05,
+      "loss": 0.2865,
+      "step": 40345
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2016774467392106e-05,
+      "loss": 0.3012,
+      "step": 40350
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.201468741429222e-05,
+      "loss": 0.2862,
+      "step": 40355
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.201260014026782e-05,
+      "loss": 0.2935,
+      "step": 40360
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2010512645346034e-05,
+      "loss": 0.2814,
+      "step": 40365
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2008424929553934e-05,
+      "loss": 0.3095,
+      "step": 40370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.200633699291866e-05,
+      "loss": 0.2908,
+      "step": 40375
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.20042488354673e-05,
+      "loss": 0.2814,
+      "step": 40380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.200216045722698e-05,
+      "loss": 0.2764,
+      "step": 40385
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.200007185822481e-05,
+      "loss": 0.2903,
+      "step": 40390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.199798303848791e-05,
+      "loss": 0.2933,
+      "step": 40395
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.19958939980434e-05,
+      "loss": 0.2782,
+      "step": 40400
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.199380473691841e-05,
+      "loss": 0.2777,
+      "step": 40405
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.199171525514006e-05,
+      "loss": 0.2911,
+      "step": 40410
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.198962555273549e-05,
+      "loss": 0.2774,
+      "step": 40415
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1987535629731835e-05,
+      "loss": 0.3019,
+      "step": 40420
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1985445486156216e-05,
+      "loss": 0.2958,
+      "step": 40425
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.198335512203578e-05,
+      "loss": 0.2622,
+      "step": 40430
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1981264537397674e-05,
+      "loss": 0.2592,
+      "step": 40435
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.197917373226904e-05,
+      "loss": 0.2674,
+      "step": 40440
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1977082706677026e-05,
+      "loss": 0.2983,
+      "step": 40445
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.197499146064877e-05,
+      "loss": 0.2961,
+      "step": 40450
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.197289999421145e-05,
+      "loss": 0.2746,
+      "step": 40455
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1970808307392204e-05,
+      "loss": 0.2803,
+      "step": 40460
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1968716400218196e-05,
+      "loss": 0.2771,
+      "step": 40465
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1966624272716594e-05,
+      "loss": 0.2805,
+      "step": 40470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.196453192491455e-05,
+      "loss": 0.2842,
+      "step": 40475
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1962439356839236e-05,
+      "loss": 0.2983,
+      "step": 40480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.196034656851783e-05,
+      "loss": 0.2932,
+      "step": 40485
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.1958253559977506e-05,
+      "loss": 0.2799,
+      "step": 40490
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.195616033124543e-05,
+      "loss": 0.2829,
+      "step": 40495
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.195406688234879e-05,
+      "loss": 0.3083,
+      "step": 40500
+    },
+    {
+      "epoch": 1.42,
+      "eval_loss": 0.2803879678249359,
+      "eval_runtime": 10.5501,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 40500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.195197321331476e-05,
+      "loss": 0.3007,
+      "step": 40505
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194987932417054e-05,
+      "loss": 0.3135,
+      "step": 40510
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194778521494329e-05,
+      "loss": 0.2749,
+      "step": 40515
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194569088566023e-05,
+      "loss": 0.2535,
+      "step": 40520
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194359633634854e-05,
+      "loss": 0.2984,
+      "step": 40525
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.194150156703542e-05,
+      "loss": 0.2547,
+      "step": 40530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1939406577748065e-05,
+      "loss": 0.3001,
+      "step": 40535
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.193731136851368e-05,
+      "loss": 0.2765,
+      "step": 40540
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.193521593935946e-05,
+      "loss": 0.2817,
+      "step": 40545
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1933120290312636e-05,
+      "loss": 0.2948,
+      "step": 40550
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.19310244214004e-05,
+      "loss": 0.3034,
+      "step": 40555
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1928928332649965e-05,
+      "loss": 0.3064,
+      "step": 40560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.192683202408855e-05,
+      "loss": 0.2673,
+      "step": 40565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1924735495743384e-05,
+      "loss": 0.2798,
+      "step": 40570
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1922638747641676e-05,
+      "loss": 0.2704,
+      "step": 40575
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.192054177981065e-05,
+      "loss": 0.2958,
+      "step": 40580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.191844459227756e-05,
+      "loss": 0.2943,
+      "step": 40585
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1916347185069596e-05,
+      "loss": 0.2997,
+      "step": 40590
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1914249558214025e-05,
+      "loss": 0.3073,
+      "step": 40595
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.191215171173807e-05,
+      "loss": 0.3079,
+      "step": 40600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1910053645668965e-05,
+      "loss": 0.2856,
+      "step": 40605
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.190795536003396e-05,
+      "loss": 0.2899,
+      "step": 40610
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.190585685486029e-05,
+      "loss": 0.2806,
+      "step": 40615
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.190375813017522e-05,
+      "loss": 0.2648,
+      "step": 40620
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1901659186005996e-05,
+      "loss": 0.2655,
+      "step": 40625
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1899560022379856e-05,
+      "loss": 0.317,
+      "step": 40630
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.189746063932407e-05,
+      "loss": 0.2878,
+      "step": 40635
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.189536103686589e-05,
+      "loss": 0.2867,
+      "step": 40640
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.189326121503258e-05,
+      "loss": 0.2737,
+      "step": 40645
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1891161173851415e-05,
+      "loss": 0.2758,
+      "step": 40650
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.188906091334964e-05,
+      "loss": 0.2886,
+      "step": 40655
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.188696043355455e-05,
+      "loss": 0.2923,
+      "step": 40660
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.188485973449341e-05,
+      "loss": 0.2712,
+      "step": 40665
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1882758816193484e-05,
+      "loss": 0.3056,
+      "step": 40670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.188065767868207e-05,
+      "loss": 0.3079,
+      "step": 40675
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187855632198643e-05,
+      "loss": 0.279,
+      "step": 40680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187645474613387e-05,
+      "loss": 0.2859,
+      "step": 40685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187435295115166e-05,
+      "loss": 0.2993,
+      "step": 40690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.18722509370671e-05,
+      "loss": 0.2918,
+      "step": 40695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.187014870390749e-05,
+      "loss": 0.2852,
+      "step": 40700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.186804625170011e-05,
+      "loss": 0.2632,
+      "step": 40705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.186594358047226e-05,
+      "loss": 0.3012,
+      "step": 40710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1863840690251255e-05,
+      "loss": 0.2882,
+      "step": 40715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1861737581064386e-05,
+      "loss": 0.2928,
+      "step": 40720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1859634252938974e-05,
+      "loss": 0.2965,
+      "step": 40725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1857530705902316e-05,
+      "loss": 0.2917,
+      "step": 40730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.185542693998173e-05,
+      "loss": 0.2961,
+      "step": 40735
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1853322955204545e-05,
+      "loss": 0.2923,
+      "step": 40740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1851218751598056e-05,
+      "loss": 0.2859,
+      "step": 40745
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.18491143291896e-05,
+      "loss": 0.2739,
+      "step": 40750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.18470096880065e-05,
+      "loss": 0.2854,
+      "step": 40755
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1844904828076086e-05,
+      "loss": 0.2799,
+      "step": 40760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.184279974942568e-05,
+      "loss": 0.3104,
+      "step": 40765
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.184069445208262e-05,
+      "loss": 0.2642,
+      "step": 40770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.183858893607423e-05,
+      "loss": 0.2641,
+      "step": 40775
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.1836483201427876e-05,
+      "loss": 0.3153,
+      "step": 40780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.183437724817087e-05,
+      "loss": 0.2718,
+      "step": 40785
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.183227107633058e-05,
+      "loss": 0.2915,
+      "step": 40790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.183016468593434e-05,
+      "loss": 0.2779,
+      "step": 40795
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.18280580770095e-05,
+      "loss": 0.2813,
+      "step": 40800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.182595124958342e-05,
+      "loss": 0.2932,
+      "step": 40805
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1823844203683456e-05,
+      "loss": 0.2671,
+      "step": 40810
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.182173693933695e-05,
+      "loss": 0.3033,
+      "step": 40815
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1819629456571285e-05,
+      "loss": 0.3036,
+      "step": 40820
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.181752175541381e-05,
+      "loss": 0.3172,
+      "step": 40825
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1815413835891904e-05,
+      "loss": 0.2817,
+      "step": 40830
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.181330569803293e-05,
+      "loss": 0.2732,
+      "step": 40835
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.181119734186426e-05,
+      "loss": 0.2779,
+      "step": 40840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.180908876741327e-05,
+      "loss": 0.2961,
+      "step": 40845
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.180697997470734e-05,
+      "loss": 0.2648,
+      "step": 40850
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1804870963773855e-05,
+      "loss": 0.2808,
+      "step": 40855
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.180276173464019e-05,
+      "loss": 0.2897,
+      "step": 40860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1800652287333745e-05,
+      "loss": 0.2875,
+      "step": 40865
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.17985426218819e-05,
+      "loss": 0.2783,
+      "step": 40870
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1796432738312046e-05,
+      "loss": 0.2906,
+      "step": 40875
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.179432263665159e-05,
+      "loss": 0.2797,
+      "step": 40880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1792212316927917e-05,
+      "loss": 0.3023,
+      "step": 40885
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1790101779168425e-05,
+      "loss": 0.301,
+      "step": 40890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1787991023400534e-05,
+      "loss": 0.2779,
+      "step": 40895
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.178588004965165e-05,
+      "loss": 0.2833,
+      "step": 40900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1783768857949166e-05,
+      "loss": 0.3119,
+      "step": 40905
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.178165744832051e-05,
+      "loss": 0.293,
+      "step": 40910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1779545820793085e-05,
+      "loss": 0.2687,
+      "step": 40915
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.177743397539432e-05,
+      "loss": 0.2857,
+      "step": 40920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.177532191215163e-05,
+      "loss": 0.3076,
+      "step": 40925
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.177320963109244e-05,
+      "loss": 0.2961,
+      "step": 40930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.177109713224418e-05,
+      "loss": 0.2973,
+      "step": 40935
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.176898441563428e-05,
+      "loss": 0.298,
+      "step": 40940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.176687148129016e-05,
+      "loss": 0.2911,
+      "step": 40945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.176475832923927e-05,
+      "loss": 0.3035,
+      "step": 40950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1762644959509035e-05,
+      "loss": 0.2867,
+      "step": 40955
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.176053137212691e-05,
+      "loss": 0.2858,
+      "step": 40960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.175841756712032e-05,
+      "loss": 0.296,
+      "step": 40965
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.175630354451673e-05,
+      "loss": 0.2876,
+      "step": 40970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1754189304343584e-05,
+      "loss": 0.2937,
+      "step": 40975
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.175207484662833e-05,
+      "loss": 0.2828,
+      "step": 40980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1749960171398413e-05,
+      "loss": 0.262,
+      "step": 40985
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.174784527868132e-05,
+      "loss": 0.3013,
+      "step": 40990
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.174573016850448e-05,
+      "loss": 0.3128,
+      "step": 40995
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.174361484089537e-05,
+      "loss": 0.2979,
+      "step": 41000
+    },
+    {
+      "epoch": 1.44,
+      "eval_loss": 0.28040316700935364,
+      "eval_runtime": 10.542,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 41000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.174149929588146e-05,
+      "loss": 0.2828,
+      "step": 41005
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.173938353349022e-05,
+      "loss": 0.296,
+      "step": 41010
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1737267553749105e-05,
+      "loss": 0.3014,
+      "step": 41015
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.173515135668561e-05,
+      "loss": 0.2585,
+      "step": 41020
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.17330349423272e-05,
+      "loss": 0.2728,
+      "step": 41025
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.173091831070135e-05,
+      "loss": 0.2747,
+      "step": 41030
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172880146183556e-05,
+      "loss": 0.2934,
+      "step": 41035
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172668439575731e-05,
+      "loss": 0.2874,
+      "step": 41040
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.1724567112494087e-05,
+      "loss": 0.2871,
+      "step": 41045
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172244961207338e-05,
+      "loss": 0.2739,
+      "step": 41050
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.172033189452268e-05,
+      "loss": 0.2983,
+      "step": 41055
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.17182139598695e-05,
+      "loss": 0.2867,
+      "step": 41060
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.171609580814132e-05,
+      "loss": 0.2692,
+      "step": 41065
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.171397743936565e-05,
+      "loss": 0.2837,
+      "step": 41070
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.171185885357001e-05,
+      "loss": 0.2832,
+      "step": 41075
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.170974005078189e-05,
+      "loss": 0.2724,
+      "step": 41080
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.170762103102881e-05,
+      "loss": 0.2777,
+      "step": 41085
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1705501794338274e-05,
+      "loss": 0.3046,
+      "step": 41090
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.170338234073781e-05,
+      "loss": 0.2767,
+      "step": 41095
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.170126267025494e-05,
+      "loss": 0.2689,
+      "step": 41100
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.169914278291718e-05,
+      "loss": 0.2693,
+      "step": 41105
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.169702267875205e-05,
+      "loss": 0.2948,
+      "step": 41110
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.169490235778709e-05,
+      "loss": 0.295,
+      "step": 41115
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.169278182004982e-05,
+      "loss": 0.2893,
+      "step": 41120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.169066106556778e-05,
+      "loss": 0.2922,
+      "step": 41125
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1688540094368505e-05,
+      "loss": 0.3208,
+      "step": 41130
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.168641890647954e-05,
+      "loss": 0.2858,
+      "step": 41135
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1684297501928416e-05,
+      "loss": 0.2928,
+      "step": 41140
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1682175880742694e-05,
+      "loss": 0.2622,
+      "step": 41145
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.16800540429499e-05,
+      "loss": 0.2929,
+      "step": 41150
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.167793198857761e-05,
+      "loss": 0.2812,
+      "step": 41155
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.167580971765335e-05,
+      "loss": 0.2923,
+      "step": 41160
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1673687230204695e-05,
+      "loss": 0.3185,
+      "step": 41165
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.16715645262592e-05,
+      "loss": 0.2985,
+      "step": 41170
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.166944160584443e-05,
+      "loss": 0.3063,
+      "step": 41175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.166731846898795e-05,
+      "loss": 0.2869,
+      "step": 41180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.166519511571731e-05,
+      "loss": 0.3082,
+      "step": 41185
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.166307154606011e-05,
+      "loss": 0.2986,
+      "step": 41190
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.16609477600439e-05,
+      "loss": 0.2614,
+      "step": 41195
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1658823757696265e-05,
+      "loss": 0.2923,
+      "step": 41200
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.165669953904478e-05,
+      "loss": 0.2819,
+      "step": 41205
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.165457510411703e-05,
+      "loss": 0.2844,
+      "step": 41210
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1652450452940594e-05,
+      "loss": 0.2923,
+      "step": 41215
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1650325585543073e-05,
+      "loss": 0.2964,
+      "step": 41220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.164820050195204e-05,
+      "loss": 0.2783,
+      "step": 41225
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.164607520219509e-05,
+      "loss": 0.2901,
+      "step": 41230
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.164394968629983e-05,
+      "loss": 0.2434,
+      "step": 41235
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1641823954293855e-05,
+      "loss": 0.2941,
+      "step": 41240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.163969800620476e-05,
+      "loss": 0.3284,
+      "step": 41245
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1637571842060156e-05,
+      "loss": 0.3151,
+      "step": 41250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.163544546188764e-05,
+      "loss": 0.2672,
+      "step": 41255
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.163331886571483e-05,
+      "loss": 0.2914,
+      "step": 41260
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.163119205356934e-05,
+      "loss": 0.2725,
+      "step": 41265
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.162906502547877e-05,
+      "loss": 0.2869,
+      "step": 41270
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.162693778147077e-05,
+      "loss": 0.2658,
+      "step": 41275
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.162481032157293e-05,
+      "loss": 0.2572,
+      "step": 41280
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1622682645812886e-05,
+      "loss": 0.2777,
+      "step": 41285
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.162055475421825e-05,
+      "loss": 0.307,
+      "step": 41290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.161842664681668e-05,
+      "loss": 0.2797,
+      "step": 41295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1616298323635776e-05,
+      "loss": 0.2818,
+      "step": 41300
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.161416978470321e-05,
+      "loss": 0.298,
+      "step": 41305
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1612041030046576e-05,
+      "loss": 0.2989,
+      "step": 41310
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1609912059693545e-05,
+      "loss": 0.3038,
+      "step": 41315
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1607782873671756e-05,
+      "loss": 0.2784,
+      "step": 41320
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.160565347200885e-05,
+      "loss": 0.3006,
+      "step": 41325
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.160352385473248e-05,
+      "loss": 0.2945,
+      "step": 41330
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.160139402187029e-05,
+      "loss": 0.2757,
+      "step": 41335
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1599263973449944e-05,
+      "loss": 0.2844,
+      "step": 41340
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.1597133709499095e-05,
+      "loss": 0.2848,
+      "step": 41345
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.159500323004539e-05,
+      "loss": 0.2781,
+      "step": 41350
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.159287253511652e-05,
+      "loss": 0.2788,
+      "step": 41355
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.159074162474013e-05,
+      "loss": 0.2766,
+      "step": 41360
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1588610498943895e-05,
+      "loss": 0.3108,
+      "step": 41365
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1586479157755483e-05,
+      "loss": 0.3063,
+      "step": 41370
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.158434760120257e-05,
+      "loss": 0.2931,
+      "step": 41375
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.158221582931283e-05,
+      "loss": 0.2764,
+      "step": 41380
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.158008384211395e-05,
+      "loss": 0.2896,
+      "step": 41385
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.15779516396336e-05,
+      "loss": 0.3045,
+      "step": 41390
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.157581922189949e-05,
+      "loss": 0.3006,
+      "step": 41395
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.157368658893928e-05,
+      "loss": 0.2764,
+      "step": 41400
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1571553740780665e-05,
+      "loss": 0.287,
+      "step": 41405
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.156942067745136e-05,
+      "loss": 0.3057,
+      "step": 41410
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1567287398979036e-05,
+      "loss": 0.2763,
+      "step": 41415
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.156515390539142e-05,
+      "loss": 0.2769,
+      "step": 41420
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.156302019671618e-05,
+      "loss": 0.2767,
+      "step": 41425
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1560886272981045e-05,
+      "loss": 0.2798,
+      "step": 41430
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.155875213421372e-05,
+      "loss": 0.318,
+      "step": 41435
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1556617780441895e-05,
+      "loss": 0.3104,
+      "step": 41440
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.155448321169332e-05,
+      "loss": 0.2767,
+      "step": 41445
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.155234842799568e-05,
+      "loss": 0.3217,
+      "step": 41450
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1550213429376705e-05,
+      "loss": 0.2916,
+      "step": 41455
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.154807821586412e-05,
+      "loss": 0.2939,
+      "step": 41460
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1545942787485636e-05,
+      "loss": 0.3086,
+      "step": 41465
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1543807144268996e-05,
+      "loss": 0.266,
+      "step": 41470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.154167128624191e-05,
+      "loss": 0.3067,
+      "step": 41475
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.153953521343214e-05,
+      "loss": 0.2628,
+      "step": 41480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.15373989258674e-05,
+      "loss": 0.2992,
+      "step": 41485
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.153526242357543e-05,
+      "loss": 0.2798,
+      "step": 41490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1533125706583974e-05,
+      "loss": 0.3018,
+      "step": 41495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.153098877492079e-05,
+      "loss": 0.2966,
+      "step": 41500
+    },
+    {
+      "epoch": 1.46,
+      "eval_loss": 0.28131207823753357,
+      "eval_runtime": 10.5489,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 41500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.15288516286136e-05,
+      "loss": 0.2731,
+      "step": 41505
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.152671426769016e-05,
+      "loss": 0.2762,
+      "step": 41510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1524576692178235e-05,
+      "loss": 0.2886,
+      "step": 41515
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1522438902105564e-05,
+      "loss": 0.299,
+      "step": 41520
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.152030089749992e-05,
+      "loss": 0.282,
+      "step": 41525
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1518162678389054e-05,
+      "loss": 0.3056,
+      "step": 41530
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1516024244800734e-05,
+      "loss": 0.2771,
+      "step": 41535
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.151388559676272e-05,
+      "loss": 0.2835,
+      "step": 41540
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1511746734302786e-05,
+      "loss": 0.2921,
+      "step": 41545
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1509607657448704e-05,
+      "loss": 0.2958,
+      "step": 41550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.150746836622824e-05,
+      "loss": 0.2752,
+      "step": 41555
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.150532886066918e-05,
+      "loss": 0.3075,
+      "step": 41560
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1503189140799315e-05,
+      "loss": 0.2905,
+      "step": 41565
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.15010492066464e-05,
+      "loss": 0.298,
+      "step": 41570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1498909058238246e-05,
+      "loss": 0.2838,
+      "step": 41575
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.149676869560262e-05,
+      "loss": 0.2928,
+      "step": 41580
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.149462811876733e-05,
+      "loss": 0.2737,
+      "step": 41585
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1492487327760165e-05,
+      "loss": 0.2936,
+      "step": 41590
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.149034632260892e-05,
+      "loss": 0.2766,
+      "step": 41595
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1488205103341395e-05,
+      "loss": 0.279,
+      "step": 41600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.14860636699854e-05,
+      "loss": 0.2808,
+      "step": 41605
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.148392202256872e-05,
+      "loss": 0.2937,
+      "step": 41610
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.148178016111919e-05,
+      "loss": 0.2873,
+      "step": 41615
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.147963808566459e-05,
+      "loss": 0.2778,
+      "step": 41620
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.147749579623276e-05,
+      "loss": 0.2742,
+      "step": 41625
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.14753532928515e-05,
+      "loss": 0.296,
+      "step": 41630
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.1473210575548644e-05,
+      "loss": 0.2929,
+      "step": 41635
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1471067644352e-05,
+      "loss": 0.2728,
+      "step": 41640
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1468924499289396e-05,
+      "loss": 0.2815,
+      "step": 41645
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1466781140388666e-05,
+      "loss": 0.3016,
+      "step": 41650
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1464637567677636e-05,
+      "loss": 0.2837,
+      "step": 41655
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.146249378118413e-05,
+      "loss": 0.28,
+      "step": 41660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1460349780936e-05,
+      "loss": 0.2909,
+      "step": 41665
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.145820556696107e-05,
+      "loss": 0.313,
+      "step": 41670
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1456061139287186e-05,
+      "loss": 0.2919,
+      "step": 41675
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.14539164979422e-05,
+      "loss": 0.2965,
+      "step": 41680
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.145177164295395e-05,
+      "loss": 0.2947,
+      "step": 41685
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144962657435028e-05,
+      "loss": 0.2752,
+      "step": 41690
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144748129215905e-05,
+      "loss": 0.2919,
+      "step": 41695
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144533579640812e-05,
+      "loss": 0.2917,
+      "step": 41700
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1443190087125344e-05,
+      "loss": 0.2771,
+      "step": 41705
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.144104416433858e-05,
+      "loss": 0.3088,
+      "step": 41710
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.143889802807569e-05,
+      "loss": 0.3076,
+      "step": 41715
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.143675167836455e-05,
+      "loss": 0.2778,
+      "step": 41720
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.143460511523302e-05,
+      "loss": 0.276,
+      "step": 41725
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1432458338708965e-05,
+      "loss": 0.2951,
+      "step": 41730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.143031134882027e-05,
+      "loss": 0.2916,
+      "step": 41735
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.142816414559481e-05,
+      "loss": 0.2642,
+      "step": 41740
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.142601672906047e-05,
+      "loss": 0.2762,
+      "step": 41745
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.142386909924512e-05,
+      "loss": 0.3077,
+      "step": 41750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.142172125617666e-05,
+      "loss": 0.2859,
+      "step": 41755
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.141957319988297e-05,
+      "loss": 0.2824,
+      "step": 41760
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1417424930391946e-05,
+      "loss": 0.2843,
+      "step": 41765
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1415276447731476e-05,
+      "loss": 0.3122,
+      "step": 41770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.141312775192946e-05,
+      "loss": 0.2629,
+      "step": 41775
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.141097884301379e-05,
+      "loss": 0.2667,
+      "step": 41780
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.140882972101239e-05,
+      "loss": 0.27,
+      "step": 41785
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1406680385953134e-05,
+      "loss": 0.304,
+      "step": 41790
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.140453083786396e-05,
+      "loss": 0.3035,
+      "step": 41795
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.140238107677276e-05,
+      "loss": 0.2816,
+      "step": 41800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.140023110270744e-05,
+      "loss": 0.2717,
+      "step": 41805
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1398080915695936e-05,
+      "loss": 0.2917,
+      "step": 41810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.139593051576616e-05,
+      "loss": 0.3025,
+      "step": 41815
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1393779902946025e-05,
+      "loss": 0.2898,
+      "step": 41820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.139162907726347e-05,
+      "loss": 0.2947,
+      "step": 41825
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.138947803874641e-05,
+      "loss": 0.3029,
+      "step": 41830
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1387326787422774e-05,
+      "loss": 0.3071,
+      "step": 41835
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.138517532332051e-05,
+      "loss": 0.3212,
+      "step": 41840
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1383023646467536e-05,
+      "loss": 0.2873,
+      "step": 41845
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.138087175689179e-05,
+      "loss": 0.2885,
+      "step": 41850
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.137871965462124e-05,
+      "loss": 0.2966,
+      "step": 41855
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.137656733968379e-05,
+      "loss": 0.2855,
+      "step": 41860
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.137441481210741e-05,
+      "loss": 0.2517,
+      "step": 41865
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.137226207192005e-05,
+      "loss": 0.2752,
+      "step": 41870
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.137010911914965e-05,
+      "loss": 0.2747,
+      "step": 41875
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1367955953824176e-05,
+      "loss": 0.2791,
+      "step": 41880
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.136580257597158e-05,
+      "loss": 0.2903,
+      "step": 41885
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.136364898561982e-05,
+      "loss": 0.2824,
+      "step": 41890
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1361495182796866e-05,
+      "loss": 0.2816,
+      "step": 41895
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1359341167530684e-05,
+      "loss": 0.2877,
+      "step": 41900
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.1357186939849225e-05,
+      "loss": 0.2658,
+      "step": 41905
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.135503249978049e-05,
+      "loss": 0.2948,
+      "step": 41910
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.135287784735242e-05,
+      "loss": 0.3228,
+      "step": 41915
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.135072298259301e-05,
+      "loss": 0.2755,
+      "step": 41920
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.134856790553025e-05,
+      "loss": 0.2892,
+      "step": 41925
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.13464126161921e-05,
+      "loss": 0.2966,
+      "step": 41930
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.134425711460656e-05,
+      "loss": 0.2837,
+      "step": 41935
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.134210140080161e-05,
+      "loss": 0.2627,
+      "step": 41940
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1339945474805244e-05,
+      "loss": 0.3242,
+      "step": 41945
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.133778933664546e-05,
+      "loss": 0.2873,
+      "step": 41950
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1335632986350235e-05,
+      "loss": 0.29,
+      "step": 41955
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1333476423947594e-05,
+      "loss": 0.2837,
+      "step": 41960
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1331319649465515e-05,
+      "loss": 0.282,
+      "step": 41965
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.132916266293202e-05,
+      "loss": 0.3029,
+      "step": 41970
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1327005464375106e-05,
+      "loss": 0.283,
+      "step": 41975
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.13248480538228e-05,
+      "loss": 0.3077,
+      "step": 41980
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.132269043130308e-05,
+      "loss": 0.286,
+      "step": 41985
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1320532596843985e-05,
+      "loss": 0.2866,
+      "step": 41990
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1318374550473536e-05,
+      "loss": 0.3032,
+      "step": 41995
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1316216292219746e-05,
+      "loss": 0.3057,
+      "step": 42000
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 0.2798173427581787,
+      "eval_runtime": 10.5336,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 42000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.131405782211064e-05,
+      "loss": 0.2677,
+      "step": 42005
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1311899140174246e-05,
+      "loss": 0.287,
+      "step": 42010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1309740246438596e-05,
+      "loss": 0.2707,
+      "step": 42015
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1307581140931705e-05,
+      "loss": 0.2812,
+      "step": 42020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1305421823681635e-05,
+      "loss": 0.3009,
+      "step": 42025
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.13032622947164e-05,
+      "loss": 0.2792,
+      "step": 42030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1301102554064046e-05,
+      "loss": 0.2933,
+      "step": 42035
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.129894260175262e-05,
+      "loss": 0.2755,
+      "step": 42040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1296782437810166e-05,
+      "loss": 0.2816,
+      "step": 42045
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.129462206226473e-05,
+      "loss": 0.2795,
+      "step": 42050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.129246147514437e-05,
+      "loss": 0.2853,
+      "step": 42055
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.129030067647713e-05,
+      "loss": 0.2732,
+      "step": 42060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.128813966629108e-05,
+      "loss": 0.2671,
+      "step": 42065
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1285978444614266e-05,
+      "loss": 0.3028,
+      "step": 42070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1283817011474755e-05,
+      "loss": 0.2778,
+      "step": 42075
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.128165536690061e-05,
+      "loss": 0.2685,
+      "step": 42080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.12794935109199e-05,
+      "loss": 0.289,
+      "step": 42085
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.12773314435607e-05,
+      "loss": 0.2809,
+      "step": 42090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1275169164851066e-05,
+      "loss": 0.3002,
+      "step": 42095
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.127300667481909e-05,
+      "loss": 0.296,
+      "step": 42100
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1270843973492854e-05,
+      "loss": 0.2916,
+      "step": 42105
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.126868106090043e-05,
+      "loss": 0.2953,
+      "step": 42110
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.12665179370699e-05,
+      "loss": 0.2817,
+      "step": 42115
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1264354602029364e-05,
+      "loss": 0.2931,
+      "step": 42120
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.126219105580689e-05,
+      "loss": 0.2771,
+      "step": 42125
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1260027298430594e-05,
+      "loss": 0.3055,
+      "step": 42130
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.125786332992855e-05,
+      "loss": 0.2824,
+      "step": 42135
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.125613200313527e-05,
+      "loss": 0.2557,
+      "step": 42140
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1253967654677706e-05,
+      "loss": 0.3008,
+      "step": 42145
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.125180309517308e-05,
+      "loss": 0.2772,
+      "step": 42150
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.124963832464951e-05,
+      "loss": 0.2838,
+      "step": 42155
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.124747334313509e-05,
+      "loss": 0.2992,
+      "step": 42160
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.124530815065794e-05,
+      "loss": 0.3114,
+      "step": 42165
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1243142747246164e-05,
+      "loss": 0.3013,
+      "step": 42170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.1240977132927886e-05,
+      "loss": 0.3294,
+      "step": 42175
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123881130773123e-05,
+      "loss": 0.2592,
+      "step": 42180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123664527168431e-05,
+      "loss": 0.302,
+      "step": 42185
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123447902481524e-05,
+      "loss": 0.2871,
+      "step": 42190
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123231256715217e-05,
+      "loss": 0.276,
+      "step": 42195
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.123014589872321e-05,
+      "loss": 0.2924,
+      "step": 42200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.122797901955651e-05,
+      "loss": 0.2593,
+      "step": 42205
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.122581192968019e-05,
+      "loss": 0.276,
+      "step": 42210
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1223644629122395e-05,
+      "loss": 0.2861,
+      "step": 42215
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.122147711791128e-05,
+      "loss": 0.2924,
+      "step": 42220
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1219309396074956e-05,
+      "loss": 0.309,
+      "step": 42225
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.12171414636416e-05,
+      "loss": 0.2984,
+      "step": 42230
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.121497332063935e-05,
+      "loss": 0.2866,
+      "step": 42235
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1212804967096355e-05,
+      "loss": 0.284,
+      "step": 42240
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.121063640304077e-05,
+      "loss": 0.3037,
+      "step": 42245
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.120846762850076e-05,
+      "loss": 0.3191,
+      "step": 42250
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.120629864350447e-05,
+      "loss": 0.2827,
+      "step": 42255
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.120412944808008e-05,
+      "loss": 0.3025,
+      "step": 42260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.120196004225574e-05,
+      "loss": 0.276,
+      "step": 42265
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.119979042605964e-05,
+      "loss": 0.277,
+      "step": 42270
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1197620599519927e-05,
+      "loss": 0.306,
+      "step": 42275
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.119545056266478e-05,
+      "loss": 0.2725,
+      "step": 42280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.11932803155224e-05,
+      "loss": 0.279,
+      "step": 42285
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.119110985812093e-05,
+      "loss": 0.2873,
+      "step": 42290
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.118893919048858e-05,
+      "loss": 0.281,
+      "step": 42295
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.118676831265351e-05,
+      "loss": 0.2841,
+      "step": 42300
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1184597224643936e-05,
+      "loss": 0.2785,
+      "step": 42305
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.118242592648802e-05,
+      "loss": 0.2958,
+      "step": 42310
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1180254418213976e-05,
+      "loss": 0.2794,
+      "step": 42315
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1178082699849984e-05,
+      "loss": 0.2772,
+      "step": 42320
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1175910771424266e-05,
+      "loss": 0.2904,
+      "step": 42325
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1173738632964986e-05,
+      "loss": 0.2881,
+      "step": 42330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.117156628450039e-05,
+      "loss": 0.2859,
+      "step": 42335
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1169393726058645e-05,
+      "loss": 0.2855,
+      "step": 42340
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.116722095766798e-05,
+      "loss": 0.3053,
+      "step": 42345
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.116504797935662e-05,
+      "loss": 0.2883,
+      "step": 42350
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1162874791152755e-05,
+      "loss": 0.2792,
+      "step": 42355
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.116070139308461e-05,
+      "loss": 0.292,
+      "step": 42360
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.115852778518041e-05,
+      "loss": 0.2843,
+      "step": 42365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1156353967468384e-05,
+      "loss": 0.2932,
+      "step": 42370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.115417993997674e-05,
+      "loss": 0.3182,
+      "step": 42375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.115200570273372e-05,
+      "loss": 0.2978,
+      "step": 42380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.114983125576754e-05,
+      "loss": 0.3059,
+      "step": 42385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.114765659910646e-05,
+      "loss": 0.2767,
+      "step": 42390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1145481732778695e-05,
+      "loss": 0.3192,
+      "step": 42395
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.114330665681249e-05,
+      "loss": 0.2982,
+      "step": 42400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.114113137123608e-05,
+      "loss": 0.3278,
+      "step": 42405
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1138955876077725e-05,
+      "loss": 0.2876,
+      "step": 42410
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1136780171365656e-05,
+      "loss": 0.2876,
+      "step": 42415
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.113460425712814e-05,
+      "loss": 0.2808,
+      "step": 42420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1132428133393414e-05,
+      "loss": 0.3196,
+      "step": 42425
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.113025180018974e-05,
+      "loss": 0.2973,
+      "step": 42430
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1128075257545384e-05,
+      "loss": 0.2506,
+      "step": 42435
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1125898505488595e-05,
+      "loss": 0.3138,
+      "step": 42440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.112372154404765e-05,
+      "loss": 0.2755,
+      "step": 42445
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.112154437325079e-05,
+      "loss": 0.2799,
+      "step": 42450
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.111936699312631e-05,
+      "loss": 0.3056,
+      "step": 42455
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.111718940370247e-05,
+      "loss": 0.2922,
+      "step": 42460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1115011605007545e-05,
+      "loss": 0.2564,
+      "step": 42465
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.111283359706982e-05,
+      "loss": 0.2812,
+      "step": 42470
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.111065537991757e-05,
+      "loss": 0.3041,
+      "step": 42475
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.1108476953579076e-05,
+      "loss": 0.2833,
+      "step": 42480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.110629831808263e-05,
+      "loss": 0.2885,
+      "step": 42485
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.11041194734565e-05,
+      "loss": 0.2598,
+      "step": 42490
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1101940419729e-05,
+      "loss": 0.2646,
+      "step": 42495
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.109976115692842e-05,
+      "loss": 0.2901,
+      "step": 42500
+    },
+    {
+      "epoch": 1.5,
+      "eval_loss": 0.28041115403175354,
+      "eval_runtime": 10.5442,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 42500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.109758168508304e-05,
+      "loss": 0.285,
+      "step": 42505
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1095402004221184e-05,
+      "loss": 0.2893,
+      "step": 42510
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1093222114371134e-05,
+      "loss": 0.2786,
+      "step": 42515
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.10910420155612e-05,
+      "loss": 0.2808,
+      "step": 42520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.108886170781969e-05,
+      "loss": 0.2984,
+      "step": 42525
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.108668119117491e-05,
+      "loss": 0.2864,
+      "step": 42530
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1084500465655186e-05,
+      "loss": 0.2877,
+      "step": 42535
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1082319531288816e-05,
+      "loss": 0.2936,
+      "step": 42540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.108013838810413e-05,
+      "loss": 0.2693,
+      "step": 42545
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1077957036129436e-05,
+      "loss": 0.2713,
+      "step": 42550
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1075775475393077e-05,
+      "loss": 0.2775,
+      "step": 42555
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.107359370592335e-05,
+      "loss": 0.2905,
+      "step": 42560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.107141172774862e-05,
+      "loss": 0.2979,
+      "step": 42565
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.106922954089719e-05,
+      "loss": 0.2906,
+      "step": 42570
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.106704714539741e-05,
+      "loss": 0.2724,
+      "step": 42575
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.106486454127761e-05,
+      "loss": 0.2972,
+      "step": 42580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.106268172856613e-05,
+      "loss": 0.2951,
+      "step": 42585
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1060498707291314e-05,
+      "loss": 0.3015,
+      "step": 42590
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1058315477481506e-05,
+      "loss": 0.2937,
+      "step": 42595
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.105613203916505e-05,
+      "loss": 0.2863,
+      "step": 42600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1053948392370315e-05,
+      "loss": 0.2809,
+      "step": 42605
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.105176453712563e-05,
+      "loss": 0.2942,
+      "step": 42610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.104958047345936e-05,
+      "loss": 0.2738,
+      "step": 42615
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.104739620139987e-05,
+      "loss": 0.2744,
+      "step": 42620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1045211720975506e-05,
+      "loss": 0.2867,
+      "step": 42625
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1043027032214656e-05,
+      "loss": 0.2808,
+      "step": 42630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.104084213514566e-05,
+      "loss": 0.2839,
+      "step": 42635
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1038657029796904e-05,
+      "loss": 0.2892,
+      "step": 42640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1036471716196755e-05,
+      "loss": 0.293,
+      "step": 42645
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.103428619437359e-05,
+      "loss": 0.301,
+      "step": 42650
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.103210046435579e-05,
+      "loss": 0.2923,
+      "step": 42655
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1029914526171734e-05,
+      "loss": 0.2789,
+      "step": 42660
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1027728379849794e-05,
+      "loss": 0.3124,
+      "step": 42665
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1025542025418365e-05,
+      "loss": 0.2955,
+      "step": 42670
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.102335546290583e-05,
+      "loss": 0.266,
+      "step": 42675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.102116869234059e-05,
+      "loss": 0.291,
+      "step": 42680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1018981713751035e-05,
+      "loss": 0.2801,
+      "step": 42685
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.101679452716556e-05,
+      "loss": 0.2897,
+      "step": 42690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.101460713261256e-05,
+      "loss": 0.2788,
+      "step": 42695
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.101241953012044e-05,
+      "loss": 0.2689,
+      "step": 42700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.101023171971761e-05,
+      "loss": 0.2894,
+      "step": 42705
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1008043701432465e-05,
+      "loss": 0.2889,
+      "step": 42710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1005855475293426e-05,
+      "loss": 0.2879,
+      "step": 42715
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.10036670413289e-05,
+      "loss": 0.2864,
+      "step": 42720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.1001478399567305e-05,
+      "loss": 0.2618,
+      "step": 42725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099928955003706e-05,
+      "loss": 0.2802,
+      "step": 42730
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099710049276658e-05,
+      "loss": 0.2785,
+      "step": 42735
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0994911227784296e-05,
+      "loss": 0.2576,
+      "step": 42740
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099272175511863e-05,
+      "loss": 0.3038,
+      "step": 42745
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.099053207479802e-05,
+      "loss": 0.312,
+      "step": 42750
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.098834218685087e-05,
+      "loss": 0.3024,
+      "step": 42755
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0986152091305645e-05,
+      "loss": 0.2952,
+      "step": 42760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0983961788190774e-05,
+      "loss": 0.3006,
+      "step": 42765
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0981771277534684e-05,
+      "loss": 0.322,
+      "step": 42770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.0979580559365835e-05,
+      "loss": 0.2838,
+      "step": 42775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.097738963371266e-05,
+      "loss": 0.2807,
+      "step": 42780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.097519850060361e-05,
+      "loss": 0.2987,
+      "step": 42785
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.097300716006713e-05,
+      "loss": 0.2989,
+      "step": 42790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.097081561213169e-05,
+      "loss": 0.2826,
+      "step": 42795
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.096862385682572e-05,
+      "loss": 0.277,
+      "step": 42800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.09664318941777e-05,
+      "loss": 0.2774,
+      "step": 42805
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0964239724216094e-05,
+      "loss": 0.2809,
+      "step": 42810
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.096204734696935e-05,
+      "loss": 0.2972,
+      "step": 42815
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.095985476246593e-05,
+      "loss": 0.2467,
+      "step": 42820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0957661970734326e-05,
+      "loss": 0.3159,
+      "step": 42825
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.095546897180299e-05,
+      "loss": 0.2756,
+      "step": 42830
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.095327576570041e-05,
+      "loss": 0.2908,
+      "step": 42835
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.095108235245505e-05,
+      "loss": 0.3008,
+      "step": 42840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0948888732095405e-05,
+      "loss": 0.2964,
+      "step": 42845
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0946694904649954e-05,
+      "loss": 0.3011,
+      "step": 42850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0944500870147176e-05,
+      "loss": 0.2992,
+      "step": 42855
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.094230662861555e-05,
+      "loss": 0.2891,
+      "step": 42860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.094011218008359e-05,
+      "loss": 0.2977,
+      "step": 42865
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.093791752457978e-05,
+      "loss": 0.2708,
+      "step": 42870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.093572266213261e-05,
+      "loss": 0.2906,
+      "step": 42875
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.093352759277058e-05,
+      "loss": 0.2713,
+      "step": 42880
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0931332316522206e-05,
+      "loss": 0.2765,
+      "step": 42885
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.092913683341597e-05,
+      "loss": 0.3059,
+      "step": 42890
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0926941143480397e-05,
+      "loss": 0.2758,
+      "step": 42895
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.092474524674399e-05,
+      "loss": 0.267,
+      "step": 42900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.092254914323526e-05,
+      "loss": 0.2707,
+      "step": 42905
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0920352832982714e-05,
+      "loss": 0.3051,
+      "step": 42910
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0918156316014884e-05,
+      "loss": 0.3182,
+      "step": 42915
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.091595959236029e-05,
+      "loss": 0.3193,
+      "step": 42920
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.091376266204744e-05,
+      "loss": 0.2697,
+      "step": 42925
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0911565525104876e-05,
+      "loss": 0.2995,
+      "step": 42930
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.090936818156112e-05,
+      "loss": 0.2751,
+      "step": 42935
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0907170631444703e-05,
+      "loss": 0.3034,
+      "step": 42940
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.090497287478415e-05,
+      "loss": 0.2835,
+      "step": 42945
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.090277491160802e-05,
+      "loss": 0.2839,
+      "step": 42950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.090057674194483e-05,
+      "loss": 0.3095,
+      "step": 42955
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.089837836582313e-05,
+      "loss": 0.2662,
+      "step": 42960
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.089617978327146e-05,
+      "loss": 0.2959,
+      "step": 42965
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.089398099431838e-05,
+      "loss": 0.3037,
+      "step": 42970
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.089178199899243e-05,
+      "loss": 0.3023,
+      "step": 42975
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0889582797322155e-05,
+      "loss": 0.2804,
+      "step": 42980
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.088738338933613e-05,
+      "loss": 0.2966,
+      "step": 42985
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0885183775062895e-05,
+      "loss": 0.269,
+      "step": 42990
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.088298395453102e-05,
+      "loss": 0.274,
+      "step": 42995
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.088078392776906e-05,
+      "loss": 0.2803,
+      "step": 43000
+    },
+    {
+      "epoch": 1.51,
+      "eval_loss": 0.2803913950920105,
+      "eval_runtime": 10.5528,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 43000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.08785836948056e-05,
+      "loss": 0.2946,
+      "step": 43005
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0876383255669184e-05,
+      "loss": 0.2873,
+      "step": 43010
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.08741826103884e-05,
+      "loss": 0.2766,
+      "step": 43015
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.087198175899181e-05,
+      "loss": 0.295,
+      "step": 43020
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0869780701508e-05,
+      "loss": 0.3029,
+      "step": 43025
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.086757943796555e-05,
+      "loss": 0.2777,
+      "step": 43030
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.086537796839304e-05,
+      "loss": 0.3146,
+      "step": 43035
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.086317629281904e-05,
+      "loss": 0.2991,
+      "step": 43040
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.086097441127216e-05,
+      "loss": 0.2706,
+      "step": 43045
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.085877232378098e-05,
+      "loss": 0.2813,
+      "step": 43050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.085657003037409e-05,
+      "loss": 0.3014,
+      "step": 43055
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.0854367531080086e-05,
+      "loss": 0.2965,
+      "step": 43060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.085216482592757e-05,
+      "loss": 0.2942,
+      "step": 43065
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0849961914945146e-05,
+      "loss": 0.291,
+      "step": 43070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.084775879816141e-05,
+      "loss": 0.2719,
+      "step": 43075
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0845555475604966e-05,
+      "loss": 0.2727,
+      "step": 43080
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0843351947304425e-05,
+      "loss": 0.2802,
+      "step": 43085
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.084114821328841e-05,
+      "loss": 0.2944,
+      "step": 43090
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.083894427358551e-05,
+      "loss": 0.2627,
+      "step": 43095
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0836740128224374e-05,
+      "loss": 0.3068,
+      "step": 43100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.083453577723359e-05,
+      "loss": 0.2858,
+      "step": 43105
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.083233122064179e-05,
+      "loss": 0.304,
+      "step": 43110
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.083012645847762e-05,
+      "loss": 0.3016,
+      "step": 43115
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0827921490769683e-05,
+      "loss": 0.2655,
+      "step": 43120
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0825716317546616e-05,
+      "loss": 0.3103,
+      "step": 43125
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0823510938837055e-05,
+      "loss": 0.2703,
+      "step": 43130
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0821305354669624e-05,
+      "loss": 0.2977,
+      "step": 43135
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081909956507298e-05,
+      "loss": 0.2772,
+      "step": 43140
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081689357007575e-05,
+      "loss": 0.2902,
+      "step": 43145
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081468736970658e-05,
+      "loss": 0.2871,
+      "step": 43150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081248096399412e-05,
+      "loss": 0.2702,
+      "step": 43155
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.081027435296701e-05,
+      "loss": 0.2834,
+      "step": 43160
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.08080675366539e-05,
+      "loss": 0.2735,
+      "step": 43165
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.080586051508347e-05,
+      "loss": 0.2994,
+      "step": 43170
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.080365328828435e-05,
+      "loss": 0.2786,
+      "step": 43175
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0801445856285205e-05,
+      "loss": 0.2774,
+      "step": 43180
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.07992382191147e-05,
+      "loss": 0.2731,
+      "step": 43185
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.079703037680149e-05,
+      "loss": 0.2878,
+      "step": 43190
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.079482232937426e-05,
+      "loss": 0.2821,
+      "step": 43195
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.079261407686167e-05,
+      "loss": 0.2589,
+      "step": 43200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.079040561929239e-05,
+      "loss": 0.2859,
+      "step": 43205
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.078819695669511e-05,
+      "loss": 0.2743,
+      "step": 43210
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0785988089098494e-05,
+      "loss": 0.2989,
+      "step": 43215
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.078377901653122e-05,
+      "loss": 0.288,
+      "step": 43220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.078156973902198e-05,
+      "loss": 0.2737,
+      "step": 43225
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.077936025659945e-05,
+      "loss": 0.3033,
+      "step": 43230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0777150569292345e-05,
+      "loss": 0.2956,
+      "step": 43235
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0774940677129316e-05,
+      "loss": 0.2846,
+      "step": 43240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.07727305801391e-05,
+      "loss": 0.3056,
+      "step": 43245
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0770520278350354e-05,
+      "loss": 0.2888,
+      "step": 43250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.07683097717918e-05,
+      "loss": 0.2872,
+      "step": 43255
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.076609906049214e-05,
+      "loss": 0.2991,
+      "step": 43260
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0763888144480075e-05,
+      "loss": 0.292,
+      "step": 43265
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.076167702378431e-05,
+      "loss": 0.2684,
+      "step": 43270
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0759465698433545e-05,
+      "loss": 0.2921,
+      "step": 43275
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0757254168456514e-05,
+      "loss": 0.2779,
+      "step": 43280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.075504243388192e-05,
+      "loss": 0.2974,
+      "step": 43285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.075283049473848e-05,
+      "loss": 0.2597,
+      "step": 43290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0750618351054924e-05,
+      "loss": 0.268,
+      "step": 43295
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0748406002859963e-05,
+      "loss": 0.2818,
+      "step": 43300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.074619345018233e-05,
+      "loss": 0.2867,
+      "step": 43305
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0743980693050753e-05,
+      "loss": 0.2596,
+      "step": 43310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.074176773149396e-05,
+      "loss": 0.3037,
+      "step": 43315
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0739554565540697e-05,
+      "loss": 0.2669,
+      "step": 43320
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.073734119521968e-05,
+      "loss": 0.2882,
+      "step": 43325
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0735127620559664e-05,
+      "loss": 0.3036,
+      "step": 43330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.0732913841589385e-05,
+      "loss": 0.2811,
+      "step": 43335
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.073069985833758e-05,
+      "loss": 0.2856,
+      "step": 43340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.072848567083302e-05,
+      "loss": 0.3084,
+      "step": 43345
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0726271279104435e-05,
+      "loss": 0.2959,
+      "step": 43350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.072405668318057e-05,
+      "loss": 0.2694,
+      "step": 43355
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.07218418830902e-05,
+      "loss": 0.3004,
+      "step": 43360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0719626878862074e-05,
+      "loss": 0.246,
+      "step": 43365
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.071741167052495e-05,
+      "loss": 0.2825,
+      "step": 43370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.071519625810759e-05,
+      "loss": 0.2835,
+      "step": 43375
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0712980641638774e-05,
+      "loss": 0.2787,
+      "step": 43380
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.071076482114725e-05,
+      "loss": 0.27,
+      "step": 43385
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0708548796661805e-05,
+      "loss": 0.3059,
+      "step": 43390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.07063325682112e-05,
+      "loss": 0.2701,
+      "step": 43395
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0704116135824226e-05,
+      "loss": 0.3022,
+      "step": 43400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.070189949952965e-05,
+      "loss": 0.2763,
+      "step": 43405
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0699682659356254e-05,
+      "loss": 0.3014,
+      "step": 43410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.069746561533282e-05,
+      "loss": 0.3238,
+      "step": 43415
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.069524836748815e-05,
+      "loss": 0.2848,
+      "step": 43420
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.069303091585102e-05,
+      "loss": 0.2805,
+      "step": 43425
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.069081326045023e-05,
+      "loss": 0.2921,
+      "step": 43430
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.068859540131457e-05,
+      "loss": 0.277,
+      "step": 43435
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.068637733847283e-05,
+      "loss": 0.2988,
+      "step": 43440
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.068415907195382e-05,
+      "loss": 0.295,
+      "step": 43445
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0681940601786345e-05,
+      "loss": 0.2782,
+      "step": 43450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06797219279992e-05,
+      "loss": 0.2732,
+      "step": 43455
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06775030506212e-05,
+      "loss": 0.3081,
+      "step": 43460
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.067528396968116e-05,
+      "loss": 0.3017,
+      "step": 43465
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.067306468520788e-05,
+      "loss": 0.2942,
+      "step": 43470
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.06708451972302e-05,
+      "loss": 0.3089,
+      "step": 43475
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0668625505776906e-05,
+      "loss": 0.2851,
+      "step": 43480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.066640561087684e-05,
+      "loss": 0.2862,
+      "step": 43485
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.066418551255883e-05,
+      "loss": 0.2883,
+      "step": 43490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.066196521085169e-05,
+      "loss": 0.2865,
+      "step": 43495
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0659744705784254e-05,
+      "loss": 0.2946,
+      "step": 43500
+    },
+    {
+      "epoch": 1.53,
+      "eval_loss": 0.27933764457702637,
+      "eval_runtime": 10.6369,
+      "eval_samples_per_second": 9.401,
+      "eval_steps_per_second": 9.401,
+      "step": 43500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065752399738535e-05,
+      "loss": 0.2491,
+      "step": 43505
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065530308568381e-05,
+      "loss": 0.2951,
+      "step": 43510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0653081970708486e-05,
+      "loss": 0.2764,
+      "step": 43515
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.065086065248821e-05,
+      "loss": 0.2887,
+      "step": 43520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0648639131051815e-05,
+      "loss": 0.272,
+      "step": 43525
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.064641740642815e-05,
+      "loss": 0.2902,
+      "step": 43530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.064419547864607e-05,
+      "loss": 0.2942,
+      "step": 43535
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.064197334773443e-05,
+      "loss": 0.2664,
+      "step": 43540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063975101372207e-05,
+      "loss": 0.2929,
+      "step": 43545
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063752847663785e-05,
+      "loss": 0.2745,
+      "step": 43550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063530573651063e-05,
+      "loss": 0.3009,
+      "step": 43555
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0633082793369256e-05,
+      "loss": 0.3144,
+      "step": 43560
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.063085964724263e-05,
+      "loss": 0.2595,
+      "step": 43565
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.062863629815957e-05,
+      "loss": 0.2784,
+      "step": 43570
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.062641274614897e-05,
+      "loss": 0.2907,
+      "step": 43575
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.062418899123971e-05,
+      "loss": 0.2899,
+      "step": 43580
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.062196503346064e-05,
+      "loss": 0.2848,
+      "step": 43585
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.061974087284066e-05,
+      "loss": 0.2721,
+      "step": 43590
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0617516509408636e-05,
+      "loss": 0.2915,
+      "step": 43595
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0615291943193446e-05,
+      "loss": 0.2946,
+      "step": 43600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0613067174223987e-05,
+      "loss": 0.2876,
+      "step": 43605
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0610842202529144e-05,
+      "loss": 0.2818,
+      "step": 43610
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.0608617028137793e-05,
+      "loss": 0.2997,
+      "step": 43615
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.060639165107885e-05,
+      "loss": 0.2639,
+      "step": 43620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.060416607138118e-05,
+      "loss": 0.289,
+      "step": 43625
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.06019402890737e-05,
+      "loss": 0.3012,
+      "step": 43630
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0599714304185315e-05,
+      "loss": 0.3031,
+      "step": 43635
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.059748811674492e-05,
+      "loss": 0.2745,
+      "step": 43640
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.059526172678142e-05,
+      "loss": 0.2801,
+      "step": 43645
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.059303513432372e-05,
+      "loss": 0.3029,
+      "step": 43650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0590808339400735e-05,
+      "loss": 0.2762,
+      "step": 43655
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0588581342041385e-05,
+      "loss": 0.2592,
+      "step": 43660
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0586354142274575e-05,
+      "loss": 0.3027,
+      "step": 43665
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0584126740129225e-05,
+      "loss": 0.2809,
+      "step": 43670
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0581899135634263e-05,
+      "loss": 0.2901,
+      "step": 43675
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.057967132881861e-05,
+      "loss": 0.2773,
+      "step": 43680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0577443319711197e-05,
+      "loss": 0.2837,
+      "step": 43685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.057521510834093e-05,
+      "loss": 0.2884,
+      "step": 43690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0572986694736784e-05,
+      "loss": 0.2677,
+      "step": 43695
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.057075807892765e-05,
+      "loss": 0.2645,
+      "step": 43700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.05685292609425e-05,
+      "loss": 0.2882,
+      "step": 43705
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0566300240810245e-05,
+      "loss": 0.285,
+      "step": 43710
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.056407101855985e-05,
+      "loss": 0.277,
+      "step": 43715
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.056184159422024e-05,
+      "loss": 0.2951,
+      "step": 43720
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0559611967820375e-05,
+      "loss": 0.2924,
+      "step": 43725
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.05573821393892e-05,
+      "loss": 0.3046,
+      "step": 43730
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.055515210895568e-05,
+      "loss": 0.3116,
+      "step": 43735
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.055292187654875e-05,
+      "loss": 0.2598,
+      "step": 43740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.055069144219739e-05,
+      "loss": 0.3018,
+      "step": 43745
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.054846080593054e-05,
+      "loss": 0.3049,
+      "step": 43750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.054622996777718e-05,
+      "loss": 0.3057,
+      "step": 43755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.054399892776627e-05,
+      "loss": 0.2754,
+      "step": 43760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.054176768592678e-05,
+      "loss": 0.2652,
+      "step": 43765
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.053953624228767e-05,
+      "loss": 0.2693,
+      "step": 43770
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.053730459687792e-05,
+      "loss": 0.297,
+      "step": 43775
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.053507274972652e-05,
+      "loss": 0.264,
+      "step": 43780
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.053284070086244e-05,
+      "loss": 0.314,
+      "step": 43785
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0530608450314654e-05,
+      "loss": 0.2932,
+      "step": 43790
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.052837599811216e-05,
+      "loss": 0.3142,
+      "step": 43795
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0526143344283916e-05,
+      "loss": 0.2902,
+      "step": 43800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0523910488858954e-05,
+      "loss": 0.294,
+      "step": 43805
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.052167743186623e-05,
+      "loss": 0.3107,
+      "step": 43810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0519444173334766e-05,
+      "loss": 0.2692,
+      "step": 43815
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.051721071329354e-05,
+      "loss": 0.2847,
+      "step": 43820
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0514977051771566e-05,
+      "loss": 0.2965,
+      "step": 43825
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.051274318879784e-05,
+      "loss": 0.2907,
+      "step": 43830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.051050912440135e-05,
+      "loss": 0.2911,
+      "step": 43835
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.050827485861114e-05,
+      "loss": 0.2847,
+      "step": 43840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.050604039145619e-05,
+      "loss": 0.2794,
+      "step": 43845
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0503805722965535e-05,
+      "loss": 0.2913,
+      "step": 43850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0501570853168166e-05,
+      "loss": 0.2668,
+      "step": 43855
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.049933578209313e-05,
+      "loss": 0.3133,
+      "step": 43860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.049710050976943e-05,
+      "loss": 0.2685,
+      "step": 43865
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.049486503622609e-05,
+      "loss": 0.289,
+      "step": 43870
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.049262936149213e-05,
+      "loss": 0.2724,
+      "step": 43875
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0490393485596605e-05,
+      "loss": 0.2723,
+      "step": 43880
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0488157408568516e-05,
+      "loss": 0.2928,
+      "step": 43885
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.048592113043692e-05,
+      "loss": 0.2693,
+      "step": 43890
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0483684651230835e-05,
+      "loss": 0.3124,
+      "step": 43895
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0481447970979316e-05,
+      "loss": 0.2454,
+      "step": 43900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.047921108971139e-05,
+      "loss": 0.2742,
+      "step": 43905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.0476974007456114e-05,
+      "loss": 0.2796,
+      "step": 43910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.047473672424253e-05,
+      "loss": 0.2946,
+      "step": 43915
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0472499240099695e-05,
+      "loss": 0.309,
+      "step": 43920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.047026155505664e-05,
+      "loss": 0.288,
+      "step": 43925
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.046802366914244e-05,
+      "loss": 0.286,
+      "step": 43930
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.046578558238615e-05,
+      "loss": 0.2997,
+      "step": 43935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0463547294816815e-05,
+      "loss": 0.2772,
+      "step": 43940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0461308806463525e-05,
+      "loss": 0.2691,
+      "step": 43945
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045907011735532e-05,
+      "loss": 0.2804,
+      "step": 43950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045683122752128e-05,
+      "loss": 0.2867,
+      "step": 43955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045459213699047e-05,
+      "loss": 0.2909,
+      "step": 43960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045235284579198e-05,
+      "loss": 0.3073,
+      "step": 43965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.045011335395486e-05,
+      "loss": 0.3136,
+      "step": 43970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.04478736615082e-05,
+      "loss": 0.288,
+      "step": 43975
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.044563376848107e-05,
+      "loss": 0.2732,
+      "step": 43980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0443393674902585e-05,
+      "loss": 0.3115,
+      "step": 43985
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.04411533808018e-05,
+      "loss": 0.2983,
+      "step": 43990
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.043891288620781e-05,
+      "loss": 0.2848,
+      "step": 43995
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0436672191149724e-05,
+      "loss": 0.2671,
+      "step": 44000
+    },
+    {
+      "epoch": 1.55,
+      "eval_loss": 0.279005765914917,
+      "eval_runtime": 10.5327,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 44000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.043443129565662e-05,
+      "loss": 0.2622,
+      "step": 44005
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.043219019975759e-05,
+      "loss": 0.2953,
+      "step": 44010
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.042994890348174e-05,
+      "loss": 0.2774,
+      "step": 44015
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.042770740685818e-05,
+      "loss": 0.3034,
+      "step": 44020
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0425465709916e-05,
+      "loss": 0.2754,
+      "step": 44025
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.042322381268432e-05,
+      "loss": 0.2879,
+      "step": 44030
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0420981715192235e-05,
+      "loss": 0.297,
+      "step": 44035
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.041873941746887e-05,
+      "loss": 0.282,
+      "step": 44040
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0416496919543336e-05,
+      "loss": 0.3116,
+      "step": 44045
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.041425422144475e-05,
+      "loss": 0.2781,
+      "step": 44050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.041201132320223e-05,
+      "loss": 0.2828,
+      "step": 44055
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.040976822484489e-05,
+      "loss": 0.3013,
+      "step": 44060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0407524926401874e-05,
+      "loss": 0.3005,
+      "step": 44065
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.04052814279023e-05,
+      "loss": 0.3076,
+      "step": 44070
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.04030377293753e-05,
+      "loss": 0.3003,
+      "step": 44075
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.040079383085001e-05,
+      "loss": 0.2927,
+      "step": 44080
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.039854973235556e-05,
+      "loss": 0.2855,
+      "step": 44085
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0396305433921086e-05,
+      "loss": 0.2697,
+      "step": 44090
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.039406093557573e-05,
+      "loss": 0.2696,
+      "step": 44095
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0391816237348646e-05,
+      "loss": 0.3029,
+      "step": 44100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038957133926896e-05,
+      "loss": 0.2862,
+      "step": 44105
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038732624136584e-05,
+      "loss": 0.2742,
+      "step": 44110
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038508094366843e-05,
+      "loss": 0.2863,
+      "step": 44115
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.038283544620588e-05,
+      "loss": 0.2809,
+      "step": 44120
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0380589749007357e-05,
+      "loss": 0.2874,
+      "step": 44125
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.037834385210201e-05,
+      "loss": 0.2926,
+      "step": 44130
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.037609775551899e-05,
+      "loss": 0.2669,
+      "step": 44135
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.037385145928749e-05,
+      "loss": 0.2645,
+      "step": 44140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0371604963436656e-05,
+      "loss": 0.2976,
+      "step": 44145
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.036935826799566e-05,
+      "loss": 0.2779,
+      "step": 44150
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0367111372993674e-05,
+      "loss": 0.2729,
+      "step": 44155
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.036486427845987e-05,
+      "loss": 0.2845,
+      "step": 44160
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.036261698442344e-05,
+      "loss": 0.276,
+      "step": 44165
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.036036949091354e-05,
+      "loss": 0.2767,
+      "step": 44170
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.035812179795938e-05,
+      "loss": 0.2749,
+      "step": 44175
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.035587390559012e-05,
+      "loss": 0.2965,
+      "step": 44180
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.035362581383496e-05,
+      "loss": 0.3134,
+      "step": 44185
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.035137752272309e-05,
+      "loss": 0.2744,
+      "step": 44190
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.0349129032283694e-05,
+      "loss": 0.2862,
+      "step": 44195
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0346880342545975e-05,
+      "loss": 0.2865,
+      "step": 44200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.034463145353913e-05,
+      "loss": 0.2837,
+      "step": 44205
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.034238236529235e-05,
+      "loss": 0.2748,
+      "step": 44210
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.034013307783486e-05,
+      "loss": 0.3044,
+      "step": 44215
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.033788359119585e-05,
+      "loss": 0.2856,
+      "step": 44220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0335633905404526e-05,
+      "loss": 0.307,
+      "step": 44225
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.033338402049009e-05,
+      "loss": 0.2752,
+      "step": 44230
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.033113393648178e-05,
+      "loss": 0.2831,
+      "step": 44235
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.03288836534088e-05,
+      "loss": 0.2967,
+      "step": 44240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0326633171300363e-05,
+      "loss": 0.2885,
+      "step": 44245
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.03243824901857e-05,
+      "loss": 0.2944,
+      "step": 44250
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.032213161009403e-05,
+      "loss": 0.2789,
+      "step": 44255
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.031988053105457e-05,
+      "loss": 0.2865,
+      "step": 44260
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.031762925309657e-05,
+      "loss": 0.2864,
+      "step": 44265
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0315377776249244e-05,
+      "loss": 0.2824,
+      "step": 44270
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.031312610054182e-05,
+      "loss": 0.2908,
+      "step": 44275
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.031087422600356e-05,
+      "loss": 0.2705,
+      "step": 44280
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.030862215266368e-05,
+      "loss": 0.2942,
+      "step": 44285
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.030636988055143e-05,
+      "loss": 0.2905,
+      "step": 44290
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.030411740969606e-05,
+      "loss": 0.2792,
+      "step": 44295
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.030186474012681e-05,
+      "loss": 0.2969,
+      "step": 44300
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.029961187187293e-05,
+      "loss": 0.2886,
+      "step": 44305
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.029735880496367e-05,
+      "loss": 0.2571,
+      "step": 44310
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.029510553942829e-05,
+      "loss": 0.3031,
+      "step": 44315
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.029285207529604e-05,
+      "loss": 0.2603,
+      "step": 44320
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.029059841259619e-05,
+      "loss": 0.2936,
+      "step": 44325
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028834455135799e-05,
+      "loss": 0.2819,
+      "step": 44330
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028609049161072e-05,
+      "loss": 0.3126,
+      "step": 44335
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028383623338363e-05,
+      "loss": 0.2795,
+      "step": 44340
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.028158177670601e-05,
+      "loss": 0.2877,
+      "step": 44345
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027932712160711e-05,
+      "loss": 0.2691,
+      "step": 44350
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027707226811622e-05,
+      "loss": 0.2824,
+      "step": 44355
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027481721626262e-05,
+      "loss": 0.2895,
+      "step": 44360
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0272561966075586e-05,
+      "loss": 0.2697,
+      "step": 44365
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.027030651758439e-05,
+      "loss": 0.2739,
+      "step": 44370
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.026805087081833e-05,
+      "loss": 0.3059,
+      "step": 44375
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.02657950258067e-05,
+      "loss": 0.2956,
+      "step": 44380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0263538982578774e-05,
+      "loss": 0.2596,
+      "step": 44385
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0261282741163855e-05,
+      "loss": 0.2915,
+      "step": 44390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025902630159124e-05,
+      "loss": 0.2936,
+      "step": 44395
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025676966389021e-05,
+      "loss": 0.2569,
+      "step": 44400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.02545128280901e-05,
+      "loss": 0.2738,
+      "step": 44405
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025225579422018e-05,
+      "loss": 0.3099,
+      "step": 44410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.025045002453369e-05,
+      "loss": 0.2883,
+      "step": 44415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0248192634212e-05,
+      "loss": 0.291,
+      "step": 44420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0245935045902575e-05,
+      "loss": 0.2972,
+      "step": 44425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0243677259634734e-05,
+      "loss": 0.291,
+      "step": 44430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0241419275437783e-05,
+      "loss": 0.2713,
+      "step": 44435
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023916109334105e-05,
+      "loss": 0.2927,
+      "step": 44440
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023690271337385e-05,
+      "loss": 0.3123,
+      "step": 44445
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0234644135565526e-05,
+      "loss": 0.2886,
+      "step": 44450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023238535994537e-05,
+      "loss": 0.2892,
+      "step": 44455
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.023012638654273e-05,
+      "loss": 0.2858,
+      "step": 44460
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022786721538694e-05,
+      "loss": 0.2648,
+      "step": 44465
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022560784650733e-05,
+      "loss": 0.2844,
+      "step": 44470
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.022334827993324e-05,
+      "loss": 0.2985,
+      "step": 44475
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.0221088515693995e-05,
+      "loss": 0.2637,
+      "step": 44480
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021882855381895e-05,
+      "loss": 0.2832,
+      "step": 44485
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021656839433745e-05,
+      "loss": 0.3021,
+      "step": 44490
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0214308037278836e-05,
+      "loss": 0.2737,
+      "step": 44495
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.021204748267246e-05,
+      "loss": 0.2878,
+      "step": 44500
+    },
+    {
+      "epoch": 1.57,
+      "eval_loss": 0.2789715826511383,
+      "eval_runtime": 10.538,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 44500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020978673054767e-05,
+      "loss": 0.2947,
+      "step": 44505
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020752578093382e-05,
+      "loss": 0.2945,
+      "step": 44510
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020526463386028e-05,
+      "loss": 0.2918,
+      "step": 44515
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020300328935639e-05,
+      "loss": 0.3068,
+      "step": 44520
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.020074174745153e-05,
+      "loss": 0.2898,
+      "step": 44525
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0198480008175044e-05,
+      "loss": 0.2909,
+      "step": 44530
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019621807155632e-05,
+      "loss": 0.2707,
+      "step": 44535
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019395593762472e-05,
+      "loss": 0.2816,
+      "step": 44540
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.019169360640961e-05,
+      "loss": 0.2972,
+      "step": 44545
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.018943107794038e-05,
+      "loss": 0.2918,
+      "step": 44550
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.018716835224638e-05,
+      "loss": 0.2743,
+      "step": 44555
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0184905429357026e-05,
+      "loss": 0.2786,
+      "step": 44560
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.018264230930167e-05,
+      "loss": 0.2878,
+      "step": 44565
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0180378992109714e-05,
+      "loss": 0.2979,
+      "step": 44570
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.017811547781054e-05,
+      "loss": 0.2852,
+      "step": 44575
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0175851766433536e-05,
+      "loss": 0.2772,
+      "step": 44580
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.01735878580081e-05,
+      "loss": 0.287,
+      "step": 44585
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0171323752563636e-05,
+      "loss": 0.2819,
+      "step": 44590
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016905945012952e-05,
+      "loss": 0.2993,
+      "step": 44595
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0166794950735164e-05,
+      "loss": 0.2784,
+      "step": 44600
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016453025440997e-05,
+      "loss": 0.2872,
+      "step": 44605
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016226536118335e-05,
+      "loss": 0.2724,
+      "step": 44610
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.016000027108471e-05,
+      "loss": 0.2988,
+      "step": 44615
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.015773498414345e-05,
+      "loss": 0.2687,
+      "step": 44620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.015546950038899e-05,
+      "loss": 0.2678,
+      "step": 44625
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.015320381985075e-05,
+      "loss": 0.2811,
+      "step": 44630
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.015093794255814e-05,
+      "loss": 0.2692,
+      "step": 44635
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0148671868540596e-05,
+      "loss": 0.2624,
+      "step": 44640
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.014640559782752e-05,
+      "loss": 0.3,
+      "step": 44645
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.014413913044835e-05,
+      "loss": 0.2471,
+      "step": 44650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0141872466432515e-05,
+      "loss": 0.2739,
+      "step": 44655
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.013960560580945e-05,
+      "loss": 0.2639,
+      "step": 44660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.013733854860858e-05,
+      "loss": 0.2871,
+      "step": 44665
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.013507129485935e-05,
+      "loss": 0.2904,
+      "step": 44670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0132803844591186e-05,
+      "loss": 0.2661,
+      "step": 44675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.013053619783355e-05,
+      "loss": 0.2731,
+      "step": 44680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.012826835461585e-05,
+      "loss": 0.2718,
+      "step": 44685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.012600031496757e-05,
+      "loss": 0.2877,
+      "step": 44690
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.012373207891814e-05,
+      "loss": 0.2779,
+      "step": 44695
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.012146364649702e-05,
+      "loss": 0.288,
+      "step": 44700
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.011919501773366e-05,
+      "loss": 0.2802,
+      "step": 44705
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.011692619265751e-05,
+      "loss": 0.2772,
+      "step": 44710
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.011465717129805e-05,
+      "loss": 0.2637,
+      "step": 44715
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.011238795368471e-05,
+      "loss": 0.2746,
+      "step": 44720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0110118539846986e-05,
+      "loss": 0.29,
+      "step": 44725
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0107848929814325e-05,
+      "loss": 0.3046,
+      "step": 44730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.01055791236162e-05,
+      "loss": 0.3014,
+      "step": 44735
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.0103309121282085e-05,
+      "loss": 0.2705,
+      "step": 44740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.010103892284146e-05,
+      "loss": 0.3123,
+      "step": 44745
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.009876852832379e-05,
+      "loss": 0.2675,
+      "step": 44750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.009649793775857e-05,
+      "loss": 0.2937,
+      "step": 44755
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.009422715117526e-05,
+      "loss": 0.2638,
+      "step": 44760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.009195616860337e-05,
+      "loss": 0.3036,
+      "step": 44765
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.008968499007237e-05,
+      "loss": 0.2876,
+      "step": 44770
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0087413615611756e-05,
+      "loss": 0.2855,
+      "step": 44775
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.008514204525101e-05,
+      "loss": 0.3001,
+      "step": 44780
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.008287027901965e-05,
+      "loss": 0.3207,
+      "step": 44785
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.008059831694716e-05,
+      "loss": 0.2864,
+      "step": 44790
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0078326159063026e-05,
+      "loss": 0.2738,
+      "step": 44795
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.007605380539677e-05,
+      "loss": 0.2664,
+      "step": 44800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0073781255977894e-05,
+      "loss": 0.2699,
+      "step": 44805
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.00715085108359e-05,
+      "loss": 0.2927,
+      "step": 44810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.00692355700003e-05,
+      "loss": 0.2615,
+      "step": 44815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.006696243350061e-05,
+      "loss": 0.2782,
+      "step": 44820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0064689101366336e-05,
+      "loss": 0.2823,
+      "step": 44825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.006241557362701e-05,
+      "loss": 0.2789,
+      "step": 44830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0060141850312135e-05,
+      "loss": 0.2832,
+      "step": 44835
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.005786793145126e-05,
+      "loss": 0.2645,
+      "step": 44840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.005559381707388e-05,
+      "loss": 0.275,
+      "step": 44845
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.005331950720953e-05,
+      "loss": 0.283,
+      "step": 44850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0051045001887755e-05,
+      "loss": 0.283,
+      "step": 44855
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.004877030113808e-05,
+      "loss": 0.3178,
+      "step": 44860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0046495404990043e-05,
+      "loss": 0.2851,
+      "step": 44865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.004422031347318e-05,
+      "loss": 0.2787,
+      "step": 44870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0041945026617024e-05,
+      "loss": 0.274,
+      "step": 44875
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0039669544451126e-05,
+      "loss": 0.278,
+      "step": 44880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.003739386700503e-05,
+      "loss": 0.2996,
+      "step": 44885
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.003511799430829e-05,
+      "loss": 0.3108,
+      "step": 44890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.003284192639045e-05,
+      "loss": 0.28,
+      "step": 44895
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.003056566328106e-05,
+      "loss": 0.2945,
+      "step": 44900
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.002828920500968e-05,
+      "loss": 0.3199,
+      "step": 44905
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.002601255160587e-05,
+      "loss": 0.2892,
+      "step": 44910
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.00237357030992e-05,
+      "loss": 0.3298,
+      "step": 44915
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.002145865951921e-05,
+      "loss": 0.2979,
+      "step": 44920
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.001918142089548e-05,
+      "loss": 0.2916,
+      "step": 44925
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.001690398725758e-05,
+      "loss": 0.2871,
+      "step": 44930
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.0014626358635084e-05,
+      "loss": 0.2833,
+      "step": 44935
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.001234853505755e-05,
+      "loss": 0.2868,
+      "step": 44940
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.001007051655457e-05,
+      "loss": 0.2803,
+      "step": 44945
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.000779230315571e-05,
+      "loss": 0.2863,
+      "step": 44950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.000551389489057e-05,
+      "loss": 0.2994,
+      "step": 44955
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.000323529178871e-05,
+      "loss": 0.3044,
+      "step": 44960
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.000095649387973e-05,
+      "loss": 0.2818,
+      "step": 44965
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9998677501193214e-05,
+      "loss": 0.2852,
+      "step": 44970
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.999639831375876e-05,
+      "loss": 0.2878,
+      "step": 44975
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.999411893160595e-05,
+      "loss": 0.283,
+      "step": 44980
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9991839354764394e-05,
+      "loss": 0.2982,
+      "step": 44985
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.998955958326368e-05,
+      "loss": 0.2903,
+      "step": 44990
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.998727961713342e-05,
+      "loss": 0.305,
+      "step": 44995
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.99849994564032e-05,
+      "loss": 0.2845,
+      "step": 45000
+    },
+    {
+      "epoch": 1.58,
+      "eval_loss": 0.27851128578186035,
+      "eval_runtime": 10.5579,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 9.472,
+      "step": 45000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.998271910110264e-05,
+      "loss": 0.3026,
+      "step": 45005
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.998043855126135e-05,
+      "loss": 0.2869,
+      "step": 45010
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.997815780690895e-05,
+      "loss": 0.2622,
+      "step": 45015
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.997587686807502e-05,
+      "loss": 0.3,
+      "step": 45020
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.997359573478921e-05,
+      "loss": 0.2866,
+      "step": 45025
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.997131440708112e-05,
+      "loss": 0.2857,
+      "step": 45030
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9969032884980386e-05,
+      "loss": 0.2889,
+      "step": 45035
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9966751168516626e-05,
+      "loss": 0.2875,
+      "step": 45040
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.996446925771946e-05,
+      "loss": 0.3088,
+      "step": 45045
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9962187152618525e-05,
+      "loss": 0.3145,
+      "step": 45050
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.995990485324345e-05,
+      "loss": 0.291,
+      "step": 45055
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.995762235962387e-05,
+      "loss": 0.2996,
+      "step": 45060
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.995533967178941e-05,
+      "loss": 0.2641,
+      "step": 45065
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.995305678976974e-05,
+      "loss": 0.293,
+      "step": 45070
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.995077371359448e-05,
+      "loss": 0.2948,
+      "step": 45075
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.994849044329326e-05,
+      "loss": 0.2961,
+      "step": 45080
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9946206978895754e-05,
+      "loss": 0.2595,
+      "step": 45085
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.99439233204316e-05,
+      "loss": 0.294,
+      "step": 45090
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.994163946793045e-05,
+      "loss": 0.291,
+      "step": 45095
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.993935542142196e-05,
+      "loss": 0.2901,
+      "step": 45100
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.993707118093578e-05,
+      "loss": 0.2719,
+      "step": 45105
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.993478674650158e-05,
+      "loss": 0.2859,
+      "step": 45110
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.993250211814901e-05,
+      "loss": 0.2932,
+      "step": 45115
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.993021729590775e-05,
+      "loss": 0.2849,
+      "step": 45120
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.992793227980744e-05,
+      "loss": 0.2834,
+      "step": 45125
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.992564706987779e-05,
+      "loss": 0.2918,
+      "step": 45130
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.992336166614844e-05,
+      "loss": 0.2689,
+      "step": 45135
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9921076068649073e-05,
+      "loss": 0.3011,
+      "step": 45140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.991879027740936e-05,
+      "loss": 0.2934,
+      "step": 45145
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9916504292459003e-05,
+      "loss": 0.2646,
+      "step": 45150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.991421811382765e-05,
+      "loss": 0.313,
+      "step": 45155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9911931741545014e-05,
+      "loss": 0.2736,
+      "step": 45160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.990964517564077e-05,
+      "loss": 0.288,
+      "step": 45165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9907358416144606e-05,
+      "loss": 0.315,
+      "step": 45170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.990507146308622e-05,
+      "loss": 0.2904,
+      "step": 45175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.99027843164953e-05,
+      "loss": 0.2748,
+      "step": 45180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.990049697640156e-05,
+      "loss": 0.2799,
+      "step": 45185
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9898209442834674e-05,
+      "loss": 0.2663,
+      "step": 45190
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9895921715824355e-05,
+      "loss": 0.2832,
+      "step": 45195
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.989363379540032e-05,
+      "loss": 0.2778,
+      "step": 45200
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.989134568159226e-05,
+      "loss": 0.3048,
+      "step": 45205
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.988905737442989e-05,
+      "loss": 0.283,
+      "step": 45210
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.988676887394292e-05,
+      "loss": 0.2761,
+      "step": 45215
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.988448018016107e-05,
+      "loss": 0.2957,
+      "step": 45220
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9882191293114055e-05,
+      "loss": 0.2997,
+      "step": 45225
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9879902212831586e-05,
+      "loss": 0.2857,
+      "step": 45230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.98776129393434e-05,
+      "loss": 0.2847,
+      "step": 45235
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.987532347267922e-05,
+      "loss": 0.318,
+      "step": 45240
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.987303381286876e-05,
+      "loss": 0.2782,
+      "step": 45245
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.987074395994176e-05,
+      "loss": 0.2763,
+      "step": 45250
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.986845391392795e-05,
+      "loss": 0.2799,
+      "step": 45255
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.986616367485706e-05,
+      "loss": 0.2632,
+      "step": 45260
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.986387324275884e-05,
+      "loss": 0.3109,
+      "step": 45265
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9861582617663014e-05,
+      "loss": 0.3049,
+      "step": 45270
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9859291799599334e-05,
+      "loss": 0.287,
+      "step": 45275
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.985700078859754e-05,
+      "loss": 0.2702,
+      "step": 45280
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.985470958468739e-05,
+      "loss": 0.3012,
+      "step": 45285
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.985241818789862e-05,
+      "loss": 0.3009,
+      "step": 45290
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.985012659826099e-05,
+      "loss": 0.3043,
+      "step": 45295
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.984783481580425e-05,
+      "loss": 0.2747,
+      "step": 45300
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.984554284055816e-05,
+      "loss": 0.3035,
+      "step": 45305
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.984325067255248e-05,
+      "loss": 0.2666,
+      "step": 45310
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.984095831181698e-05,
+      "loss": 0.3017,
+      "step": 45315
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.983866575838141e-05,
+      "loss": 0.2895,
+      "step": 45320
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.9836373012275543e-05,
+      "loss": 0.3074,
+      "step": 45325
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 3.983408007352916e-05,
+      "loss": 0.2672,
+      "step": 45330
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.983178694217201e-05,
+      "loss": 0.2651,
+      "step": 45335
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.982949361823388e-05,
+      "loss": 0.3096,
+      "step": 45340
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.982720010174456e-05,
+      "loss": 0.2666,
+      "step": 45345
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.982490639273381e-05,
+      "loss": 0.2852,
+      "step": 45350
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9822612491231425e-05,
+      "loss": 0.2856,
+      "step": 45355
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.982031839726718e-05,
+      "loss": 0.2854,
+      "step": 45360
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.981802411087087e-05,
+      "loss": 0.2934,
+      "step": 45365
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.981572963207229e-05,
+      "loss": 0.3026,
+      "step": 45370
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.981343496090121e-05,
+      "loss": 0.3136,
+      "step": 45375
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9811140097387456e-05,
+      "loss": 0.2767,
+      "step": 45380
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.98088450415608e-05,
+      "loss": 0.299,
+      "step": 45385
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9806549793451046e-05,
+      "loss": 0.2696,
+      "step": 45390
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9804254353088003e-05,
+      "loss": 0.2684,
+      "step": 45395
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.980195872050147e-05,
+      "loss": 0.2764,
+      "step": 45400
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.979966289572127e-05,
+      "loss": 0.269,
+      "step": 45405
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.97973668787772e-05,
+      "loss": 0.269,
+      "step": 45410
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9795070669699064e-05,
+      "loss": 0.2676,
+      "step": 45415
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9792774268516687e-05,
+      "loss": 0.2857,
+      "step": 45420
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.979047767525989e-05,
+      "loss": 0.3097,
+      "step": 45425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.978818088995848e-05,
+      "loss": 0.2589,
+      "step": 45430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.97858839126423e-05,
+      "loss": 0.2776,
+      "step": 45435
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.978358674334115e-05,
+      "loss": 0.2829,
+      "step": 45440
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9781289382084875e-05,
+      "loss": 0.3155,
+      "step": 45445
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9778991828903286e-05,
+      "loss": 0.2622,
+      "step": 45450
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.977669408382625e-05,
+      "loss": 0.2948,
+      "step": 45455
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.977439614688356e-05,
+      "loss": 0.2935,
+      "step": 45460
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9772098018105085e-05,
+      "loss": 0.259,
+      "step": 45465
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9769799697520645e-05,
+      "loss": 0.2842,
+      "step": 45470
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.976750118516009e-05,
+      "loss": 0.3044,
+      "step": 45475
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.976520248105326e-05,
+      "loss": 0.2931,
+      "step": 45480
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.976290358523002e-05,
+      "loss": 0.272,
+      "step": 45485
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9760604497720196e-05,
+      "loss": 0.2871,
+      "step": 45490
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.975830521855366e-05,
+      "loss": 0.2799,
+      "step": 45495
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.975600574776024e-05,
+      "loss": 0.2895,
+      "step": 45500
+    },
+    {
+      "epoch": 1.6,
+      "eval_loss": 0.2773810625076294,
+      "eval_runtime": 10.5526,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 45500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.975370608536982e-05,
+      "loss": 0.2877,
+      "step": 45505
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.975140623141226e-05,
+      "loss": 0.2763,
+      "step": 45510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9749106185917394e-05,
+      "loss": 0.2933,
+      "step": 45515
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.974680594891511e-05,
+      "loss": 0.2566,
+      "step": 45520
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.974450552043527e-05,
+      "loss": 0.2687,
+      "step": 45525
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.974220490050775e-05,
+      "loss": 0.2882,
+      "step": 45530
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.97399040891624e-05,
+      "loss": 0.2873,
+      "step": 45535
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.973760308642912e-05,
+      "loss": 0.2725,
+      "step": 45540
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.973530189233777e-05,
+      "loss": 0.297,
+      "step": 45545
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9733000506918245e-05,
+      "loss": 0.2573,
+      "step": 45550
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.973069893020041e-05,
+      "loss": 0.284,
+      "step": 45555
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.972839716221416e-05,
+      "loss": 0.2903,
+      "step": 45560
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.972609520298938e-05,
+      "loss": 0.2788,
+      "step": 45565
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9723793052555957e-05,
+      "loss": 0.2796,
+      "step": 45570
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9721490710943784e-05,
+      "loss": 0.2792,
+      "step": 45575
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.971918817818276e-05,
+      "loss": 0.3132,
+      "step": 45580
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.971688545430277e-05,
+      "loss": 0.2981,
+      "step": 45585
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.971458253933373e-05,
+      "loss": 0.2772,
+      "step": 45590
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.971227943330553e-05,
+      "loss": 0.3021,
+      "step": 45595
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9709976136248074e-05,
+      "loss": 0.2925,
+      "step": 45600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9707672648191274e-05,
+      "loss": 0.2876,
+      "step": 45605
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9705368969165036e-05,
+      "loss": 0.293,
+      "step": 45610
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9703065099199275e-05,
+      "loss": 0.2589,
+      "step": 45615
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.970076103832391e-05,
+      "loss": 0.296,
+      "step": 45620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.969845678656884e-05,
+      "loss": 0.2924,
+      "step": 45625
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9696152343963996e-05,
+      "loss": 0.3112,
+      "step": 45630
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9693847710539305e-05,
+      "loss": 0.3033,
+      "step": 45635
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9691542886324684e-05,
+      "loss": 0.2816,
+      "step": 45640
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.968923787135006e-05,
+      "loss": 0.2761,
+      "step": 45645
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9686932665645357e-05,
+      "loss": 0.2845,
+      "step": 45650
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.968462726924052e-05,
+      "loss": 0.2957,
+      "step": 45655
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9682321682165474e-05,
+      "loss": 0.2873,
+      "step": 45660
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.968001590445016e-05,
+      "loss": 0.2757,
+      "step": 45665
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.967770993612451e-05,
+      "loss": 0.2825,
+      "step": 45670
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.967540377721847e-05,
+      "loss": 0.2928,
+      "step": 45675
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9673097427761975e-05,
+      "loss": 0.2877,
+      "step": 45680
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.967079088778499e-05,
+      "loss": 0.2864,
+      "step": 45685
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9668484157317455e-05,
+      "loss": 0.2722,
+      "step": 45690
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.966617723638931e-05,
+      "loss": 0.281,
+      "step": 45695
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.966387012503053e-05,
+      "loss": 0.2646,
+      "step": 45700
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.966156282327106e-05,
+      "loss": 0.2854,
+      "step": 45705
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.965925533114085e-05,
+      "loss": 0.2768,
+      "step": 45710
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.965694764866988e-05,
+      "loss": 0.2905,
+      "step": 45715
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9654639775888093e-05,
+      "loss": 0.2845,
+      "step": 45720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9652331712825474e-05,
+      "loss": 0.2775,
+      "step": 45725
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9650023459511983e-05,
+      "loss": 0.2828,
+      "step": 45730
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9647715015977594e-05,
+      "loss": 0.3034,
+      "step": 45735
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.964540638225227e-05,
+      "loss": 0.2993,
+      "step": 45740
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.964309755836601e-05,
+      "loss": 0.3023,
+      "step": 45745
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.964078854434876e-05,
+      "loss": 0.2665,
+      "step": 45750
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9638479340230535e-05,
+      "loss": 0.2827,
+      "step": 45755
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.96361699460413e-05,
+      "loss": 0.2882,
+      "step": 45760
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9633860361811036e-05,
+      "loss": 0.2943,
+      "step": 45765
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.963155058756975e-05,
+      "loss": 0.2901,
+      "step": 45770
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.962924062334742e-05,
+      "loss": 0.283,
+      "step": 45775
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.962693046917404e-05,
+      "loss": 0.2823,
+      "step": 45780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.96246201250796e-05,
+      "loss": 0.2808,
+      "step": 45785
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.962230959109411e-05,
+      "loss": 0.2731,
+      "step": 45790
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961999886724756e-05,
+      "loss": 0.2726,
+      "step": 45795
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961768795356998e-05,
+      "loss": 0.306,
+      "step": 45800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961537685009134e-05,
+      "loss": 0.2753,
+      "step": 45805
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961306555684167e-05,
+      "loss": 0.2986,
+      "step": 45810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.961075407385097e-05,
+      "loss": 0.2763,
+      "step": 45815
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.960844240114926e-05,
+      "loss": 0.2872,
+      "step": 45820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.960613053876655e-05,
+      "loss": 0.2956,
+      "step": 45825
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.960381848673286e-05,
+      "loss": 0.306,
+      "step": 45830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9601506245078216e-05,
+      "loss": 0.2837,
+      "step": 45835
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.959919381383262e-05,
+      "loss": 0.2831,
+      "step": 45840
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.959688119302614e-05,
+      "loss": 0.2889,
+      "step": 45845
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9594568382688755e-05,
+      "loss": 0.2778,
+      "step": 45850
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.959225538285053e-05,
+      "loss": 0.2599,
+      "step": 45855
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9589942193541486e-05,
+      "loss": 0.2797,
+      "step": 45860
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.958762881479165e-05,
+      "loss": 0.293,
+      "step": 45865
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9585315246631074e-05,
+      "loss": 0.283,
+      "step": 45870
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9583001489089786e-05,
+      "loss": 0.2734,
+      "step": 45875
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9580687542197834e-05,
+      "loss": 0.3061,
+      "step": 45880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.957837340598527e-05,
+      "loss": 0.2923,
+      "step": 45885
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.957605908048213e-05,
+      "loss": 0.3052,
+      "step": 45890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.9573744565718476e-05,
+      "loss": 0.2921,
+      "step": 45895
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 3.957142986172434e-05,
+      "loss": 0.2828,
+      "step": 45900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.95691149685298e-05,
+      "loss": 0.29,
+      "step": 45905
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.956679988616489e-05,
+      "loss": 0.313,
+      "step": 45910
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9564484614659704e-05,
+      "loss": 0.2805,
+      "step": 45915
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.956216915404427e-05,
+      "loss": 0.2935,
+      "step": 45920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.955985350434868e-05,
+      "loss": 0.3025,
+      "step": 45925
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.955753766560297e-05,
+      "loss": 0.2691,
+      "step": 45930
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.955522163783724e-05,
+      "loss": 0.2919,
+      "step": 45935
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.955290542108154e-05,
+      "loss": 0.289,
+      "step": 45940
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.955058901536596e-05,
+      "loss": 0.2994,
+      "step": 45945
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9548272420720566e-05,
+      "loss": 0.2907,
+      "step": 45950
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9545955637175446e-05,
+      "loss": 0.2685,
+      "step": 45955
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9543638664760675e-05,
+      "loss": 0.2935,
+      "step": 45960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9541321503506346e-05,
+      "loss": 0.279,
+      "step": 45965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.953900415344254e-05,
+      "loss": 0.3079,
+      "step": 45970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9536686614599346e-05,
+      "loss": 0.2906,
+      "step": 45975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.953436888700684e-05,
+      "loss": 0.2907,
+      "step": 45980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9532050970695153e-05,
+      "loss": 0.2623,
+      "step": 45985
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.952973286569436e-05,
+      "loss": 0.2741,
+      "step": 45990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9527414572034555e-05,
+      "loss": 0.2883,
+      "step": 45995
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.952509608974585e-05,
+      "loss": 0.2835,
+      "step": 46000
+    },
+    {
+      "epoch": 1.62,
+      "eval_loss": 0.2772733271121979,
+      "eval_runtime": 10.5485,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 46000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.952277741885834e-05,
+      "loss": 0.2889,
+      "step": 46005
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9520458559402135e-05,
+      "loss": 0.298,
+      "step": 46010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.951813951140735e-05,
+      "loss": 0.2769,
+      "step": 46015
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.951582027490409e-05,
+      "loss": 0.2866,
+      "step": 46020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.951350084992246e-05,
+      "loss": 0.3019,
+      "step": 46025
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.95111812364926e-05,
+      "loss": 0.3141,
+      "step": 46030
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.950886143464461e-05,
+      "loss": 0.2843,
+      "step": 46035
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.950654144440862e-05,
+      "loss": 0.2648,
+      "step": 46040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9504221265814746e-05,
+      "loss": 0.2859,
+      "step": 46045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.950190089889311e-05,
+      "loss": 0.3057,
+      "step": 46050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9499580343673856e-05,
+      "loss": 0.2912,
+      "step": 46055
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.949725960018711e-05,
+      "loss": 0.2906,
+      "step": 46060
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9494938668463e-05,
+      "loss": 0.2849,
+      "step": 46065
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.949261754853166e-05,
+      "loss": 0.2912,
+      "step": 46070
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.949029624042323e-05,
+      "loss": 0.3034,
+      "step": 46075
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.948797474416786e-05,
+      "loss": 0.3001,
+      "step": 46080
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.948565305979568e-05,
+      "loss": 0.2599,
+      "step": 46085
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.948333118733685e-05,
+      "loss": 0.3083,
+      "step": 46090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.94810091268215e-05,
+      "loss": 0.2711,
+      "step": 46095
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.947868687827979e-05,
+      "loss": 0.2913,
+      "step": 46100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9476364441741873e-05,
+      "loss": 0.2639,
+      "step": 46105
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.947404181723791e-05,
+      "loss": 0.2943,
+      "step": 46110
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.947171900479805e-05,
+      "loss": 0.293,
+      "step": 46115
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9469396004452453e-05,
+      "loss": 0.2659,
+      "step": 46120
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.946707281623129e-05,
+      "loss": 0.271,
+      "step": 46125
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.946474944016472e-05,
+      "loss": 0.2975,
+      "step": 46130
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9462425876282904e-05,
+      "loss": 0.2955,
+      "step": 46135
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9460102124616027e-05,
+      "loss": 0.2788,
+      "step": 46140
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.945777818519425e-05,
+      "loss": 0.2802,
+      "step": 46145
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9455454058047753e-05,
+      "loss": 0.2871,
+      "step": 46150
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9453129743206714e-05,
+      "loss": 0.2792,
+      "step": 46155
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9450805240701304e-05,
+      "loss": 0.301,
+      "step": 46160
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.944848055056172e-05,
+      "loss": 0.2779,
+      "step": 46165
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.944615567281812e-05,
+      "loss": 0.2917,
+      "step": 46170
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.944383060750072e-05,
+      "loss": 0.2802,
+      "step": 46175
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.944150535463971e-05,
+      "loss": 0.308,
+      "step": 46180
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 3.9439179914265255e-05,
+      "loss": 0.2955,
+      "step": 46185
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9436854286407565e-05,
+      "loss": 0.2685,
+      "step": 46190
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.943452847109684e-05,
+      "loss": 0.277,
+      "step": 46195
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.943220246836328e-05,
+      "loss": 0.3088,
+      "step": 46200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9429876278237083e-05,
+      "loss": 0.265,
+      "step": 46205
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9427549900748445e-05,
+      "loss": 0.2596,
+      "step": 46210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.942522333592759e-05,
+      "loss": 0.2739,
+      "step": 46215
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.94228965838047e-05,
+      "loss": 0.2872,
+      "step": 46220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.942056964441001e-05,
+      "loss": 0.2797,
+      "step": 46225
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.941824251777373e-05,
+      "loss": 0.278,
+      "step": 46230
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9415915203926066e-05,
+      "loss": 0.291,
+      "step": 46235
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.941358770289725e-05,
+      "loss": 0.2806,
+      "step": 46240
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.94112600147175e-05,
+      "loss": 0.2903,
+      "step": 46245
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.940893213941703e-05,
+      "loss": 0.2905,
+      "step": 46250
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9406604077026073e-05,
+      "loss": 0.2838,
+      "step": 46255
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9404275827574855e-05,
+      "loss": 0.2762,
+      "step": 46260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9401947391093606e-05,
+      "loss": 0.2821,
+      "step": 46265
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.939961876761257e-05,
+      "loss": 0.3049,
+      "step": 46270
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.939728995716196e-05,
+      "loss": 0.2822,
+      "step": 46275
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9394960959772045e-05,
+      "loss": 0.2667,
+      "step": 46280
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.939263177547304e-05,
+      "loss": 0.2875,
+      "step": 46285
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9390302404295195e-05,
+      "loss": 0.2996,
+      "step": 46290
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.938797284626876e-05,
+      "loss": 0.2871,
+      "step": 46295
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9385643101423985e-05,
+      "loss": 0.2833,
+      "step": 46300
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9383313169791104e-05,
+      "loss": 0.2626,
+      "step": 46305
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9380983051400387e-05,
+      "loss": 0.2652,
+      "step": 46310
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.937865274628208e-05,
+      "loss": 0.2765,
+      "step": 46315
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.937632225446645e-05,
+      "loss": 0.2799,
+      "step": 46320
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.937399157598374e-05,
+      "loss": 0.2789,
+      "step": 46325
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.937166071086422e-05,
+      "loss": 0.2795,
+      "step": 46330
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9369329659138165e-05,
+      "loss": 0.2931,
+      "step": 46335
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.936699842083583e-05,
+      "loss": 0.2742,
+      "step": 46340
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.936466699598749e-05,
+      "loss": 0.2872,
+      "step": 46345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9362335384623414e-05,
+      "loss": 0.2813,
+      "step": 46350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9360003586773886e-05,
+      "loss": 0.2804,
+      "step": 46355
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.935767160246917e-05,
+      "loss": 0.2565,
+      "step": 46360
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.935533943173954e-05,
+      "loss": 0.2673,
+      "step": 46365
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.93530070746153e-05,
+      "loss": 0.2911,
+      "step": 46370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.935067453112673e-05,
+      "loss": 0.2806,
+      "step": 46375
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.934834180130409e-05,
+      "loss": 0.2451,
+      "step": 46380
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.93460088851777e-05,
+      "loss": 0.2758,
+      "step": 46385
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.934367578277783e-05,
+      "loss": 0.2824,
+      "step": 46390
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.934134249413479e-05,
+      "loss": 0.2827,
+      "step": 46395
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.933900901927887e-05,
+      "loss": 0.2792,
+      "step": 46400
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.933667535824036e-05,
+      "loss": 0.2811,
+      "step": 46405
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9334341511049576e-05,
+      "loss": 0.283,
+      "step": 46410
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9332007477736816e-05,
+      "loss": 0.2761,
+      "step": 46415
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.932967325833238e-05,
+      "loss": 0.2941,
+      "step": 46420
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.932733885286658e-05,
+      "loss": 0.2831,
+      "step": 46425
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.932500426136973e-05,
+      "loss": 0.2813,
+      "step": 46430
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.932266948387213e-05,
+      "loss": 0.2816,
+      "step": 46435
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.932033452040411e-05,
+      "loss": 0.3044,
+      "step": 46440
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.931799937099599e-05,
+      "loss": 0.2882,
+      "step": 46445
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.931566403567808e-05,
+      "loss": 0.2827,
+      "step": 46450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.931332851448071e-05,
+      "loss": 0.2983,
+      "step": 46455
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.931099280743419e-05,
+      "loss": 0.2995,
+      "step": 46460
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.9308656914568875e-05,
+      "loss": 0.2944,
+      "step": 46465
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 3.930632083591508e-05,
+      "loss": 0.2639,
+      "step": 46470
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.930398457150313e-05,
+      "loss": 0.2724,
+      "step": 46475
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.930164812136336e-05,
+      "loss": 0.2594,
+      "step": 46480
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9299311485526124e-05,
+      "loss": 0.2702,
+      "step": 46485
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.929697466402175e-05,
+      "loss": 0.2883,
+      "step": 46490
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9294637656880586e-05,
+      "loss": 0.2708,
+      "step": 46495
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.929230046413297e-05,
+      "loss": 0.2786,
+      "step": 46500
+    },
+    {
+      "epoch": 1.64,
+      "eval_loss": 0.2775975167751312,
+      "eval_runtime": 10.5364,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 46500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9289963085809246e-05,
+      "loss": 0.2771,
+      "step": 46505
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.928762552193977e-05,
+      "loss": 0.2936,
+      "step": 46510
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9285287772554905e-05,
+      "loss": 0.2759,
+      "step": 46515
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.928294983768498e-05,
+      "loss": 0.2953,
+      "step": 46520
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.928061171736036e-05,
+      "loss": 0.297,
+      "step": 46525
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9278273411611425e-05,
+      "loss": 0.2788,
+      "step": 46530
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.92759349204685e-05,
+      "loss": 0.2689,
+      "step": 46535
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.927359624396199e-05,
+      "loss": 0.2782,
+      "step": 46540
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.927125738212222e-05,
+      "loss": 0.2997,
+      "step": 46545
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9268918334979584e-05,
+      "loss": 0.2866,
+      "step": 46550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.926657910256445e-05,
+      "loss": 0.2718,
+      "step": 46555
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.926423968490719e-05,
+      "loss": 0.273,
+      "step": 46560
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9261900082038174e-05,
+      "loss": 0.2883,
+      "step": 46565
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9259560293987785e-05,
+      "loss": 0.2827,
+      "step": 46570
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.92572203207864e-05,
+      "loss": 0.2716,
+      "step": 46575
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.925488016246441e-05,
+      "loss": 0.3,
+      "step": 46580
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.92525398190522e-05,
+      "loss": 0.3001,
+      "step": 46585
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9250199290580144e-05,
+      "loss": 0.2725,
+      "step": 46590
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.924785857707865e-05,
+      "loss": 0.2713,
+      "step": 46595
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.924551767857809e-05,
+      "loss": 0.2979,
+      "step": 46600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.924317659510888e-05,
+      "loss": 0.2895,
+      "step": 46605
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.924083532670141e-05,
+      "loss": 0.285,
+      "step": 46610
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9238493873386075e-05,
+      "loss": 0.2717,
+      "step": 46615
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.923615223519328e-05,
+      "loss": 0.2959,
+      "step": 46620
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.923381041215343e-05,
+      "loss": 0.2938,
+      "step": 46625
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.923146840429693e-05,
+      "loss": 0.2913,
+      "step": 46630
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.92291262116542e-05,
+      "loss": 0.2844,
+      "step": 46635
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9226783834255634e-05,
+      "loss": 0.2926,
+      "step": 46640
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9224441272131664e-05,
+      "loss": 0.2879,
+      "step": 46645
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.922209852531269e-05,
+      "loss": 0.2757,
+      "step": 46650
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.921975559382914e-05,
+      "loss": 0.2858,
+      "step": 46655
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.921741247771144e-05,
+      "loss": 0.2691,
+      "step": 46660
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9215069176990006e-05,
+      "loss": 0.2943,
+      "step": 46665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.921272569169527e-05,
+      "loss": 0.2809,
+      "step": 46670
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9210382021857655e-05,
+      "loss": 0.2826,
+      "step": 46675
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.920803816750759e-05,
+      "loss": 0.2968,
+      "step": 46680
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.920569412867552e-05,
+      "loss": 0.2786,
+      "step": 46685
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9203349905391876e-05,
+      "loss": 0.3007,
+      "step": 46690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.920100549768708e-05,
+      "loss": 0.3029,
+      "step": 46695
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.91986609055916e-05,
+      "loss": 0.2705,
+      "step": 46700
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.919631612913586e-05,
+      "loss": 0.3101,
+      "step": 46705
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.919397116835031e-05,
+      "loss": 0.2752,
+      "step": 46710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.91916260232654e-05,
+      "loss": 0.3007,
+      "step": 46715
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9189280693911576e-05,
+      "loss": 0.268,
+      "step": 46720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.918693518031929e-05,
+      "loss": 0.2829,
+      "step": 46725
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.918458948251902e-05,
+      "loss": 0.2913,
+      "step": 46730
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9182243600541186e-05,
+      "loss": 0.2815,
+      "step": 46735
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.917989753441627e-05,
+      "loss": 0.2865,
+      "step": 46740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9177551284174724e-05,
+      "loss": 0.2695,
+      "step": 46745
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.917520484984702e-05,
+      "loss": 0.2711,
+      "step": 46750
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 3.9173327569863296e-05,
+      "loss": 0.2899,
+      "step": 46755
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.917098080425729e-05,
+      "loss": 0.2844,
+      "step": 46760
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.916863385465043e-05,
+      "loss": 0.2959,
+      "step": 46765
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.91662867210732e-05,
+      "loss": 0.276,
+      "step": 46770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9163939403556074e-05,
+      "loss": 0.2636,
+      "step": 46775
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.916159190212952e-05,
+      "loss": 0.2678,
+      "step": 46780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.915924421682404e-05,
+      "loss": 0.2807,
+      "step": 46785
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.91568963476701e-05,
+      "loss": 0.299,
+      "step": 46790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.915454829469819e-05,
+      "loss": 0.2929,
+      "step": 46795
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9152200057938796e-05,
+      "loss": 0.301,
+      "step": 46800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9149851637422415e-05,
+      "loss": 0.2898,
+      "step": 46805
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.914750303317953e-05,
+      "loss": 0.3064,
+      "step": 46810
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.914515424524065e-05,
+      "loss": 0.2903,
+      "step": 46815
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9142805273636265e-05,
+      "loss": 0.2944,
+      "step": 46820
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9140456118396865e-05,
+      "loss": 0.2811,
+      "step": 46825
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.913810677955297e-05,
+      "loss": 0.3244,
+      "step": 46830
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.913575725713507e-05,
+      "loss": 0.3122,
+      "step": 46835
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9133407551173676e-05,
+      "loss": 0.2729,
+      "step": 46840
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.91310576616993e-05,
+      "loss": 0.2804,
+      "step": 46845
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.912870758874246e-05,
+      "loss": 0.2933,
+      "step": 46850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9126357332333656e-05,
+      "loss": 0.2887,
+      "step": 46855
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.912400689250342e-05,
+      "loss": 0.2587,
+      "step": 46860
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9121656269282245e-05,
+      "loss": 0.3161,
+      "step": 46865
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9119305462700696e-05,
+      "loss": 0.3236,
+      "step": 46870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.911695447278925e-05,
+      "loss": 0.2894,
+      "step": 46875
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.911460329957846e-05,
+      "loss": 0.273,
+      "step": 46880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9112251943098844e-05,
+      "loss": 0.2766,
+      "step": 46885
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.910990040338094e-05,
+      "loss": 0.2898,
+      "step": 46890
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.910754868045528e-05,
+      "loss": 0.2967,
+      "step": 46895
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.910519677435239e-05,
+      "loss": 0.3025,
+      "step": 46900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.910284468510283e-05,
+      "loss": 0.2868,
+      "step": 46905
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.910049241273712e-05,
+      "loss": 0.3248,
+      "step": 46910
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.909813995728581e-05,
+      "loss": 0.2966,
+      "step": 46915
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.909578731877943e-05,
+      "loss": 0.276,
+      "step": 46920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.909343449724856e-05,
+      "loss": 0.2901,
+      "step": 46925
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.909108149272371e-05,
+      "loss": 0.2952,
+      "step": 46930
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.908872830523547e-05,
+      "loss": 0.2676,
+      "step": 46935
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9086374934814375e-05,
+      "loss": 0.2738,
+      "step": 46940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9084021381490986e-05,
+      "loss": 0.2668,
+      "step": 46945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.908166764529585e-05,
+      "loss": 0.2843,
+      "step": 46950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.907931372625955e-05,
+      "loss": 0.2964,
+      "step": 46955
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.907695962441264e-05,
+      "loss": 0.2733,
+      "step": 46960
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.907460533978569e-05,
+      "loss": 0.2851,
+      "step": 46965
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.907225087240925e-05,
+      "loss": 0.2835,
+      "step": 46970
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.906989622231392e-05,
+      "loss": 0.2729,
+      "step": 46975
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9067541389530256e-05,
+      "loss": 0.3028,
+      "step": 46980
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.906518637408884e-05,
+      "loss": 0.3031,
+      "step": 46985
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.906283117602024e-05,
+      "loss": 0.3047,
+      "step": 46990
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.906047579535504e-05,
+      "loss": 0.2737,
+      "step": 46995
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.905812023212385e-05,
+      "loss": 0.3082,
+      "step": 47000
+    },
+    {
+      "epoch": 1.65,
+      "eval_loss": 0.27732065320014954,
+      "eval_runtime": 10.5336,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 47000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.905576448635722e-05,
+      "loss": 0.3076,
+      "step": 47005
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.905340855808576e-05,
+      "loss": 0.2795,
+      "step": 47010
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.905105244734004e-05,
+      "loss": 0.2857,
+      "step": 47015
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9048696154150666e-05,
+      "loss": 0.2797,
+      "step": 47020
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9046339678548234e-05,
+      "loss": 0.2599,
+      "step": 47025
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.904398302056333e-05,
+      "loss": 0.2587,
+      "step": 47030
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.9041626180226575e-05,
+      "loss": 0.2786,
+      "step": 47035
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 3.903926915756856e-05,
+      "loss": 0.2823,
+      "step": 47040
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.903691195261987e-05,
+      "loss": 0.2879,
+      "step": 47045
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.903455456541114e-05,
+      "loss": 0.2644,
+      "step": 47050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.903219699597297e-05,
+      "loss": 0.2788,
+      "step": 47055
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.902983924433596e-05,
+      "loss": 0.2764,
+      "step": 47060
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.902748131053074e-05,
+      "loss": 0.2835,
+      "step": 47065
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9025123194587927e-05,
+      "loss": 0.2733,
+      "step": 47070
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.902276489653812e-05,
+      "loss": 0.2824,
+      "step": 47075
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.902040641641196e-05,
+      "loss": 0.2928,
+      "step": 47080
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9018047754240054e-05,
+      "loss": 0.2955,
+      "step": 47085
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.901568891005305e-05,
+      "loss": 0.284,
+      "step": 47090
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9013329883881544e-05,
+      "loss": 0.316,
+      "step": 47095
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.90109706757562e-05,
+      "loss": 0.297,
+      "step": 47100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.900861128570763e-05,
+      "loss": 0.2845,
+      "step": 47105
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.9006251713766474e-05,
+      "loss": 0.2836,
+      "step": 47110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.900389195996337e-05,
+      "loss": 0.3025,
+      "step": 47115
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.900153202432895e-05,
+      "loss": 0.2968,
+      "step": 47120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8999171906893875e-05,
+      "loss": 0.2915,
+      "step": 47125
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8996811607688774e-05,
+      "loss": 0.2617,
+      "step": 47130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8994451126744305e-05,
+      "loss": 0.2753,
+      "step": 47135
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.89920904640911e-05,
+      "loss": 0.3031,
+      "step": 47140
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.898972961975983e-05,
+      "loss": 0.3031,
+      "step": 47145
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.898736859378114e-05,
+      "loss": 0.2961,
+      "step": 47150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.898500738618568e-05,
+      "loss": 0.2889,
+      "step": 47155
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.898264599700412e-05,
+      "loss": 0.2899,
+      "step": 47160
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.898028442626712e-05,
+      "loss": 0.2805,
+      "step": 47165
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.897792267400533e-05,
+      "loss": 0.2881,
+      "step": 47170
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.897556074024944e-05,
+      "loss": 0.2914,
+      "step": 47175
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8973198625030094e-05,
+      "loss": 0.3109,
+      "step": 47180
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.897083632837797e-05,
+      "loss": 0.2974,
+      "step": 47185
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.896847385032375e-05,
+      "loss": 0.2844,
+      "step": 47190
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.89661111908981e-05,
+      "loss": 0.2585,
+      "step": 47195
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.89637483501317e-05,
+      "loss": 0.2966,
+      "step": 47200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.896138532805523e-05,
+      "loss": 0.2913,
+      "step": 47205
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8959022124699375e-05,
+      "loss": 0.2973,
+      "step": 47210
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.895665874009482e-05,
+      "loss": 0.3064,
+      "step": 47215
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8954295174272245e-05,
+      "loss": 0.2975,
+      "step": 47220
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.895193142726234e-05,
+      "loss": 0.2924,
+      "step": 47225
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.894956749909581e-05,
+      "loss": 0.2741,
+      "step": 47230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.894720338980332e-05,
+      "loss": 0.2908,
+      "step": 47235
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.89448390994156e-05,
+      "loss": 0.3165,
+      "step": 47240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.894247462796333e-05,
+      "loss": 0.2858,
+      "step": 47245
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.894010997547722e-05,
+      "loss": 0.268,
+      "step": 47250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.893774514198797e-05,
+      "loss": 0.2751,
+      "step": 47255
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8935380127526284e-05,
+      "loss": 0.2815,
+      "step": 47260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.893301493212286e-05,
+      "loss": 0.3071,
+      "step": 47265
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.8930649555808434e-05,
+      "loss": 0.2742,
+      "step": 47270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.89282839986137e-05,
+      "loss": 0.3003,
+      "step": 47275
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.892591826056937e-05,
+      "loss": 0.2946,
+      "step": 47280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.892355234170617e-05,
+      "loss": 0.3135,
+      "step": 47285
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.892118624205483e-05,
+      "loss": 0.2809,
+      "step": 47290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.891881996164605e-05,
+      "loss": 0.2887,
+      "step": 47295
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.891645350051057e-05,
+      "loss": 0.2931,
+      "step": 47300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.891408685867911e-05,
+      "loss": 0.3101,
+      "step": 47305
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.891172003618241e-05,
+      "loss": 0.3029,
+      "step": 47310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.890935303305118e-05,
+      "loss": 0.279,
+      "step": 47315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 3.890698584931618e-05,
+      "loss": 0.2807,
+      "step": 47320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.890461848500813e-05,
+      "loss": 0.2789,
+      "step": 47325
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.890225094015777e-05,
+      "loss": 0.2878,
+      "step": 47330
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.889988321479584e-05,
+      "loss": 0.2847,
+      "step": 47335
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.889751530895309e-05,
+      "loss": 0.2574,
+      "step": 47340
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8895147222660264e-05,
+      "loss": 0.2951,
+      "step": 47345
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8892778955948104e-05,
+      "loss": 0.3026,
+      "step": 47350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8890410508847376e-05,
+      "loss": 0.2767,
+      "step": 47355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.888804188138882e-05,
+      "loss": 0.303,
+      "step": 47360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.888567307360318e-05,
+      "loss": 0.2889,
+      "step": 47365
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8883304085521236e-05,
+      "loss": 0.2902,
+      "step": 47370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8880934917173735e-05,
+      "loss": 0.2759,
+      "step": 47375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.887856556859144e-05,
+      "loss": 0.3042,
+      "step": 47380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.887619603980512e-05,
+      "loss": 0.3202,
+      "step": 47385
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8873826330845545e-05,
+      "loss": 0.2898,
+      "step": 47390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.887145644174347e-05,
+      "loss": 0.2714,
+      "step": 47395
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.886908637252967e-05,
+      "loss": 0.2699,
+      "step": 47400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8866716123234934e-05,
+      "loss": 0.269,
+      "step": 47405
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8864345693890024e-05,
+      "loss": 0.2756,
+      "step": 47410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.886197508452572e-05,
+      "loss": 0.2592,
+      "step": 47415
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8859604295172816e-05,
+      "loss": 0.273,
+      "step": 47420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8857233325862074e-05,
+      "loss": 0.2826,
+      "step": 47425
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.885486217662428e-05,
+      "loss": 0.3095,
+      "step": 47430
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.885249084749025e-05,
+      "loss": 0.2889,
+      "step": 47435
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8850119338490745e-05,
+      "loss": 0.2691,
+      "step": 47440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.884774764965658e-05,
+      "loss": 0.279,
+      "step": 47445
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8845375781018524e-05,
+      "loss": 0.3029,
+      "step": 47450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.884300373260739e-05,
+      "loss": 0.2701,
+      "step": 47455
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8840631504453975e-05,
+      "loss": 0.2512,
+      "step": 47460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.883825909658908e-05,
+      "loss": 0.2756,
+      "step": 47465
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.883588650904352e-05,
+      "loss": 0.2821,
+      "step": 47470
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.883351374184808e-05,
+      "loss": 0.2952,
+      "step": 47475
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.883114079503359e-05,
+      "loss": 0.2917,
+      "step": 47480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.882876766863085e-05,
+      "loss": 0.2885,
+      "step": 47485
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8826394362670684e-05,
+      "loss": 0.3046,
+      "step": 47490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8824020877183886e-05,
+      "loss": 0.2945,
+      "step": 47495
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8821647212201286e-05,
+      "loss": 0.275,
+      "step": 47500
+    },
+    {
+      "epoch": 1.67,
+      "eval_loss": 0.2766549289226532,
+      "eval_runtime": 10.5338,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 47500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.881927336775371e-05,
+      "loss": 0.3027,
+      "step": 47505
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.881689934387198e-05,
+      "loss": 0.2904,
+      "step": 47510
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.881452514058691e-05,
+      "loss": 0.2958,
+      "step": 47515
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.881215075792934e-05,
+      "loss": 0.2804,
+      "step": 47520
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8809776195930095e-05,
+      "loss": 0.2713,
+      "step": 47525
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8807401454620004e-05,
+      "loss": 0.2963,
+      "step": 47530
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.880502653402991e-05,
+      "loss": 0.2641,
+      "step": 47535
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.880265143419065e-05,
+      "loss": 0.2932,
+      "step": 47540
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8800276155133044e-05,
+      "loss": 0.2958,
+      "step": 47545
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.879790069688795e-05,
+      "loss": 0.2934,
+      "step": 47550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8795525059486216e-05,
+      "loss": 0.284,
+      "step": 47555
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.879314924295867e-05,
+      "loss": 0.2561,
+      "step": 47560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8790773247336176e-05,
+      "loss": 0.2807,
+      "step": 47565
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8788397072649584e-05,
+      "loss": 0.2702,
+      "step": 47570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8786020718929737e-05,
+      "loss": 0.2759,
+      "step": 47575
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8783644186207494e-05,
+      "loss": 0.2952,
+      "step": 47580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.878126747451373e-05,
+      "loss": 0.2847,
+      "step": 47585
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.877889058387927e-05,
+      "loss": 0.2939,
+      "step": 47590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8776513514335014e-05,
+      "loss": 0.3119,
+      "step": 47595
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.8774136265911795e-05,
+      "loss": 0.2785,
+      "step": 47600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 3.87717588386405e-05,
+      "loss": 0.3148,
+      "step": 47605
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.876938123255199e-05,
+      "loss": 0.2824,
+      "step": 47610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8767003447677145e-05,
+      "loss": 0.3029,
+      "step": 47615
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.876462548404683e-05,
+      "loss": 0.2829,
+      "step": 47620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.876224734169193e-05,
+      "loss": 0.2836,
+      "step": 47625
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.87598690206433e-05,
+      "loss": 0.2844,
+      "step": 47630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.875749052093186e-05,
+      "loss": 0.2848,
+      "step": 47635
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.875511184258845e-05,
+      "loss": 0.287,
+      "step": 47640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8752732985644e-05,
+      "loss": 0.2968,
+      "step": 47645
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.875035395012936e-05,
+      "loss": 0.3014,
+      "step": 47650
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8747974736075444e-05,
+      "loss": 0.2758,
+      "step": 47655
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.874559534351314e-05,
+      "loss": 0.2948,
+      "step": 47660
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8743215772473326e-05,
+      "loss": 0.3045,
+      "step": 47665
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.874083602298693e-05,
+      "loss": 0.2855,
+      "step": 47670
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.873845609508482e-05,
+      "loss": 0.2766,
+      "step": 47675
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.873607598879791e-05,
+      "loss": 0.3062,
+      "step": 47680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8733695704157115e-05,
+      "loss": 0.2835,
+      "step": 47685
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8731315241193324e-05,
+      "loss": 0.3016,
+      "step": 47690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.872893459993745e-05,
+      "loss": 0.2987,
+      "step": 47695
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.872655378042043e-05,
+      "loss": 0.2697,
+      "step": 47700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.872417278267313e-05,
+      "loss": 0.2787,
+      "step": 47705
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.87217916067265e-05,
+      "loss": 0.2992,
+      "step": 47710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.871941025261146e-05,
+      "loss": 0.2912,
+      "step": 47715
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8717028720358904e-05,
+      "loss": 0.2638,
+      "step": 47720
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.871464700999978e-05,
+      "loss": 0.3052,
+      "step": 47725
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.871226512156499e-05,
+      "loss": 0.2839,
+      "step": 47730
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.870988305508549e-05,
+      "loss": 0.291,
+      "step": 47735
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.870750081059219e-05,
+      "loss": 0.2973,
+      "step": 47740
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.870511838811601e-05,
+      "loss": 0.2885,
+      "step": 47745
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.870273578768792e-05,
+      "loss": 0.2954,
+      "step": 47750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.870035300933883e-05,
+      "loss": 0.2994,
+      "step": 47755
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.869797005309969e-05,
+      "loss": 0.3017,
+      "step": 47760
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.869558691900142e-05,
+      "loss": 0.3055,
+      "step": 47765
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.869320360707499e-05,
+      "loss": 0.2589,
+      "step": 47770
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.869082011735133e-05,
+      "loss": 0.2804,
+      "step": 47775
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8688436449861406e-05,
+      "loss": 0.259,
+      "step": 47780
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8686052604636144e-05,
+      "loss": 0.2794,
+      "step": 47785
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.868366858170651e-05,
+      "loss": 0.285,
+      "step": 47790
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.868128438110346e-05,
+      "loss": 0.3079,
+      "step": 47795
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.867890000285795e-05,
+      "loss": 0.2741,
+      "step": 47800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.867651544700093e-05,
+      "loss": 0.2971,
+      "step": 47805
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8674130713563376e-05,
+      "loss": 0.285,
+      "step": 47810
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8671745802576254e-05,
+      "loss": 0.2907,
+      "step": 47815
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.866936071407051e-05,
+      "loss": 0.2883,
+      "step": 47820
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8666975448077135e-05,
+      "loss": 0.2868,
+      "step": 47825
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.866459000462708e-05,
+      "loss": 0.3158,
+      "step": 47830
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.866220438375134e-05,
+      "loss": 0.3045,
+      "step": 47835
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865981858548087e-05,
+      "loss": 0.2977,
+      "step": 47840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865743260984667e-05,
+      "loss": 0.2843,
+      "step": 47845
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865504645687969e-05,
+      "loss": 0.3085,
+      "step": 47850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865266012661095e-05,
+      "loss": 0.3038,
+      "step": 47855
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.865027361907141e-05,
+      "loss": 0.2652,
+      "step": 47860
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.864788693429206e-05,
+      "loss": 0.2831,
+      "step": 47865
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8645500072303886e-05,
+      "loss": 0.2692,
+      "step": 47870
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.86431130331379e-05,
+      "loss": 0.2932,
+      "step": 47875
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8640725816825075e-05,
+      "loss": 0.2795,
+      "step": 47880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.8638338423396415e-05,
+      "loss": 0.2969,
+      "step": 47885
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 3.863595085288292e-05,
+      "loss": 0.2695,
+      "step": 47890
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8633563105315585e-05,
+      "loss": 0.2813,
+      "step": 47895
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.863117518072542e-05,
+      "loss": 0.2802,
+      "step": 47900
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8628787079143434e-05,
+      "loss": 0.2588,
+      "step": 47905
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.862639880060063e-05,
+      "loss": 0.298,
+      "step": 47910
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8624010345128015e-05,
+      "loss": 0.2723,
+      "step": 47915
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8621621712756596e-05,
+      "loss": 0.2967,
+      "step": 47920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.86192329035174e-05,
+      "loss": 0.2764,
+      "step": 47925
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8616843917441447e-05,
+      "loss": 0.2625,
+      "step": 47930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.861445475455974e-05,
+      "loss": 0.3208,
+      "step": 47935
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.861206541490332e-05,
+      "loss": 0.2718,
+      "step": 47940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.86096758985032e-05,
+      "loss": 0.2794,
+      "step": 47945
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.86072862053904e-05,
+      "loss": 0.2741,
+      "step": 47950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.860489633559596e-05,
+      "loss": 0.2908,
+      "step": 47955
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.860250628915091e-05,
+      "loss": 0.2994,
+      "step": 47960
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.860011606608628e-05,
+      "loss": 0.3038,
+      "step": 47965
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.85977256664331e-05,
+      "loss": 0.2753,
+      "step": 47970
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8595335090222414e-05,
+      "loss": 0.286,
+      "step": 47975
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.859294433748527e-05,
+      "loss": 0.297,
+      "step": 47980
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.85905534082527e-05,
+      "loss": 0.2689,
+      "step": 47985
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8588162302555744e-05,
+      "loss": 0.2753,
+      "step": 47990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8585771020425454e-05,
+      "loss": 0.2754,
+      "step": 47995
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.858337956189289e-05,
+      "loss": 0.2888,
+      "step": 48000
+    },
+    {
+      "epoch": 1.69,
+      "eval_loss": 0.2754231095314026,
+      "eval_runtime": 10.5474,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 48000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.858098792698909e-05,
+      "loss": 0.2768,
+      "step": 48005
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8578596115745114e-05,
+      "loss": 0.2832,
+      "step": 48010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.857620412819201e-05,
+      "loss": 0.2764,
+      "step": 48015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8573811964360844e-05,
+      "loss": 0.2743,
+      "step": 48020
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8571419624282676e-05,
+      "loss": 0.2823,
+      "step": 48025
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.856902710798857e-05,
+      "loss": 0.2791,
+      "step": 48030
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8566634415509584e-05,
+      "loss": 0.2706,
+      "step": 48035
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.85642415468768e-05,
+      "loss": 0.2822,
+      "step": 48040
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.856184850212128e-05,
+      "loss": 0.2911,
+      "step": 48045
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.855945528127408e-05,
+      "loss": 0.26,
+      "step": 48050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8557061884366294e-05,
+      "loss": 0.2653,
+      "step": 48055
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8554668311429e-05,
+      "loss": 0.3004,
+      "step": 48060
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.855227456249327e-05,
+      "loss": 0.2668,
+      "step": 48065
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.854988063759019e-05,
+      "loss": 0.2933,
+      "step": 48070
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.854748653675083e-05,
+      "loss": 0.3116,
+      "step": 48075
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8545092260006287e-05,
+      "loss": 0.2593,
+      "step": 48080
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8542697807387656e-05,
+      "loss": 0.2967,
+      "step": 48085
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.854030317892601e-05,
+      "loss": 0.2844,
+      "step": 48090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8537908374652455e-05,
+      "loss": 0.3032,
+      "step": 48095
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.853551339459808e-05,
+      "loss": 0.2862,
+      "step": 48100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8533118238793985e-05,
+      "loss": 0.2812,
+      "step": 48105
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.853072290727127e-05,
+      "loss": 0.2666,
+      "step": 48110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.852832740006103e-05,
+      "loss": 0.2676,
+      "step": 48115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.852593171719437e-05,
+      "loss": 0.2995,
+      "step": 48120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.852353585870242e-05,
+      "loss": 0.297,
+      "step": 48125
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.852113982461625e-05,
+      "loss": 0.2822,
+      "step": 48130
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8518743614967e-05,
+      "loss": 0.2708,
+      "step": 48135
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8516347229785765e-05,
+      "loss": 0.2885,
+      "step": 48140
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.8513950669103676e-05,
+      "loss": 0.2671,
+      "step": 48145
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.851155393295184e-05,
+      "loss": 0.2828,
+      "step": 48150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.850915702136139e-05,
+      "loss": 0.2785,
+      "step": 48155
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.850675993436342e-05,
+      "loss": 0.266,
+      "step": 48160
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.850436267198909e-05,
+      "loss": 0.28,
+      "step": 48165
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.85019652342695e-05,
+      "loss": 0.2823,
+      "step": 48170
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 3.84995676212358e-05,
+      "loss": 0.299,
+      "step": 48175
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8497169832919105e-05,
+      "loss": 0.2796,
+      "step": 48180
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8494771869350555e-05,
+      "loss": 0.2853,
+      "step": 48185
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.849237373056128e-05,
+      "loss": 0.2868,
+      "step": 48190
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8489975416582436e-05,
+      "loss": 0.2774,
+      "step": 48195
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.848757692744515e-05,
+      "loss": 0.2843,
+      "step": 48200
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.848517826318055e-05,
+      "loss": 0.2581,
+      "step": 48205
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.848277942381981e-05,
+      "loss": 0.3146,
+      "step": 48210
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8480380409394066e-05,
+      "loss": 0.298,
+      "step": 48215
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8477981219934464e-05,
+      "loss": 0.2889,
+      "step": 48220
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8475581855472156e-05,
+      "loss": 0.2909,
+      "step": 48225
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8473182316038306e-05,
+      "loss": 0.3038,
+      "step": 48230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.847078260166406e-05,
+      "loss": 0.3052,
+      "step": 48235
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.846838271238058e-05,
+      "loss": 0.2841,
+      "step": 48240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.846598264821901e-05,
+      "loss": 0.2738,
+      "step": 48245
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8463582409210554e-05,
+      "loss": 0.3019,
+      "step": 48250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8461181995386335e-05,
+      "loss": 0.2845,
+      "step": 48255
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.845878140677756e-05,
+      "loss": 0.3173,
+      "step": 48260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.845638064341536e-05,
+      "loss": 0.269,
+      "step": 48265
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8453979705330926e-05,
+      "loss": 0.2784,
+      "step": 48270
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.845157859255543e-05,
+      "loss": 0.2945,
+      "step": 48275
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.844917730512005e-05,
+      "loss": 0.3041,
+      "step": 48280
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.844677584305598e-05,
+      "loss": 0.2842,
+      "step": 48285
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.844437420639439e-05,
+      "loss": 0.2791,
+      "step": 48290
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.844197239516645e-05,
+      "loss": 0.2651,
+      "step": 48295
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8439570409403355e-05,
+      "loss": 0.2748,
+      "step": 48300
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8437168249136294e-05,
+      "loss": 0.302,
+      "step": 48305
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8434765914396465e-05,
+      "loss": 0.259,
+      "step": 48310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.843236340521505e-05,
+      "loss": 0.2918,
+      "step": 48315
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8429960721623255e-05,
+      "loss": 0.2759,
+      "step": 48320
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.842755786365227e-05,
+      "loss": 0.3081,
+      "step": 48325
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8425154831333287e-05,
+      "loss": 0.2736,
+      "step": 48330
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8422751624697526e-05,
+      "loss": 0.2834,
+      "step": 48335
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.842034824377617e-05,
+      "loss": 0.2849,
+      "step": 48340
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8417944688600446e-05,
+      "loss": 0.2805,
+      "step": 48345
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8415540959201544e-05,
+      "loss": 0.308,
+      "step": 48350
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8413137055610696e-05,
+      "loss": 0.259,
+      "step": 48355
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8410732977859086e-05,
+      "loss": 0.2816,
+      "step": 48360
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8408328725977966e-05,
+      "loss": 0.2959,
+      "step": 48365
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.840592429999852e-05,
+      "loss": 0.2828,
+      "step": 48370
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8403519699951996e-05,
+      "loss": 0.3202,
+      "step": 48375
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.840111492586959e-05,
+      "loss": 0.2739,
+      "step": 48380
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.839870997778253e-05,
+      "loss": 0.2863,
+      "step": 48385
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.839630485572206e-05,
+      "loss": 0.2788,
+      "step": 48390
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.83938995597194e-05,
+      "loss": 0.2885,
+      "step": 48395
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.839149408980579e-05,
+      "loss": 0.2836,
+      "step": 48400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8389088446012444e-05,
+      "loss": 0.284,
+      "step": 48405
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.838668262837061e-05,
+      "loss": 0.3085,
+      "step": 48410
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.838427663691153e-05,
+      "loss": 0.2612,
+      "step": 48415
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.838187047166644e-05,
+      "loss": 0.2742,
+      "step": 48420
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.837946413266657e-05,
+      "loss": 0.3048,
+      "step": 48425
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.837705761994318e-05,
+      "loss": 0.2909,
+      "step": 48430
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8374650933527514e-05,
+      "loss": 0.2658,
+      "step": 48435
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.837224407345082e-05,
+      "loss": 0.2941,
+      "step": 48440
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.836983703974435e-05,
+      "loss": 0.3045,
+      "step": 48445
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.8367429832439356e-05,
+      "loss": 0.277,
+      "step": 48450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.836502245156709e-05,
+      "loss": 0.2902,
+      "step": 48455
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 3.836261489715882e-05,
+      "loss": 0.2796,
+      "step": 48460
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.83602071692458e-05,
+      "loss": 0.2701,
+      "step": 48465
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.83577992678593e-05,
+      "loss": 0.2802,
+      "step": 48470
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.835539119303058e-05,
+      "loss": 0.2777,
+      "step": 48475
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8352982944790905e-05,
+      "loss": 0.3027,
+      "step": 48480
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.835057452317154e-05,
+      "loss": 0.2844,
+      "step": 48485
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8348165928203766e-05,
+      "loss": 0.2777,
+      "step": 48490
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8345757159918855e-05,
+      "loss": 0.3131,
+      "step": 48495
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8343348218348086e-05,
+      "loss": 0.2874,
+      "step": 48500
+    },
+    {
+      "epoch": 1.71,
+      "eval_loss": 0.27594712376594543,
+      "eval_runtime": 10.5531,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 48500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.834093910352273e-05,
+      "loss": 0.2925,
+      "step": 48505
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.833852981547407e-05,
+      "loss": 0.2861,
+      "step": 48510
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8336120354233395e-05,
+      "loss": 0.2601,
+      "step": 48515
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8333710719831986e-05,
+      "loss": 0.2842,
+      "step": 48520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.833130091230113e-05,
+      "loss": 0.2735,
+      "step": 48525
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.832889093167211e-05,
+      "loss": 0.2881,
+      "step": 48530
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.832648077797624e-05,
+      "loss": 0.3089,
+      "step": 48535
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.832407045124479e-05,
+      "loss": 0.272,
+      "step": 48540
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.832165995150907e-05,
+      "loss": 0.2908,
+      "step": 48545
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.831924927880037e-05,
+      "loss": 0.2877,
+      "step": 48550
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.831683843315e-05,
+      "loss": 0.2693,
+      "step": 48555
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8314427414589255e-05,
+      "loss": 0.2799,
+      "step": 48560
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8312016223149446e-05,
+      "loss": 0.3004,
+      "step": 48565
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.830960485886188e-05,
+      "loss": 0.2897,
+      "step": 48570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8307193321757865e-05,
+      "loss": 0.2824,
+      "step": 48575
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.830478161186872e-05,
+      "loss": 0.3045,
+      "step": 48580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.830236972922575e-05,
+      "loss": 0.281,
+      "step": 48585
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.829995767386028e-05,
+      "loss": 0.2744,
+      "step": 48590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.829754544580362e-05,
+      "loss": 0.3124,
+      "step": 48595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.829513304508709e-05,
+      "loss": 0.287,
+      "step": 48600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8292720471742025e-05,
+      "loss": 0.2996,
+      "step": 48605
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8290307725799745e-05,
+      "loss": 0.2858,
+      "step": 48610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.828789480729158e-05,
+      "loss": 0.2869,
+      "step": 48615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.828548171624886e-05,
+      "loss": 0.2766,
+      "step": 48620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.828306845270291e-05,
+      "loss": 0.2747,
+      "step": 48625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.828065501668506e-05,
+      "loss": 0.2949,
+      "step": 48630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.827824140822667e-05,
+      "loss": 0.2672,
+      "step": 48635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.827582762735907e-05,
+      "loss": 0.2871,
+      "step": 48640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.827341367411359e-05,
+      "loss": 0.2998,
+      "step": 48645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8270999548521586e-05,
+      "loss": 0.2506,
+      "step": 48650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8268585250614397e-05,
+      "loss": 0.3298,
+      "step": 48655
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.826617078042337e-05,
+      "loss": 0.2921,
+      "step": 48660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8263756137979855e-05,
+      "loss": 0.2699,
+      "step": 48665
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.826134132331521e-05,
+      "loss": 0.2849,
+      "step": 48670
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.825892633646079e-05,
+      "loss": 0.2732,
+      "step": 48675
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.825651117744796e-05,
+      "loss": 0.2994,
+      "step": 48680
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8254095846308046e-05,
+      "loss": 0.2828,
+      "step": 48685
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.825168034307245e-05,
+      "loss": 0.2586,
+      "step": 48690
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.824926466777251e-05,
+      "loss": 0.2924,
+      "step": 48695
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.824684882043961e-05,
+      "loss": 0.2748,
+      "step": 48700
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.824443280110509e-05,
+      "loss": 0.2621,
+      "step": 48705
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8242016609800345e-05,
+      "loss": 0.2377,
+      "step": 48710
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.823960024655675e-05,
+      "loss": 0.2684,
+      "step": 48715
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.8237183711405666e-05,
+      "loss": 0.2967,
+      "step": 48720
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.823476700437847e-05,
+      "loss": 0.3064,
+      "step": 48725
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.823235012550655e-05,
+      "loss": 0.2617,
+      "step": 48730
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.822993307482129e-05,
+      "loss": 0.2875,
+      "step": 48735
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.822751585235406e-05,
+      "loss": 0.2954,
+      "step": 48740
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 3.822509845813626e-05,
+      "loss": 0.2939,
+      "step": 48745
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8222680892199266e-05,
+      "loss": 0.2767,
+      "step": 48750
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.822026315457448e-05,
+      "loss": 0.2901,
+      "step": 48755
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.821784524529328e-05,
+      "loss": 0.2757,
+      "step": 48760
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8215427164387086e-05,
+      "loss": 0.2656,
+      "step": 48765
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.821300891188727e-05,
+      "loss": 0.2772,
+      "step": 48770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.821059048782524e-05,
+      "loss": 0.3002,
+      "step": 48775
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8208171892232406e-05,
+      "loss": 0.2943,
+      "step": 48780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.820575312514015e-05,
+      "loss": 0.2894,
+      "step": 48785
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.820333418657991e-05,
+      "loss": 0.3017,
+      "step": 48790
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8200915076583064e-05,
+      "loss": 0.2886,
+      "step": 48795
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8198495795181036e-05,
+      "loss": 0.2696,
+      "step": 48800
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.819607634240525e-05,
+      "loss": 0.2835,
+      "step": 48805
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8193656718287097e-05,
+      "loss": 0.2748,
+      "step": 48810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.819123692285802e-05,
+      "loss": 0.2899,
+      "step": 48815
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.818881695614941e-05,
+      "loss": 0.2798,
+      "step": 48820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.818639681819271e-05,
+      "loss": 0.286,
+      "step": 48825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8183976509019334e-05,
+      "loss": 0.2705,
+      "step": 48830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.818155602866072e-05,
+      "loss": 0.2901,
+      "step": 48835
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.817913537714828e-05,
+      "loss": 0.285,
+      "step": 48840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.817671455451346e-05,
+      "loss": 0.2788,
+      "step": 48845
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8174293560787676e-05,
+      "loss": 0.281,
+      "step": 48850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.817187239600238e-05,
+      "loss": 0.2748,
+      "step": 48855
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8169451060188996e-05,
+      "loss": 0.2743,
+      "step": 48860
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8167029553378966e-05,
+      "loss": 0.2853,
+      "step": 48865
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8164607875603744e-05,
+      "loss": 0.2706,
+      "step": 48870
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.816218602689477e-05,
+      "loss": 0.2811,
+      "step": 48875
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.815976400728346e-05,
+      "loss": 0.2945,
+      "step": 48880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8157341816801305e-05,
+      "loss": 0.3165,
+      "step": 48885
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.815491945547974e-05,
+      "loss": 0.2618,
+      "step": 48890
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.815249692335021e-05,
+      "loss": 0.296,
+      "step": 48895
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8150074220444175e-05,
+      "loss": 0.288,
+      "step": 48900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.814765134679308e-05,
+      "loss": 0.29,
+      "step": 48905
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8145228302428416e-05,
+      "loss": 0.2876,
+      "step": 48910
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.814280508738162e-05,
+      "loss": 0.2828,
+      "step": 48915
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.814038170168417e-05,
+      "loss": 0.2716,
+      "step": 48920
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.81379581453675e-05,
+      "loss": 0.275,
+      "step": 48925
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.813553441846312e-05,
+      "loss": 0.2987,
+      "step": 48930
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.813311052100248e-05,
+      "loss": 0.2832,
+      "step": 48935
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.813068645301705e-05,
+      "loss": 0.2787,
+      "step": 48940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8128262214538316e-05,
+      "loss": 0.2945,
+      "step": 48945
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8125837805597743e-05,
+      "loss": 0.28,
+      "step": 48950
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8123413226226824e-05,
+      "loss": 0.2776,
+      "step": 48955
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.812098847645702e-05,
+      "loss": 0.2849,
+      "step": 48960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.811856355631984e-05,
+      "loss": 0.2916,
+      "step": 48965
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.811613846584675e-05,
+      "loss": 0.2675,
+      "step": 48970
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8113713205069254e-05,
+      "loss": 0.3024,
+      "step": 48975
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.811128777401882e-05,
+      "loss": 0.2783,
+      "step": 48980
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.810886217272697e-05,
+      "loss": 0.2888,
+      "step": 48985
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.810643640122518e-05,
+      "loss": 0.3066,
+      "step": 48990
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8104010459544946e-05,
+      "loss": 0.3046,
+      "step": 48995
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.810158434771777e-05,
+      "loss": 0.311,
+      "step": 49000
+    },
+    {
+      "epoch": 1.72,
+      "eval_loss": 0.2752542495727539,
+      "eval_runtime": 10.5525,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 49000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8099158065775166e-05,
+      "loss": 0.2867,
+      "step": 49005
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.8096731613748624e-05,
+      "loss": 0.2566,
+      "step": 49010
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.809430499166965e-05,
+      "loss": 0.2895,
+      "step": 49015
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.809187819956975e-05,
+      "loss": 0.2768,
+      "step": 49020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 3.808945123748045e-05,
+      "loss": 0.3111,
+      "step": 49025
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.808702410543324e-05,
+      "loss": 0.2982,
+      "step": 49030
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.808459680345966e-05,
+      "loss": 0.2851,
+      "step": 49035
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.808216933159121e-05,
+      "loss": 0.2602,
+      "step": 49040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8079741689859404e-05,
+      "loss": 0.2768,
+      "step": 49045
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.807731387829578e-05,
+      "loss": 0.2631,
+      "step": 49050
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.807488589693186e-05,
+      "loss": 0.2645,
+      "step": 49055
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.807245774579915e-05,
+      "loss": 0.2747,
+      "step": 49060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.80700294249292e-05,
+      "loss": 0.293,
+      "step": 49065
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.806760093435352e-05,
+      "loss": 0.2619,
+      "step": 49070
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.806517227410367e-05,
+      "loss": 0.2908,
+      "step": 49075
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.806274344421115e-05,
+      "loss": 0.2813,
+      "step": 49080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.806031444470753e-05,
+      "loss": 0.2782,
+      "step": 49085
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.805788527562433e-05,
+      "loss": 0.2899,
+      "step": 49090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.805545593699309e-05,
+      "loss": 0.2656,
+      "step": 49095
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.805302642884536e-05,
+      "loss": 0.2963,
+      "step": 49100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8050596751212684e-05,
+      "loss": 0.2814,
+      "step": 49105
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.804816690412661e-05,
+      "loss": 0.2822,
+      "step": 49110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.80457368876187e-05,
+      "loss": 0.2573,
+      "step": 49115
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.804330670172048e-05,
+      "loss": 0.2755,
+      "step": 49120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.804087634646352e-05,
+      "loss": 0.28,
+      "step": 49125
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.803844582187938e-05,
+      "loss": 0.2762,
+      "step": 49130
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8036015127999615e-05,
+      "loss": 0.2853,
+      "step": 49135
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.803358426485578e-05,
+      "loss": 0.2931,
+      "step": 49140
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.803115323247944e-05,
+      "loss": 0.2888,
+      "step": 49145
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8028722030902165e-05,
+      "loss": 0.2877,
+      "step": 49150
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.802629066015552e-05,
+      "loss": 0.2623,
+      "step": 49155
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8023859120271084e-05,
+      "loss": 0.2574,
+      "step": 49160
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.802142741128041e-05,
+      "loss": 0.2714,
+      "step": 49165
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.801899553321509e-05,
+      "loss": 0.2799,
+      "step": 49170
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.801656348610669e-05,
+      "loss": 0.2804,
+      "step": 49175
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.801413126998679e-05,
+      "loss": 0.2926,
+      "step": 49180
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8011698884886973e-05,
+      "loss": 0.2721,
+      "step": 49185
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.800926633083882e-05,
+      "loss": 0.2686,
+      "step": 49190
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.800683360787391e-05,
+      "loss": 0.2771,
+      "step": 49195
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.800440071602386e-05,
+      "loss": 0.2874,
+      "step": 49200
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.8001967655320216e-05,
+      "loss": 0.2632,
+      "step": 49205
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.799953442579459e-05,
+      "loss": 0.2797,
+      "step": 49210
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.799710102747859e-05,
+      "loss": 0.3042,
+      "step": 49215
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.799466746040379e-05,
+      "loss": 0.2858,
+      "step": 49220
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.79922337246018e-05,
+      "loss": 0.2786,
+      "step": 49225
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7989799820104215e-05,
+      "loss": 0.2892,
+      "step": 49230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.798736574694264e-05,
+      "loss": 0.2848,
+      "step": 49235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.798493150514868e-05,
+      "loss": 0.2859,
+      "step": 49240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7982497094753947e-05,
+      "loss": 0.3,
+      "step": 49245
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7980062515790036e-05,
+      "loss": 0.2706,
+      "step": 49250
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.797762776828858e-05,
+      "loss": 0.2738,
+      "step": 49255
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.797519285228117e-05,
+      "loss": 0.2571,
+      "step": 49260
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7972757767799424e-05,
+      "loss": 0.2982,
+      "step": 49265
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.797032251487498e-05,
+      "loss": 0.2826,
+      "step": 49270
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.796788709353945e-05,
+      "loss": 0.2615,
+      "step": 49275
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.796545150382445e-05,
+      "loss": 0.3103,
+      "step": 49280
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7963015745761597e-05,
+      "loss": 0.2841,
+      "step": 49285
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.796057981938253e-05,
+      "loss": 0.2699,
+      "step": 49290
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.795814372471888e-05,
+      "loss": 0.2865,
+      "step": 49295
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.795570746180227e-05,
+      "loss": 0.2647,
+      "step": 49300
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.795327103066434e-05,
+      "loss": 0.2766,
+      "step": 49305
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.7950834431336725e-05,
+      "loss": 0.2839,
+      "step": 49310
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.794839766385106e-05,
+      "loss": 0.3372,
+      "step": 49315
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.794596072823897e-05,
+      "loss": 0.2916,
+      "step": 49320
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.794352362453213e-05,
+      "loss": 0.2933,
+      "step": 49325
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7941086352762155e-05,
+      "loss": 0.3035,
+      "step": 49330
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.79386489129607e-05,
+      "loss": 0.2965,
+      "step": 49335
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.793621130515942e-05,
+      "loss": 0.2873,
+      "step": 49340
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.793377352938996e-05,
+      "loss": 0.2836,
+      "step": 49345
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.793133558568398e-05,
+      "loss": 0.2836,
+      "step": 49350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.792889747407312e-05,
+      "loss": 0.2956,
+      "step": 49355
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.792645919458906e-05,
+      "loss": 0.2759,
+      "step": 49360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7924020747263436e-05,
+      "loss": 0.2733,
+      "step": 49365
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.792158213212793e-05,
+      "loss": 0.2935,
+      "step": 49370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.791914334921418e-05,
+      "loss": 0.2866,
+      "step": 49375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7916704398553884e-05,
+      "loss": 0.2597,
+      "step": 49380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.791426528017869e-05,
+      "loss": 0.281,
+      "step": 49385
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7911825994120266e-05,
+      "loss": 0.2893,
+      "step": 49390
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.790938654041029e-05,
+      "loss": 0.2988,
+      "step": 49395
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.790694691908044e-05,
+      "loss": 0.2972,
+      "step": 49400
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.790450713016239e-05,
+      "loss": 0.2926,
+      "step": 49405
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.790206717368782e-05,
+      "loss": 0.3084,
+      "step": 49410
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7899627049688416e-05,
+      "loss": 0.2983,
+      "step": 49415
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.789718675819585e-05,
+      "loss": 0.2852,
+      "step": 49420
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.789474629924181e-05,
+      "loss": 0.278,
+      "step": 49425
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7892305672857986e-05,
+      "loss": 0.3003,
+      "step": 49430
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7889864879076074e-05,
+      "loss": 0.2931,
+      "step": 49435
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7887423917927764e-05,
+      "loss": 0.2539,
+      "step": 49440
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.788498278944475e-05,
+      "loss": 0.2906,
+      "step": 49445
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.788254149365872e-05,
+      "loss": 0.2911,
+      "step": 49450
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.788010003060137e-05,
+      "loss": 0.2989,
+      "step": 49455
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787765840030442e-05,
+      "loss": 0.2723,
+      "step": 49460
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7875216602799565e-05,
+      "loss": 0.3076,
+      "step": 49465
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.78727746381185e-05,
+      "loss": 0.2934,
+      "step": 49470
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.787033250629295e-05,
+      "loss": 0.2802,
+      "step": 49475
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7867890207354604e-05,
+      "loss": 0.2747,
+      "step": 49480
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786544774133519e-05,
+      "loss": 0.2802,
+      "step": 49485
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7863005108266417e-05,
+      "loss": 0.2939,
+      "step": 49490
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.786056230817999e-05,
+      "loss": 0.2919,
+      "step": 49495
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.785811934110764e-05,
+      "loss": 0.2557,
+      "step": 49500
+    },
+    {
+      "epoch": 1.74,
+      "eval_loss": 0.27525606751441956,
+      "eval_runtime": 10.5363,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 49500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7855676207081095e-05,
+      "loss": 0.2857,
+      "step": 49505
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7853232906132055e-05,
+      "loss": 0.2636,
+      "step": 49510
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.785078943829227e-05,
+      "loss": 0.2768,
+      "step": 49515
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7848345803593436e-05,
+      "loss": 0.3081,
+      "step": 49520
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.784590200206731e-05,
+      "loss": 0.3179,
+      "step": 49525
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7843458033745606e-05,
+      "loss": 0.2832,
+      "step": 49530
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.784101389866007e-05,
+      "loss": 0.3037,
+      "step": 49535
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.783856959684242e-05,
+      "loss": 0.2964,
+      "step": 49540
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.783612512832442e-05,
+      "loss": 0.2684,
+      "step": 49545
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7833680493137786e-05,
+      "loss": 0.2917,
+      "step": 49550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.783123569131427e-05,
+      "loss": 0.2968,
+      "step": 49555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.782879072288561e-05,
+      "loss": 0.3113,
+      "step": 49560
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.782634558788356e-05,
+      "loss": 0.291,
+      "step": 49565
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.782390028633986e-05,
+      "loss": 0.2844,
+      "step": 49570
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.782145481828628e-05,
+      "loss": 0.2514,
+      "step": 49575
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.781900918375455e-05,
+      "loss": 0.291,
+      "step": 49580
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.781656338277644e-05,
+      "loss": 0.3004,
+      "step": 49585
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.781411741538369e-05,
+      "loss": 0.2854,
+      "step": 49590
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.7811671281608073e-05,
+      "loss": 0.2908,
+      "step": 49595
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.780922498148136e-05,
+      "loss": 0.3035,
+      "step": 49600
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.780677851503529e-05,
+      "loss": 0.291,
+      "step": 49605
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.780433188230165e-05,
+      "loss": 0.2509,
+      "step": 49610
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.780188508331219e-05,
+      "loss": 0.2715,
+      "step": 49615
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.779943811809869e-05,
+      "loss": 0.2831,
+      "step": 49620
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.779699098669292e-05,
+      "loss": 0.2855,
+      "step": 49625
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.779454368912666e-05,
+      "loss": 0.2876,
+      "step": 49630
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.779209622543168e-05,
+      "loss": 0.2821,
+      "step": 49635
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.778964859563977e-05,
+      "loss": 0.2485,
+      "step": 49640
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.778720079978269e-05,
+      "loss": 0.2964,
+      "step": 49645
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7784752837892235e-05,
+      "loss": 0.2808,
+      "step": 49650
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.778230471000019e-05,
+      "loss": 0.2695,
+      "step": 49655
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.777985641613835e-05,
+      "loss": 0.2931,
+      "step": 49660
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7777407956338484e-05,
+      "loss": 0.2756,
+      "step": 49665
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.777495933063241e-05,
+      "loss": 0.2654,
+      "step": 49670
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7772510539051895e-05,
+      "loss": 0.269,
+      "step": 49675
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.777006158162876e-05,
+      "loss": 0.2727,
+      "step": 49680
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.776761245839478e-05,
+      "loss": 0.2853,
+      "step": 49685
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7765163169381776e-05,
+      "loss": 0.27,
+      "step": 49690
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7762713714621544e-05,
+      "loss": 0.301,
+      "step": 49695
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7760264094145876e-05,
+      "loss": 0.3008,
+      "step": 49700
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.775781430798659e-05,
+      "loss": 0.2869,
+      "step": 49705
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7755364356175493e-05,
+      "loss": 0.304,
+      "step": 49710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7752914238744395e-05,
+      "loss": 0.257,
+      "step": 49715
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.775046395572511e-05,
+      "loss": 0.2746,
+      "step": 49720
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.774801350714946e-05,
+      "loss": 0.2908,
+      "step": 49725
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.774556289304925e-05,
+      "loss": 0.2751,
+      "step": 49730
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.77431121134563e-05,
+      "loss": 0.2916,
+      "step": 49735
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7740661168402445e-05,
+      "loss": 0.2611,
+      "step": 49740
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.773821005791951e-05,
+      "loss": 0.3029,
+      "step": 49745
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.77357587820393e-05,
+      "loss": 0.2782,
+      "step": 49750
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7733307340793656e-05,
+      "loss": 0.2969,
+      "step": 49755
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7730855734214405e-05,
+      "loss": 0.2977,
+      "step": 49760
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7728403962333394e-05,
+      "loss": 0.3303,
+      "step": 49765
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.772595202518243e-05,
+      "loss": 0.2923,
+      "step": 49770
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.772349992279338e-05,
+      "loss": 0.2865,
+      "step": 49775
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.772104765519806e-05,
+      "loss": 0.2976,
+      "step": 49780
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.771859522242832e-05,
+      "loss": 0.2694,
+      "step": 49785
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.771614262451601e-05,
+      "loss": 0.3007,
+      "step": 49790
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.771368986149296e-05,
+      "loss": 0.2718,
+      "step": 49795
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.771123693339103e-05,
+      "loss": 0.2804,
+      "step": 49800
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.770878384024206e-05,
+      "loss": 0.2479,
+      "step": 49805
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7706330582077915e-05,
+      "loss": 0.2908,
+      "step": 49810
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.770387715893043e-05,
+      "loss": 0.2954,
+      "step": 49815
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.770142357083149e-05,
+      "loss": 0.3016,
+      "step": 49820
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.769896981781292e-05,
+      "loss": 0.3029,
+      "step": 49825
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.769651589990661e-05,
+      "loss": 0.2911,
+      "step": 49830
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.76940618171444e-05,
+      "loss": 0.2742,
+      "step": 49835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.769160756955816e-05,
+      "loss": 0.2924,
+      "step": 49840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.768915315717976e-05,
+      "loss": 0.2557,
+      "step": 49845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7686698580041075e-05,
+      "loss": 0.2791,
+      "step": 49850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7684243838173964e-05,
+      "loss": 0.2808,
+      "step": 49855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7681788931610315e-05,
+      "loss": 0.2843,
+      "step": 49860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7679333860381984e-05,
+      "loss": 0.2857,
+      "step": 49865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.767687862452086e-05,
+      "loss": 0.2519,
+      "step": 49870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7674423224058825e-05,
+      "loss": 0.2961,
+      "step": 49875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7671967659027755e-05,
+      "loss": 0.2871,
+      "step": 49880
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.766951192945954e-05,
+      "loss": 0.3087,
+      "step": 49885
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.766705603538606e-05,
+      "loss": 0.2804,
+      "step": 49890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.76645999768392e-05,
+      "loss": 0.2909,
+      "step": 49895
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7662143753850864e-05,
+      "loss": 0.2842,
+      "step": 49900
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7659687366452934e-05,
+      "loss": 0.2629,
+      "step": 49905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.76572308146773e-05,
+      "loss": 0.2873,
+      "step": 49910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7654774098555874e-05,
+      "loss": 0.289,
+      "step": 49915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.765231721812054e-05,
+      "loss": 0.2851,
+      "step": 49920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.764986017340321e-05,
+      "loss": 0.2801,
+      "step": 49925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.764740296443577e-05,
+      "loss": 0.3052,
+      "step": 49930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.764494559125016e-05,
+      "loss": 0.2867,
+      "step": 49935
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7642488053878246e-05,
+      "loss": 0.2873,
+      "step": 49940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7640030352351965e-05,
+      "loss": 0.3017,
+      "step": 49945
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.763757248670321e-05,
+      "loss": 0.2837,
+      "step": 49950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7635114456963914e-05,
+      "loss": 0.2532,
+      "step": 49955
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7632656263165984e-05,
+      "loss": 0.3039,
+      "step": 49960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7630197905341336e-05,
+      "loss": 0.2665,
+      "step": 49965
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.762773938352189e-05,
+      "loss": 0.2835,
+      "step": 49970
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7625280697739574e-05,
+      "loss": 0.2875,
+      "step": 49975
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.76228218480263e-05,
+      "loss": 0.2812,
+      "step": 49980
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.762036283441401e-05,
+      "loss": 0.2623,
+      "step": 49985
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.761790365693463e-05,
+      "loss": 0.2721,
+      "step": 49990
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.761544431562008e-05,
+      "loss": 0.2883,
+      "step": 49995
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.76129848105023e-05,
+      "loss": 0.2767,
+      "step": 50000
+    },
+    {
+      "epoch": 1.76,
+      "eval_loss": 0.2745961546897888,
+      "eval_runtime": 10.5442,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 50000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7610525141613226e-05,
+      "loss": 0.2897,
+      "step": 50005
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7608065308984786e-05,
+      "loss": 0.3061,
+      "step": 50010
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.760560531264894e-05,
+      "loss": 0.3162,
+      "step": 50015
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.760314515263761e-05,
+      "loss": 0.2929,
+      "step": 50020
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7600684828982747e-05,
+      "loss": 0.2934,
+      "step": 50025
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.759822434171629e-05,
+      "loss": 0.2888,
+      "step": 50030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.759576369087019e-05,
+      "loss": 0.3187,
+      "step": 50035
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.759330287647641e-05,
+      "loss": 0.2742,
+      "step": 50040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7590841898566884e-05,
+      "loss": 0.2864,
+      "step": 50045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7588380757173586e-05,
+      "loss": 0.2757,
+      "step": 50050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.758591945232845e-05,
+      "loss": 0.2854,
+      "step": 50055
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.758345798406344e-05,
+      "loss": 0.2772,
+      "step": 50060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.758099635241053e-05,
+      "loss": 0.3134,
+      "step": 50065
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.757853455740167e-05,
+      "loss": 0.2818,
+      "step": 50070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7576072599068834e-05,
+      "loss": 0.2701,
+      "step": 50075
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7573610477443974e-05,
+      "loss": 0.2831,
+      "step": 50080
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.757114819255908e-05,
+      "loss": 0.2732,
+      "step": 50085
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7568685744446094e-05,
+      "loss": 0.2824,
+      "step": 50090
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.756622313313702e-05,
+      "loss": 0.2878,
+      "step": 50095
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7563760358663814e-05,
+      "loss": 0.287,
+      "step": 50100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.756129742105846e-05,
+      "loss": 0.2736,
+      "step": 50105
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.755883432035294e-05,
+      "loss": 0.309,
+      "step": 50110
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.755637105657923e-05,
+      "loss": 0.2891,
+      "step": 50115
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.755390762976933e-05,
+      "loss": 0.2989,
+      "step": 50120
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7551444039955196e-05,
+      "loss": 0.29,
+      "step": 50125
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7548980287168836e-05,
+      "loss": 0.2877,
+      "step": 50130
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7546516371442244e-05,
+      "loss": 0.2558,
+      "step": 50135
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.75440522928074e-05,
+      "loss": 0.286,
+      "step": 50140
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.75415880512963e-05,
+      "loss": 0.2894,
+      "step": 50145
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.753912364694095e-05,
+      "loss": 0.2904,
+      "step": 50150
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.7536659079773336e-05,
+      "loss": 0.2695,
+      "step": 50155
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.753419434982548e-05,
+      "loss": 0.2806,
+      "step": 50160
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.753172945712936e-05,
+      "loss": 0.2857,
+      "step": 50165
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.752926440171699e-05,
+      "loss": 0.3019,
+      "step": 50170
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7526799183620384e-05,
+      "loss": 0.2934,
+      "step": 50175
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.752433380287155e-05,
+      "loss": 0.2921,
+      "step": 50180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.752186825950249e-05,
+      "loss": 0.2819,
+      "step": 50185
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7519402553545224e-05,
+      "loss": 0.2756,
+      "step": 50190
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.751693668503177e-05,
+      "loss": 0.2899,
+      "step": 50195
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7514470653994124e-05,
+      "loss": 0.3066,
+      "step": 50200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.751200446046434e-05,
+      "loss": 0.3111,
+      "step": 50205
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.750953810447443e-05,
+      "loss": 0.2624,
+      "step": 50210
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.750707158605641e-05,
+      "loss": 0.267,
+      "step": 50215
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7504604905242296e-05,
+      "loss": 0.2832,
+      "step": 50220
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.750213806206413e-05,
+      "loss": 0.271,
+      "step": 50225
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.749967105655394e-05,
+      "loss": 0.3105,
+      "step": 50230
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.749720388874377e-05,
+      "loss": 0.2642,
+      "step": 50235
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.749473655866563e-05,
+      "loss": 0.2783,
+      "step": 50240
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.749226906635159e-05,
+      "loss": 0.2739,
+      "step": 50245
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7489801411833644e-05,
+      "loss": 0.2714,
+      "step": 50250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7487333595143875e-05,
+      "loss": 0.2904,
+      "step": 50255
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.74848656163143e-05,
+      "loss": 0.293,
+      "step": 50260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.748239747537697e-05,
+      "loss": 0.2806,
+      "step": 50265
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.747992917236395e-05,
+      "loss": 0.2623,
+      "step": 50270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7477460707307265e-05,
+      "loss": 0.298,
+      "step": 50275
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.747499208023897e-05,
+      "loss": 0.2629,
+      "step": 50280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.747252329119113e-05,
+      "loss": 0.2498,
+      "step": 50285
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.74700543401958e-05,
+      "loss": 0.2924,
+      "step": 50290
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.746758522728502e-05,
+      "loss": 0.2857,
+      "step": 50295
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7465115952490876e-05,
+      "loss": 0.2857,
+      "step": 50300
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.746264651584541e-05,
+      "loss": 0.2751,
+      "step": 50305
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.746017691738069e-05,
+      "loss": 0.2849,
+      "step": 50310
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7457707157128784e-05,
+      "loss": 0.2778,
+      "step": 50315
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7455237235121765e-05,
+      "loss": 0.2708,
+      "step": 50320
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.74527671513917e-05,
+      "loss": 0.276,
+      "step": 50325
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.745029690597065e-05,
+      "loss": 0.314,
+      "step": 50330
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7447826498890705e-05,
+      "loss": 0.2613,
+      "step": 50335
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7445355930183945e-05,
+      "loss": 0.2961,
+      "step": 50340
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.744288519988243e-05,
+      "loss": 0.297,
+      "step": 50345
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.744041430801826e-05,
+      "loss": 0.2742,
+      "step": 50350
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.74379432546235e-05,
+      "loss": 0.2893,
+      "step": 50355
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.743547203973024e-05,
+      "loss": 0.2973,
+      "step": 50360
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7433000663370575e-05,
+      "loss": 0.2927,
+      "step": 50365
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.74305291255766e-05,
+      "loss": 0.2889,
+      "step": 50370
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.742805742638038e-05,
+      "loss": 0.2961,
+      "step": 50375
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.742558556581403e-05,
+      "loss": 0.3069,
+      "step": 50380
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.742311354390965e-05,
+      "loss": 0.2718,
+      "step": 50385
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7420641360699315e-05,
+      "loss": 0.2959,
+      "step": 50390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.741816901621515e-05,
+      "loss": 0.2958,
+      "step": 50395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.741569651048923e-05,
+      "loss": 0.2924,
+      "step": 50400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7413223843553694e-05,
+      "loss": 0.2873,
+      "step": 50405
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.74107510154406e-05,
+      "loss": 0.2992,
+      "step": 50410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7408278026182106e-05,
+      "loss": 0.2784,
+      "step": 50415
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7405804875810286e-05,
+      "loss": 0.2861,
+      "step": 50420
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.740333156435727e-05,
+      "loss": 0.2779,
+      "step": 50425
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.740085809185517e-05,
+      "loss": 0.2882,
+      "step": 50430
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.73983844583361e-05,
+      "loss": 0.2531,
+      "step": 50435
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7395910663832175e-05,
+      "loss": 0.2711,
+      "step": 50440
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.7393436708375516e-05,
+      "loss": 0.2866,
+      "step": 50445
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.739096259199825e-05,
+      "loss": 0.2892,
+      "step": 50450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.73884883147325e-05,
+      "loss": 0.2926,
+      "step": 50455
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.73860138766104e-05,
+      "loss": 0.3057,
+      "step": 50460
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7383539277664074e-05,
+      "loss": 0.2761,
+      "step": 50465
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.738106451792564e-05,
+      "loss": 0.2693,
+      "step": 50470
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.737858959742726e-05,
+      "loss": 0.2797,
+      "step": 50475
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7376114516201036e-05,
+      "loss": 0.2759,
+      "step": 50480
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.737363927427913e-05,
+      "loss": 0.2778,
+      "step": 50485
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.737116387169367e-05,
+      "loss": 0.3123,
+      "step": 50490
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7368688308476796e-05,
+      "loss": 0.2828,
+      "step": 50495
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.736621258466066e-05,
+      "loss": 0.2872,
+      "step": 50500
+    },
+    {
+      "epoch": 1.78,
+      "eval_loss": 0.27529093623161316,
+      "eval_runtime": 10.5518,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 50500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.73637367002774e-05,
+      "loss": 0.27,
+      "step": 50505
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.736126065535918e-05,
+      "loss": 0.2908,
+      "step": 50510
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.735878444993812e-05,
+      "loss": 0.3056,
+      "step": 50515
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.73563080840464e-05,
+      "loss": 0.2857,
+      "step": 50520
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.735383155771616e-05,
+      "loss": 0.3068,
+      "step": 50525
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.735135487097955e-05,
+      "loss": 0.2963,
+      "step": 50530
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.734887802386874e-05,
+      "loss": 0.3042,
+      "step": 50535
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.734640101641589e-05,
+      "loss": 0.2815,
+      "step": 50540
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.734392384865316e-05,
+      "loss": 0.2768,
+      "step": 50545
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.734144652061271e-05,
+      "loss": 0.2734,
+      "step": 50550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.733896903232671e-05,
+      "loss": 0.2725,
+      "step": 50555
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.733649138382734e-05,
+      "loss": 0.3065,
+      "step": 50560
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7334013575146756e-05,
+      "loss": 0.2916,
+      "step": 50565
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.733153560631712e-05,
+      "loss": 0.2769,
+      "step": 50570
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7329057477370636e-05,
+      "loss": 0.2781,
+      "step": 50575
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.732657918833946e-05,
+      "loss": 0.2919,
+      "step": 50580
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7324100739255774e-05,
+      "loss": 0.2488,
+      "step": 50585
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7321622130151775e-05,
+      "loss": 0.2713,
+      "step": 50590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7319143361059616e-05,
+      "loss": 0.2853,
+      "step": 50595
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.731666443201151e-05,
+      "loss": 0.2938,
+      "step": 50600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.731418534303963e-05,
+      "loss": 0.2751,
+      "step": 50605
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7311706094176166e-05,
+      "loss": 0.2899,
+      "step": 50610
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.730922668545331e-05,
+      "loss": 0.2622,
+      "step": 50615
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.730674711690326e-05,
+      "loss": 0.2915,
+      "step": 50620
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7304267388558215e-05,
+      "loss": 0.2604,
+      "step": 50625
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.730178750045036e-05,
+      "loss": 0.2886,
+      "step": 50630
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.729930745261191e-05,
+      "loss": 0.275,
+      "step": 50635
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.729682724507505e-05,
+      "loss": 0.3038,
+      "step": 50640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7294346877872e-05,
+      "loss": 0.2918,
+      "step": 50645
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.729186635103495e-05,
+      "loss": 0.2943,
+      "step": 50650
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.728938566459612e-05,
+      "loss": 0.2683,
+      "step": 50655
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.728690481858771e-05,
+      "loss": 0.2774,
+      "step": 50660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.728442381304194e-05,
+      "loss": 0.2812,
+      "step": 50665
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.728194264799103e-05,
+      "loss": 0.2777,
+      "step": 50670
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.727946132346719e-05,
+      "loss": 0.3065,
+      "step": 50675
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7276979839502626e-05,
+      "loss": 0.2789,
+      "step": 50680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.727449819612958e-05,
+      "loss": 0.2794,
+      "step": 50685
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7272016393380256e-05,
+      "loss": 0.2922,
+      "step": 50690
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.726953443128689e-05,
+      "loss": 0.2913,
+      "step": 50695
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.726705230988171e-05,
+      "loss": 0.2764,
+      "step": 50700
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7264570029196935e-05,
+      "loss": 0.3015,
+      "step": 50705
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.72620875892648e-05,
+      "loss": 0.2876,
+      "step": 50710
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7259604990117536e-05,
+      "loss": 0.2714,
+      "step": 50715
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.725712223178738e-05,
+      "loss": 0.2936,
+      "step": 50720
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.7254639314306585e-05,
+      "loss": 0.2801,
+      "step": 50725
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.725215623770736e-05,
+      "loss": 0.2894,
+      "step": 50730
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.724967300202196e-05,
+      "loss": 0.2929,
+      "step": 50735
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.724718960728264e-05,
+      "loss": 0.2622,
+      "step": 50740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.724470605352163e-05,
+      "loss": 0.2968,
+      "step": 50745
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7242222340771174e-05,
+      "loss": 0.283,
+      "step": 50750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7239738469063535e-05,
+      "loss": 0.2829,
+      "step": 50755
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.723725443843096e-05,
+      "loss": 0.3006,
+      "step": 50760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.723477024890569e-05,
+      "loss": 0.2904,
+      "step": 50765
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7232285900520006e-05,
+      "loss": 0.2768,
+      "step": 50770
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.722980139330614e-05,
+      "loss": 0.2764,
+      "step": 50775
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.722731672729637e-05,
+      "loss": 0.2917,
+      "step": 50780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7224831902522944e-05,
+      "loss": 0.2634,
+      "step": 50785
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.722234691901814e-05,
+      "loss": 0.2761,
+      "step": 50790
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.721986177681421e-05,
+      "loss": 0.2781,
+      "step": 50795
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.721737647594343e-05,
+      "loss": 0.2898,
+      "step": 50800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.721489101643806e-05,
+      "loss": 0.2843,
+      "step": 50805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.72124053983304e-05,
+      "loss": 0.2757,
+      "step": 50810
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.720991962165269e-05,
+      "loss": 0.2847,
+      "step": 50815
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7207433686437224e-05,
+      "loss": 0.2749,
+      "step": 50820
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7204947592716274e-05,
+      "loss": 0.2849,
+      "step": 50825
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.720295860363726e-05,
+      "loss": 0.2891,
+      "step": 50830
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.720047222468779e-05,
+      "loss": 0.2837,
+      "step": 50835
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.719798568732323e-05,
+      "loss": 0.2903,
+      "step": 50840
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.719549899157586e-05,
+      "loss": 0.3028,
+      "step": 50845
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.719301213747798e-05,
+      "loss": 0.2797,
+      "step": 50850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.719052512506187e-05,
+      "loss": 0.2833,
+      "step": 50855
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7188037954359826e-05,
+      "loss": 0.2882,
+      "step": 50860
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.718555062540414e-05,
+      "loss": 0.2785,
+      "step": 50865
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7183063138227115e-05,
+      "loss": 0.2619,
+      "step": 50870
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7180575492861034e-05,
+      "loss": 0.2815,
+      "step": 50875
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.717808768933823e-05,
+      "loss": 0.2977,
+      "step": 50880
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.717559972769096e-05,
+      "loss": 0.2808,
+      "step": 50885
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.717311160795156e-05,
+      "loss": 0.2925,
+      "step": 50890
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7170623330152324e-05,
+      "loss": 0.2829,
+      "step": 50895
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.716813489432557e-05,
+      "loss": 0.2828,
+      "step": 50900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.71656463005036e-05,
+      "loss": 0.2682,
+      "step": 50905
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.716315754871873e-05,
+      "loss": 0.3047,
+      "step": 50910
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.716066863900327e-05,
+      "loss": 0.2797,
+      "step": 50915
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.715817957138954e-05,
+      "loss": 0.2997,
+      "step": 50920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.715569034590988e-05,
+      "loss": 0.2846,
+      "step": 50925
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.715320096259658e-05,
+      "loss": 0.2702,
+      "step": 50930
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7150711421481975e-05,
+      "loss": 0.3049,
+      "step": 50935
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.714822172259839e-05,
+      "loss": 0.2774,
+      "step": 50940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.714573186597815e-05,
+      "loss": 0.276,
+      "step": 50945
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7143241851653586e-05,
+      "loss": 0.2749,
+      "step": 50950
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.714075167965703e-05,
+      "loss": 0.3144,
+      "step": 50955
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.713826135002082e-05,
+      "loss": 0.2635,
+      "step": 50960
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7135770862777286e-05,
+      "loss": 0.2974,
+      "step": 50965
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.713328021795875e-05,
+      "loss": 0.2894,
+      "step": 50970
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7130789415597587e-05,
+      "loss": 0.2923,
+      "step": 50975
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7128298455726104e-05,
+      "loss": 0.3008,
+      "step": 50980
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.712580733837666e-05,
+      "loss": 0.2606,
+      "step": 50985
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.71233160635816e-05,
+      "loss": 0.3075,
+      "step": 50990
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7120824631373274e-05,
+      "loss": 0.2828,
+      "step": 50995
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7118333041784024e-05,
+      "loss": 0.2842,
+      "step": 51000
+    },
+    {
+      "epoch": 1.79,
+      "eval_loss": 0.2750702500343323,
+      "eval_runtime": 10.538,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 51000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.711584129484621e-05,
+      "loss": 0.2738,
+      "step": 51005
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.711334939059218e-05,
+      "loss": 0.2937,
+      "step": 51010
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.7110857329054295e-05,
+      "loss": 0.29,
+      "step": 51015
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7108365110264895e-05,
+      "loss": 0.2818,
+      "step": 51020
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.710587273425636e-05,
+      "loss": 0.2837,
+      "step": 51025
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.710338020106106e-05,
+      "loss": 0.2928,
+      "step": 51030
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7100887510711326e-05,
+      "loss": 0.3059,
+      "step": 51035
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.709839466323956e-05,
+      "loss": 0.288,
+      "step": 51040
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.709590165867809e-05,
+      "loss": 0.2917,
+      "step": 51045
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7093408497059325e-05,
+      "loss": 0.2823,
+      "step": 51050
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.709091517841561e-05,
+      "loss": 0.2821,
+      "step": 51055
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.708842170277934e-05,
+      "loss": 0.29,
+      "step": 51060
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.708592807018287e-05,
+      "loss": 0.2864,
+      "step": 51065
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7083434280658605e-05,
+      "loss": 0.3002,
+      "step": 51070
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.70809403342389e-05,
+      "loss": 0.2825,
+      "step": 51075
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7078446230956143e-05,
+      "loss": 0.2741,
+      "step": 51080
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.707595197084272e-05,
+      "loss": 0.2888,
+      "step": 51085
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.707345755393103e-05,
+      "loss": 0.3029,
+      "step": 51090
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.707096298025344e-05,
+      "loss": 0.2536,
+      "step": 51095
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.706846824984236e-05,
+      "loss": 0.2956,
+      "step": 51100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7065973362730164e-05,
+      "loss": 0.2647,
+      "step": 51105
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7063478318949255e-05,
+      "loss": 0.3026,
+      "step": 51110
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.706098311853204e-05,
+      "loss": 0.2727,
+      "step": 51115
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.70584877615109e-05,
+      "loss": 0.2703,
+      "step": 51120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7055992247918247e-05,
+      "loss": 0.2926,
+      "step": 51125
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7053496577786485e-05,
+      "loss": 0.3351,
+      "step": 51130
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7051000751148e-05,
+      "loss": 0.2617,
+      "step": 51135
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.704850476803522e-05,
+      "loss": 0.2807,
+      "step": 51140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.704600862848054e-05,
+      "loss": 0.2904,
+      "step": 51145
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.704351233251638e-05,
+      "loss": 0.3106,
+      "step": 51150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.704101588017516e-05,
+      "loss": 0.2817,
+      "step": 51155
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.703851927148927e-05,
+      "loss": 0.2982,
+      "step": 51160
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7036022506491146e-05,
+      "loss": 0.2926,
+      "step": 51165
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7033525585213205e-05,
+      "loss": 0.2752,
+      "step": 51170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.703102850768786e-05,
+      "loss": 0.2834,
+      "step": 51175
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7028531273947536e-05,
+      "loss": 0.3048,
+      "step": 51180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.702603388402467e-05,
+      "loss": 0.2871,
+      "step": 51185
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.702353633795167e-05,
+      "loss": 0.28,
+      "step": 51190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.702103863576098e-05,
+      "loss": 0.2954,
+      "step": 51195
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7018540777485024e-05,
+      "loss": 0.2968,
+      "step": 51200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7016042763156236e-05,
+      "loss": 0.2944,
+      "step": 51205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7013544592807054e-05,
+      "loss": 0.2831,
+      "step": 51210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7011046266469905e-05,
+      "loss": 0.2764,
+      "step": 51215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7008547784177246e-05,
+      "loss": 0.2949,
+      "step": 51220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.70060491459615e-05,
+      "loss": 0.2917,
+      "step": 51225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.7003550351855124e-05,
+      "loss": 0.2698,
+      "step": 51230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.700105140189056e-05,
+      "loss": 0.2764,
+      "step": 51235
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.699855229610024e-05,
+      "loss": 0.3163,
+      "step": 51240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6996053034516634e-05,
+      "loss": 0.2672,
+      "step": 51245
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.699355361717218e-05,
+      "loss": 0.2975,
+      "step": 51250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.699105404409935e-05,
+      "loss": 0.3058,
+      "step": 51255
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.698855431533057e-05,
+      "loss": 0.2785,
+      "step": 51260
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.698605443089833e-05,
+      "loss": 0.2886,
+      "step": 51265
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.698355439083506e-05,
+      "loss": 0.2679,
+      "step": 51270
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.698105419517323e-05,
+      "loss": 0.2909,
+      "step": 51275
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6978553843945306e-05,
+      "loss": 0.2745,
+      "step": 51280
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.697605333718376e-05,
+      "loss": 0.2649,
+      "step": 51285
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.6973552674921067e-05,
+      "loss": 0.2923,
+      "step": 51290
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.697105185718967e-05,
+      "loss": 0.2965,
+      "step": 51295
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.696855088402204e-05,
+      "loss": 0.2692,
+      "step": 51300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.696604975545068e-05,
+      "loss": 0.2756,
+      "step": 51305
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.696354847150806e-05,
+      "loss": 0.3012,
+      "step": 51310
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.696104703222663e-05,
+      "loss": 0.267,
+      "step": 51315
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6958545437638896e-05,
+      "loss": 0.2901,
+      "step": 51320
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.695604368777733e-05,
+      "loss": 0.2705,
+      "step": 51325
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.695354178267441e-05,
+      "loss": 0.2869,
+      "step": 51330
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.695103972236263e-05,
+      "loss": 0.2809,
+      "step": 51335
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.694853750687448e-05,
+      "loss": 0.2787,
+      "step": 51340
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.694603513624244e-05,
+      "loss": 0.2472,
+      "step": 51345
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6943532610499014e-05,
+      "loss": 0.2754,
+      "step": 51350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6941029929676675e-05,
+      "loss": 0.2844,
+      "step": 51355
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.693852709380794e-05,
+      "loss": 0.2816,
+      "step": 51360
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.69360241029253e-05,
+      "loss": 0.2789,
+      "step": 51365
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.693352095706125e-05,
+      "loss": 0.3022,
+      "step": 51370
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.69310176562483e-05,
+      "loss": 0.2998,
+      "step": 51375
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.692851420051894e-05,
+      "loss": 0.2976,
+      "step": 51380
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6926010589905694e-05,
+      "loss": 0.2608,
+      "step": 51385
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.692350682444104e-05,
+      "loss": 0.2994,
+      "step": 51390
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6921002904157533e-05,
+      "loss": 0.28,
+      "step": 51395
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.691849882908764e-05,
+      "loss": 0.2896,
+      "step": 51400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.691599459926391e-05,
+      "loss": 0.2934,
+      "step": 51405
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.691349021471884e-05,
+      "loss": 0.286,
+      "step": 51410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.691098567548494e-05,
+      "loss": 0.2833,
+      "step": 51415
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.690848098159475e-05,
+      "loss": 0.283,
+      "step": 51420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6905976133080785e-05,
+      "loss": 0.266,
+      "step": 51425
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6903471129975556e-05,
+      "loss": 0.2611,
+      "step": 51430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.69009659723116e-05,
+      "loss": 0.2934,
+      "step": 51435
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.689846066012145e-05,
+      "loss": 0.2724,
+      "step": 51440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.689595519343763e-05,
+      "loss": 0.2909,
+      "step": 51445
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.689344957229267e-05,
+      "loss": 0.2729,
+      "step": 51450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.689094379671911e-05,
+      "loss": 0.2945,
+      "step": 51455
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.688843786674947e-05,
+      "loss": 0.2714,
+      "step": 51460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6885931782416304e-05,
+      "loss": 0.2573,
+      "step": 51465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6883425543752145e-05,
+      "loss": 0.2973,
+      "step": 51470
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.688091915078955e-05,
+      "loss": 0.2655,
+      "step": 51475
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.687841260356103e-05,
+      "loss": 0.3057,
+      "step": 51480
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6875905902099166e-05,
+      "loss": 0.2944,
+      "step": 51485
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.687339904643648e-05,
+      "loss": 0.2956,
+      "step": 51490
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.687089203660554e-05,
+      "loss": 0.2672,
+      "step": 51495
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6868384872638886e-05,
+      "loss": 0.3063,
+      "step": 51500
+    },
+    {
+      "epoch": 1.81,
+      "eval_loss": 0.274717777967453,
+      "eval_runtime": 10.5391,
+      "eval_samples_per_second": 9.488,
+      "eval_steps_per_second": 9.488,
+      "step": 51500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.686587755456907e-05,
+      "loss": 0.2998,
+      "step": 51505
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6863370082428674e-05,
+      "loss": 0.2863,
+      "step": 51510
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.686086245625022e-05,
+      "loss": 0.2824,
+      "step": 51515
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.685835467606628e-05,
+      "loss": 0.3065,
+      "step": 51520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.685584674190942e-05,
+      "loss": 0.287,
+      "step": 51525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.685333865381222e-05,
+      "loss": 0.2882,
+      "step": 51530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.685133207251927e-05,
+      "loss": 0.2919,
+      "step": 51535
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.684882370741149e-05,
+      "loss": 0.292,
+      "step": 51540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.684631518845454e-05,
+      "loss": 0.294,
+      "step": 51545
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.684380651568099e-05,
+      "loss": 0.2997,
+      "step": 51550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6841297689123415e-05,
+      "loss": 0.2699,
+      "step": 51555
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.68387887088144e-05,
+      "loss": 0.3079,
+      "step": 51560
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6836279574786516e-05,
+      "loss": 0.2889,
+      "step": 51565
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.683377028707233e-05,
+      "loss": 0.289,
+      "step": 51570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6831260845704445e-05,
+      "loss": 0.3018,
+      "step": 51575
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.682875125071544e-05,
+      "loss": 0.2832,
+      "step": 51580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.6826241502137884e-05,
+      "loss": 0.2844,
+      "step": 51585
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.682373160000438e-05,
+      "loss": 0.2827,
+      "step": 51590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6821221544347516e-05,
+      "loss": 0.291,
+      "step": 51595
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.681871133519988e-05,
+      "loss": 0.2522,
+      "step": 51600
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.681620097259407e-05,
+      "loss": 0.2753,
+      "step": 51605
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6813690456562674e-05,
+      "loss": 0.2762,
+      "step": 51610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.681117978713829e-05,
+      "loss": 0.2705,
+      "step": 51615
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.680866896435353e-05,
+      "loss": 0.2775,
+      "step": 51620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.680615798824097e-05,
+      "loss": 0.3353,
+      "step": 51625
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.680364685883325e-05,
+      "loss": 0.2734,
+      "step": 51630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6801135576162935e-05,
+      "loss": 0.277,
+      "step": 51635
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.679862414026266e-05,
+      "loss": 0.2991,
+      "step": 51640
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.679611255116502e-05,
+      "loss": 0.3159,
+      "step": 51645
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.679360080890264e-05,
+      "loss": 0.2819,
+      "step": 51650
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.679108891350813e-05,
+      "loss": 0.2818,
+      "step": 51655
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6788576865014084e-05,
+      "loss": 0.264,
+      "step": 51660
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6786064663453143e-05,
+      "loss": 0.2892,
+      "step": 51665
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.678355230885793e-05,
+      "loss": 0.2763,
+      "step": 51670
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6781039801261054e-05,
+      "loss": 0.2865,
+      "step": 51675
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6778527140695135e-05,
+      "loss": 0.2993,
+      "step": 51680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.67760143271928e-05,
+      "loss": 0.266,
+      "step": 51685
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.677350136078668e-05,
+      "loss": 0.2778,
+      "step": 51690
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6770988241509406e-05,
+      "loss": 0.2742,
+      "step": 51695
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6768474969393607e-05,
+      "loss": 0.2973,
+      "step": 51700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6765961544471925e-05,
+      "loss": 0.3009,
+      "step": 51705
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6763447966776975e-05,
+      "loss": 0.2739,
+      "step": 51710
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.676093423634142e-05,
+      "loss": 0.2892,
+      "step": 51715
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6758420353197863e-05,
+      "loss": 0.3095,
+      "step": 51720
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.675590631737898e-05,
+      "loss": 0.2716,
+      "step": 51725
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6753392128917396e-05,
+      "loss": 0.2891,
+      "step": 51730
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.675087778784577e-05,
+      "loss": 0.2994,
+      "step": 51735
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.674836329419673e-05,
+      "loss": 0.2867,
+      "step": 51740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.674584864800294e-05,
+      "loss": 0.2896,
+      "step": 51745
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.674333384929705e-05,
+      "loss": 0.301,
+      "step": 51750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.67408188981117e-05,
+      "loss": 0.2757,
+      "step": 51755
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6738303794479556e-05,
+      "loss": 0.2815,
+      "step": 51760
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.673578853843328e-05,
+      "loss": 0.3074,
+      "step": 51765
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.673327313000552e-05,
+      "loss": 0.2867,
+      "step": 51770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.673075756922894e-05,
+      "loss": 0.2663,
+      "step": 51775
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6728241856136204e-05,
+      "loss": 0.2919,
+      "step": 51780
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.672572599075998e-05,
+      "loss": 0.3072,
+      "step": 51785
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.672320997313292e-05,
+      "loss": 0.2797,
+      "step": 51790
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.672069380328772e-05,
+      "loss": 0.2797,
+      "step": 51795
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.671817748125702e-05,
+      "loss": 0.3024,
+      "step": 51800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.671566100707352e-05,
+      "loss": 0.2841,
+      "step": 51805
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6713144380769885e-05,
+      "loss": 0.3027,
+      "step": 51810
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.671062760237878e-05,
+      "loss": 0.2932,
+      "step": 51815
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6708110671932896e-05,
+      "loss": 0.2705,
+      "step": 51820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.670559358946491e-05,
+      "loss": 0.2865,
+      "step": 51825
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6703076355007506e-05,
+      "loss": 0.2862,
+      "step": 51830
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.670055896859337e-05,
+      "loss": 0.3049,
+      "step": 51835
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.669804143025519e-05,
+      "loss": 0.2863,
+      "step": 51840
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.669552374002564e-05,
+      "loss": 0.2776,
+      "step": 51845
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6693005897937435e-05,
+      "loss": 0.2851,
+      "step": 51850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.669048790402324e-05,
+      "loss": 0.2902,
+      "step": 51855
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6687969758315774e-05,
+      "loss": 0.2835,
+      "step": 51860
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.668545146084772e-05,
+      "loss": 0.2973,
+      "step": 51865
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.668293301165179e-05,
+      "loss": 0.2554,
+      "step": 51870
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.668041441076066e-05,
+      "loss": 0.2768,
+      "step": 51875
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.667789565820705e-05,
+      "loss": 0.2787,
+      "step": 51880
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.667537675402366e-05,
+      "loss": 0.2967,
+      "step": 51885
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.667285769824321e-05,
+      "loss": 0.2772,
+      "step": 51890
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.667033849089838e-05,
+      "loss": 0.3012,
+      "step": 51895
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.666781913202191e-05,
+      "loss": 0.2883,
+      "step": 51900
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.666529962164648e-05,
+      "loss": 0.2899,
+      "step": 51905
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6662779959804833e-05,
+      "loss": 0.2894,
+      "step": 51910
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6660260146529676e-05,
+      "loss": 0.2798,
+      "step": 51915
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.665774018185372e-05,
+      "loss": 0.2764,
+      "step": 51920
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.665522006580969e-05,
+      "loss": 0.2734,
+      "step": 51925
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.665269979843031e-05,
+      "loss": 0.3093,
+      "step": 51930
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.665017937974831e-05,
+      "loss": 0.2899,
+      "step": 51935
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.664765880979639e-05,
+      "loss": 0.2641,
+      "step": 51940
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.664513808860732e-05,
+      "loss": 0.2931,
+      "step": 51945
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.664261721621379e-05,
+      "loss": 0.2899,
+      "step": 51950
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6640096192648554e-05,
+      "loss": 0.3063,
+      "step": 51955
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.663757501794434e-05,
+      "loss": 0.2803,
+      "step": 51960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.663505369213388e-05,
+      "loss": 0.2783,
+      "step": 51965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.663253221524992e-05,
+      "loss": 0.2873,
+      "step": 51970
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.66300105873252e-05,
+      "loss": 0.306,
+      "step": 51975
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6627488808392444e-05,
+      "loss": 0.2775,
+      "step": 51980
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.662496687848442e-05,
+      "loss": 0.2764,
+      "step": 51985
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.662244479763386e-05,
+      "loss": 0.3088,
+      "step": 51990
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.661992256587351e-05,
+      "loss": 0.2778,
+      "step": 51995
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.661740018323612e-05,
+      "loss": 0.3253,
+      "step": 52000
+    },
+    {
+      "epoch": 1.83,
+      "eval_loss": 0.2752087414264679,
+      "eval_runtime": 10.5359,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 52000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.661487764975445e-05,
+      "loss": 0.2783,
+      "step": 52005
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.661235496546125e-05,
+      "loss": 0.2796,
+      "step": 52010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6609832130389274e-05,
+      "loss": 0.2624,
+      "step": 52015
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6607309144571274e-05,
+      "loss": 0.307,
+      "step": 52020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6604786008040024e-05,
+      "loss": 0.2823,
+      "step": 52025
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.660226272082826e-05,
+      "loss": 0.2917,
+      "step": 52030
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.659973928296878e-05,
+      "loss": 0.2749,
+      "step": 52035
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.659721569449431e-05,
+      "loss": 0.2948,
+      "step": 52040
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6594691955437646e-05,
+      "loss": 0.2705,
+      "step": 52045
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.659216806583155e-05,
+      "loss": 0.2612,
+      "step": 52050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6589644025708785e-05,
+      "loss": 0.2771,
+      "step": 52055
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6587119835102135e-05,
+      "loss": 0.2768,
+      "step": 52060
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.658459549404437e-05,
+      "loss": 0.2863,
+      "step": 52065
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.658207100256827e-05,
+      "loss": 0.2957,
+      "step": 52070
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6579546360706606e-05,
+      "loss": 0.2587,
+      "step": 52075
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.657702156849216e-05,
+      "loss": 0.2844,
+      "step": 52080
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.657449662595773e-05,
+      "loss": 0.2673,
+      "step": 52085
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6571971533136085e-05,
+      "loss": 0.2779,
+      "step": 52090
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.656944629006001e-05,
+      "loss": 0.2795,
+      "step": 52095
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6566920896762304e-05,
+      "loss": 0.2798,
+      "step": 52100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.656439535327575e-05,
+      "loss": 0.281,
+      "step": 52105
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.656186965963315e-05,
+      "loss": 0.2881,
+      "step": 52110
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.655934381586728e-05,
+      "loss": 0.2843,
+      "step": 52115
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6556817822010966e-05,
+      "loss": 0.2842,
+      "step": 52120
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6554291678096975e-05,
+      "loss": 0.2903,
+      "step": 52125
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.655176538415813e-05,
+      "loss": 0.2705,
+      "step": 52130
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6549238940227215e-05,
+      "loss": 0.2743,
+      "step": 52135
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6546712346337054e-05,
+      "loss": 0.283,
+      "step": 52140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.654418560252043e-05,
+      "loss": 0.3023,
+      "step": 52145
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6541658708810175e-05,
+      "loss": 0.2607,
+      "step": 52150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.6539131665239086e-05,
+      "loss": 0.2968,
+      "step": 52155
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6536604471839964e-05,
+      "loss": 0.2697,
+      "step": 52160
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6534077128645646e-05,
+      "loss": 0.2761,
+      "step": 52165
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6531549635688945e-05,
+      "loss": 0.2768,
+      "step": 52170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.652902199300266e-05,
+      "loss": 0.3106,
+      "step": 52175
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.652649420061963e-05,
+      "loss": 0.2749,
+      "step": 52180
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.652396625857267e-05,
+      "loss": 0.2913,
+      "step": 52185
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.65214381668946e-05,
+      "loss": 0.2954,
+      "step": 52190
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.651890992561824e-05,
+      "loss": 0.2702,
+      "step": 52195
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6516381534776436e-05,
+      "loss": 0.2934,
+      "step": 52200
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.651385299440201e-05,
+      "loss": 0.2758,
+      "step": 52205
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.651132430452778e-05,
+      "loss": 0.2816,
+      "step": 52210
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6508795465186596e-05,
+      "loss": 0.2823,
+      "step": 52215
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.650626647641129e-05,
+      "loss": 0.2805,
+      "step": 52220
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.65037373382347e-05,
+      "loss": 0.3046,
+      "step": 52225
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.650120805068965e-05,
+      "loss": 0.2818,
+      "step": 52230
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6498678613809e-05,
+      "loss": 0.2896,
+      "step": 52235
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.649614902762558e-05,
+      "loss": 0.2778,
+      "step": 52240
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.649361929217224e-05,
+      "loss": 0.2767,
+      "step": 52245
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.649108940748184e-05,
+      "loss": 0.2824,
+      "step": 52250
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.648855937358721e-05,
+      "loss": 0.2869,
+      "step": 52255
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.648602919052121e-05,
+      "loss": 0.3008,
+      "step": 52260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.648349885831669e-05,
+      "loss": 0.2764,
+      "step": 52265
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6480968377006495e-05,
+      "loss": 0.2816,
+      "step": 52270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.647843774662351e-05,
+      "loss": 0.2847,
+      "step": 52275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.647590696720057e-05,
+      "loss": 0.2593,
+      "step": 52280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.647337603877054e-05,
+      "loss": 0.2581,
+      "step": 52285
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.647084496136627e-05,
+      "loss": 0.2778,
+      "step": 52290
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.646831373502065e-05,
+      "loss": 0.2942,
+      "step": 52295
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6465782359766534e-05,
+      "loss": 0.3032,
+      "step": 52300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.646325083563679e-05,
+      "loss": 0.2676,
+      "step": 52305
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.646071916266429e-05,
+      "loss": 0.2821,
+      "step": 52310
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.64581873408819e-05,
+      "loss": 0.3013,
+      "step": 52315
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.64556553703225e-05,
+      "loss": 0.2605,
+      "step": 52320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.645312325101896e-05,
+      "loss": 0.2636,
+      "step": 52325
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.645059098300417e-05,
+      "loss": 0.2734,
+      "step": 52330
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.644805856631101e-05,
+      "loss": 0.3045,
+      "step": 52335
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.644552600097234e-05,
+      "loss": 0.2675,
+      "step": 52340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.644299328702106e-05,
+      "loss": 0.2913,
+      "step": 52345
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6440460424490055e-05,
+      "loss": 0.2898,
+      "step": 52350
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.643792741341221e-05,
+      "loss": 0.2949,
+      "step": 52355
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.643539425382042e-05,
+      "loss": 0.2776,
+      "step": 52360
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.643286094574757e-05,
+      "loss": 0.2794,
+      "step": 52365
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.643032748922655e-05,
+      "loss": 0.3072,
+      "step": 52370
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6427793884290265e-05,
+      "loss": 0.2788,
+      "step": 52375
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6425260130971604e-05,
+      "loss": 0.2757,
+      "step": 52380
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.642272622930347e-05,
+      "loss": 0.3123,
+      "step": 52385
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.642019217931877e-05,
+      "loss": 0.2864,
+      "step": 52390
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6417657981050396e-05,
+      "loss": 0.2776,
+      "step": 52395
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.641512363453126e-05,
+      "loss": 0.2893,
+      "step": 52400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6412589139794264e-05,
+      "loss": 0.303,
+      "step": 52405
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.641005449687231e-05,
+      "loss": 0.2639,
+      "step": 52410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.640751970579834e-05,
+      "loss": 0.3054,
+      "step": 52415
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.6404984766605235e-05,
+      "loss": 0.2825,
+      "step": 52420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.640244967932591e-05,
+      "loss": 0.2827,
+      "step": 52425
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.63999144439933e-05,
+      "loss": 0.2741,
+      "step": 52430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.639737906064031e-05,
+      "loss": 0.2635,
+      "step": 52435
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.639484352929986e-05,
+      "loss": 0.2647,
+      "step": 52440
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.639230785000489e-05,
+      "loss": 0.268,
+      "step": 52445
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6389772022788295e-05,
+      "loss": 0.2776,
+      "step": 52450
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.638723604768302e-05,
+      "loss": 0.2842,
+      "step": 52455
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.638469992472199e-05,
+      "loss": 0.2796,
+      "step": 52460
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.638216365393814e-05,
+      "loss": 0.2851,
+      "step": 52465
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6379627235364395e-05,
+      "loss": 0.3005,
+      "step": 52470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.637709066903368e-05,
+      "loss": 0.2855,
+      "step": 52475
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6374553954978955e-05,
+      "loss": 0.2723,
+      "step": 52480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.637201709323314e-05,
+      "loss": 0.2943,
+      "step": 52485
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.636948008382917e-05,
+      "loss": 0.2891,
+      "step": 52490
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.63669429268e-05,
+      "loss": 0.2916,
+      "step": 52495
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.636440562217856e-05,
+      "loss": 0.2823,
+      "step": 52500
+    },
+    {
+      "epoch": 1.85,
+      "eval_loss": 0.27478304505348206,
+      "eval_runtime": 10.5283,
+      "eval_samples_per_second": 9.498,
+      "eval_steps_per_second": 9.498,
+      "step": 52500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.636186816999781e-05,
+      "loss": 0.2729,
+      "step": 52505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6359330570290686e-05,
+      "loss": 0.2818,
+      "step": 52510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.635679282309013e-05,
+      "loss": 0.2786,
+      "step": 52515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.635425492842912e-05,
+      "loss": 0.2766,
+      "step": 52520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6351716886340585e-05,
+      "loss": 0.287,
+      "step": 52525
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6349178696857496e-05,
+      "loss": 0.2887,
+      "step": 52530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.634664036001279e-05,
+      "loss": 0.2625,
+      "step": 52535
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.634410187583944e-05,
+      "loss": 0.2751,
+      "step": 52540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6341563244370405e-05,
+      "loss": 0.303,
+      "step": 52545
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.633902446563865e-05,
+      "loss": 0.2807,
+      "step": 52550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.633648553967712e-05,
+      "loss": 0.2919,
+      "step": 52555
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.633394646651881e-05,
+      "loss": 0.2839,
+      "step": 52560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6331407246196674e-05,
+      "loss": 0.2871,
+      "step": 52565
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.632886787874368e-05,
+      "loss": 0.2677,
+      "step": 52570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6326328364192795e-05,
+      "loss": 0.2702,
+      "step": 52575
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.632378870257701e-05,
+      "loss": 0.28,
+      "step": 52580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.632124889392929e-05,
+      "loss": 0.2936,
+      "step": 52585
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.631870893828262e-05,
+      "loss": 0.2671,
+      "step": 52590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.631616883566996e-05,
+      "loss": 0.2628,
+      "step": 52595
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6313628586124317e-05,
+      "loss": 0.284,
+      "step": 52600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.631108818967865e-05,
+      "loss": 0.2837,
+      "step": 52605
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.630854764636597e-05,
+      "loss": 0.2802,
+      "step": 52610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.630600695621924e-05,
+      "loss": 0.2782,
+      "step": 52615
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.630346611927147e-05,
+      "loss": 0.2628,
+      "step": 52620
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6300925135555634e-05,
+      "loss": 0.286,
+      "step": 52625
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6298384005104736e-05,
+      "loss": 0.2759,
+      "step": 52630
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.629584272795177e-05,
+      "loss": 0.2929,
+      "step": 52635
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.629330130412974e-05,
+      "loss": 0.285,
+      "step": 52640
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.629075973367162e-05,
+      "loss": 0.2624,
+      "step": 52645
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.628821801661043e-05,
+      "loss": 0.2926,
+      "step": 52650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.628567615297916e-05,
+      "loss": 0.2976,
+      "step": 52655
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.628313414281084e-05,
+      "loss": 0.3159,
+      "step": 52660
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6280591986138446e-05,
+      "loss": 0.2822,
+      "step": 52665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6278049682995e-05,
+      "loss": 0.2769,
+      "step": 52670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6275507233413516e-05,
+      "loss": 0.2939,
+      "step": 52675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6272964637427e-05,
+      "loss": 0.2559,
+      "step": 52680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6270421895068465e-05,
+      "loss": 0.2857,
+      "step": 52685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6267879006370936e-05,
+      "loss": 0.2951,
+      "step": 52690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.626533597136742e-05,
+      "loss": 0.2767,
+      "step": 52695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.626279279009094e-05,
+      "loss": 0.2571,
+      "step": 52700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.626024946257452e-05,
+      "loss": 0.3061,
+      "step": 52705
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6257705988851176e-05,
+      "loss": 0.2548,
+      "step": 52710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6255162368953946e-05,
+      "loss": 0.2965,
+      "step": 52715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.6252618602915845e-05,
+      "loss": 0.307,
+      "step": 52720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6250074690769906e-05,
+      "loss": 0.2867,
+      "step": 52725
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6247530632549165e-05,
+      "loss": 0.2846,
+      "step": 52730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6244986428286656e-05,
+      "loss": 0.2995,
+      "step": 52735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.62424420780154e-05,
+      "loss": 0.2915,
+      "step": 52740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.623989758176844e-05,
+      "loss": 0.2876,
+      "step": 52745
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.623735293957883e-05,
+      "loss": 0.2866,
+      "step": 52750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.623480815147959e-05,
+      "loss": 0.299,
+      "step": 52755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.623226321750377e-05,
+      "loss": 0.3092,
+      "step": 52760
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.622971813768441e-05,
+      "loss": 0.3142,
+      "step": 52765
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.622717291205457e-05,
+      "loss": 0.2707,
+      "step": 52770
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6224627540647286e-05,
+      "loss": 0.2993,
+      "step": 52775
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.622208202349561e-05,
+      "loss": 0.2707,
+      "step": 52780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.621953636063259e-05,
+      "loss": 0.2887,
+      "step": 52785
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.621699055209129e-05,
+      "loss": 0.2877,
+      "step": 52790
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.621444459790476e-05,
+      "loss": 0.2692,
+      "step": 52795
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.621189849810606e-05,
+      "loss": 0.2881,
+      "step": 52800
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.620935225272823e-05,
+      "loss": 0.2931,
+      "step": 52805
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.620680586180437e-05,
+      "loss": 0.2877,
+      "step": 52810
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.620425932536751e-05,
+      "loss": 0.2648,
+      "step": 52815
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.620171264345073e-05,
+      "loss": 0.2814,
+      "step": 52820
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.619916581608709e-05,
+      "loss": 0.2752,
+      "step": 52825
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.619661884330967e-05,
+      "loss": 0.2623,
+      "step": 52830
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.619407172515153e-05,
+      "loss": 0.2936,
+      "step": 52835
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6191524461645736e-05,
+      "loss": 0.2566,
+      "step": 52840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.618897705282538e-05,
+      "loss": 0.2904,
+      "step": 52845
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6186429498723535e-05,
+      "loss": 0.2571,
+      "step": 52850
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.618388179937327e-05,
+      "loss": 0.2748,
+      "step": 52855
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6181333954807674e-05,
+      "loss": 0.3076,
+      "step": 52860
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.617878596505982e-05,
+      "loss": 0.2754,
+      "step": 52865
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6176237830162804e-05,
+      "loss": 0.2669,
+      "step": 52870
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.61736895501497e-05,
+      "loss": 0.2724,
+      "step": 52875
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6171141125053595e-05,
+      "loss": 0.2857,
+      "step": 52880
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.61685925549076e-05,
+      "loss": 0.2971,
+      "step": 52885
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.616604383974478e-05,
+      "loss": 0.2753,
+      "step": 52890
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.616349497959824e-05,
+      "loss": 0.3067,
+      "step": 52895
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6160945974501077e-05,
+      "loss": 0.2854,
+      "step": 52900
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6158396824486385e-05,
+      "loss": 0.287,
+      "step": 52905
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6155847529587264e-05,
+      "loss": 0.3051,
+      "step": 52910
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6153298089836815e-05,
+      "loss": 0.2939,
+      "step": 52915
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.615074850526815e-05,
+      "loss": 0.2884,
+      "step": 52920
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.614819877591435e-05,
+      "loss": 0.3121,
+      "step": 52925
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6145648901808535e-05,
+      "loss": 0.2799,
+      "step": 52930
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.614309888298383e-05,
+      "loss": 0.3066,
+      "step": 52935
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.614054871947332e-05,
+      "loss": 0.2771,
+      "step": 52940
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6137998411310126e-05,
+      "loss": 0.2908,
+      "step": 52945
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.613544795852736e-05,
+      "loss": 0.2769,
+      "step": 52950
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6132897361158145e-05,
+      "loss": 0.2981,
+      "step": 52955
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.613034661923559e-05,
+      "loss": 0.2712,
+      "step": 52960
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.612779573279283e-05,
+      "loss": 0.2836,
+      "step": 52965
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.612524470186297e-05,
+      "loss": 0.2873,
+      "step": 52970
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6122693526479135e-05,
+      "loss": 0.2933,
+      "step": 52975
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6120142206674455e-05,
+      "loss": 0.2809,
+      "step": 52980
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.611759074248206e-05,
+      "loss": 0.2955,
+      "step": 52985
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.611503913393508e-05,
+      "loss": 0.2707,
+      "step": 52990
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.611248738106663e-05,
+      "loss": 0.2893,
+      "step": 52995
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.6109935483909876e-05,
+      "loss": 0.3082,
+      "step": 53000
+    },
+    {
+      "epoch": 1.86,
+      "eval_loss": 0.27502691745758057,
+      "eval_runtime": 10.5394,
+      "eval_samples_per_second": 9.488,
+      "eval_steps_per_second": 9.488,
+      "step": 53000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.610738344249791e-05,
+      "loss": 0.2816,
+      "step": 53005
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6104831256863905e-05,
+      "loss": 0.2889,
+      "step": 53010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.610227892704097e-05,
+      "loss": 0.2816,
+      "step": 53015
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.609972645306227e-05,
+      "loss": 0.2639,
+      "step": 53020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6097173834960935e-05,
+      "loss": 0.2738,
+      "step": 53025
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.609462107277012e-05,
+      "loss": 0.2803,
+      "step": 53030
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.609206816652295e-05,
+      "loss": 0.2838,
+      "step": 53035
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6089515116252584e-05,
+      "loss": 0.2923,
+      "step": 53040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6086961921992173e-05,
+      "loss": 0.283,
+      "step": 53045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6084408583774875e-05,
+      "loss": 0.2805,
+      "step": 53050
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.608185510163383e-05,
+      "loss": 0.2876,
+      "step": 53055
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.60793014756022e-05,
+      "loss": 0.2843,
+      "step": 53060
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6076747705713145e-05,
+      "loss": 0.2778,
+      "step": 53065
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.607419379199982e-05,
+      "loss": 0.3115,
+      "step": 53070
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6071639734495386e-05,
+      "loss": 0.28,
+      "step": 53075
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.606908553323301e-05,
+      "loss": 0.2613,
+      "step": 53080
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.606653118824586e-05,
+      "loss": 0.2805,
+      "step": 53085
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.606397669956707e-05,
+      "loss": 0.2614,
+      "step": 53090
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.606142206722985e-05,
+      "loss": 0.293,
+      "step": 53095
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6058867291267354e-05,
+      "loss": 0.2816,
+      "step": 53100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.605631237171277e-05,
+      "loss": 0.287,
+      "step": 53105
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6053757308599235e-05,
+      "loss": 0.2713,
+      "step": 53110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6051202101959955e-05,
+      "loss": 0.3059,
+      "step": 53115
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.604864675182809e-05,
+      "loss": 0.2723,
+      "step": 53120
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6046091258236834e-05,
+      "loss": 0.2759,
+      "step": 53125
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.604353562121936e-05,
+      "loss": 0.2998,
+      "step": 53130
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6040979840808866e-05,
+      "loss": 0.2741,
+      "step": 53135
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.603842391703851e-05,
+      "loss": 0.2959,
+      "step": 53140
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.60358678499415e-05,
+      "loss": 0.2795,
+      "step": 53145
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.603331163955102e-05,
+      "loss": 0.2791,
+      "step": 53150
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6030755285900255e-05,
+      "loss": 0.2649,
+      "step": 53155
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.602819878902241e-05,
+      "loss": 0.291,
+      "step": 53160
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.602564214895067e-05,
+      "loss": 0.2913,
+      "step": 53165
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.602308536571822e-05,
+      "loss": 0.3051,
+      "step": 53170
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.602052843935828e-05,
+      "loss": 0.2615,
+      "step": 53175
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.601797136990404e-05,
+      "loss": 0.2853,
+      "step": 53180
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.601541415738871e-05,
+      "loss": 0.3029,
+      "step": 53185
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6012856801845476e-05,
+      "loss": 0.2798,
+      "step": 53190
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6010299303307564e-05,
+      "loss": 0.2792,
+      "step": 53195
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.6007741661808156e-05,
+      "loss": 0.2701,
+      "step": 53200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.60051838773805e-05,
+      "loss": 0.2634,
+      "step": 53205
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.600262595005777e-05,
+      "loss": 0.274,
+      "step": 53210
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.600006787987319e-05,
+      "loss": 0.276,
+      "step": 53215
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5997509666859986e-05,
+      "loss": 0.254,
+      "step": 53220
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.599495131105136e-05,
+      "loss": 0.2867,
+      "step": 53225
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.599239281248054e-05,
+      "loss": 0.2812,
+      "step": 53230
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.598983417118075e-05,
+      "loss": 0.2513,
+      "step": 53235
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5987275387185196e-05,
+      "loss": 0.3058,
+      "step": 53240
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5984716460527116e-05,
+      "loss": 0.2823,
+      "step": 53245
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.598215739123973e-05,
+      "loss": 0.2615,
+      "step": 53250
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.597959817935628e-05,
+      "loss": 0.2778,
+      "step": 53255
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5977038824909966e-05,
+      "loss": 0.272,
+      "step": 53260
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.597447932793404e-05,
+      "loss": 0.2803,
+      "step": 53265
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5971919688461744e-05,
+      "loss": 0.2661,
+      "step": 53270
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.596935990652629e-05,
+      "loss": 0.2634,
+      "step": 53275
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.596679998216093e-05,
+      "loss": 0.2915,
+      "step": 53280
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.59642399153989e-05,
+      "loss": 0.2551,
+      "step": 53285
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.5961679706273447e-05,
+      "loss": 0.2735,
+      "step": 53290
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.59591193548178e-05,
+      "loss": 0.2879,
+      "step": 53295
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.595655886106521e-05,
+      "loss": 0.3059,
+      "step": 53300
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5953998225048925e-05,
+      "loss": 0.2938,
+      "step": 53305
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.595143744680219e-05,
+      "loss": 0.2826,
+      "step": 53310
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.594887652635826e-05,
+      "loss": 0.2837,
+      "step": 53315
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.594631546375038e-05,
+      "loss": 0.2787,
+      "step": 53320
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5943754259011813e-05,
+      "loss": 0.2896,
+      "step": 53325
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.594119291217581e-05,
+      "loss": 0.2584,
+      "step": 53330
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5938631423275625e-05,
+      "loss": 0.2586,
+      "step": 53335
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5936069792344514e-05,
+      "loss": 0.2713,
+      "step": 53340
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.593350801941575e-05,
+      "loss": 0.286,
+      "step": 53345
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.593094610452259e-05,
+      "loss": 0.2882,
+      "step": 53350
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.59283840476983e-05,
+      "loss": 0.2778,
+      "step": 53355
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.592582184897614e-05,
+      "loss": 0.2799,
+      "step": 53360
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.592325950838938e-05,
+      "loss": 0.3066,
+      "step": 53365
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5920697025971306e-05,
+      "loss": 0.2768,
+      "step": 53370
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.591813440175518e-05,
+      "loss": 0.2807,
+      "step": 53375
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5915571635774256e-05,
+      "loss": 0.2799,
+      "step": 53380
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5913008728061845e-05,
+      "loss": 0.2797,
+      "step": 53385
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.59104456786512e-05,
+      "loss": 0.2818,
+      "step": 53390
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.59078824875756e-05,
+      "loss": 0.2837,
+      "step": 53395
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.590531915486835e-05,
+      "loss": 0.2849,
+      "step": 53400
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.590275568056272e-05,
+      "loss": 0.2767,
+      "step": 53405
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5900192064691976e-05,
+      "loss": 0.2593,
+      "step": 53410
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5897628307289435e-05,
+      "loss": 0.2942,
+      "step": 53415
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.589506440838836e-05,
+      "loss": 0.2855,
+      "step": 53420
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.589250036802206e-05,
+      "loss": 0.2643,
+      "step": 53425
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.588993618622383e-05,
+      "loss": 0.3168,
+      "step": 53430
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.588737186302695e-05,
+      "loss": 0.2947,
+      "step": 53435
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5884807398464713e-05,
+      "loss": 0.271,
+      "step": 53440
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.588224279257043e-05,
+      "loss": 0.2879,
+      "step": 53445
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.58796780453774e-05,
+      "loss": 0.2801,
+      "step": 53450
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.587711315691891e-05,
+      "loss": 0.288,
+      "step": 53455
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.587454812722829e-05,
+      "loss": 0.2895,
+      "step": 53460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.587198295633882e-05,
+      "loss": 0.2864,
+      "step": 53465
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.586941764428381e-05,
+      "loss": 0.2726,
+      "step": 53470
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.586685219109659e-05,
+      "loss": 0.2633,
+      "step": 53475
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.586428659681044e-05,
+      "loss": 0.274,
+      "step": 53480
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5861720861458695e-05,
+      "loss": 0.2746,
+      "step": 53485
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.585915498507467e-05,
+      "loss": 0.2978,
+      "step": 53490
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5856588967691664e-05,
+      "loss": 0.258,
+      "step": 53495
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.585402280934301e-05,
+      "loss": 0.3014,
+      "step": 53500
+    },
+    {
+      "epoch": 1.88,
+      "eval_loss": 0.27490025758743286,
+      "eval_runtime": 10.529,
+      "eval_samples_per_second": 9.498,
+      "eval_steps_per_second": 9.498,
+      "step": 53500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.585145651006202e-05,
+      "loss": 0.2879,
+      "step": 53505
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.584889006988202e-05,
+      "loss": 0.2526,
+      "step": 53510
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.584632348883633e-05,
+      "loss": 0.2781,
+      "step": 53515
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5843756766958285e-05,
+      "loss": 0.3115,
+      "step": 53520
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.58411899042812e-05,
+      "loss": 0.2705,
+      "step": 53525
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.583862290083841e-05,
+      "loss": 0.2759,
+      "step": 53530
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.583605575666324e-05,
+      "loss": 0.3126,
+      "step": 53535
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.583348847178903e-05,
+      "loss": 0.3062,
+      "step": 53540
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.583092104624911e-05,
+      "loss": 0.28,
+      "step": 53545
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.582835348007682e-05,
+      "loss": 0.3002,
+      "step": 53550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5825785773305496e-05,
+      "loss": 0.2674,
+      "step": 53555
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.582373150667954e-05,
+      "loss": 0.2849,
+      "step": 53560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.582116354691397e-05,
+      "loss": 0.2912,
+      "step": 53565
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.5818595446642725e-05,
+      "loss": 0.2874,
+      "step": 53570
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.581602720589915e-05,
+      "loss": 0.2762,
+      "step": 53575
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5813458824716594e-05,
+      "loss": 0.296,
+      "step": 53580
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.58108903031284e-05,
+      "loss": 0.2853,
+      "step": 53585
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.580832164116791e-05,
+      "loss": 0.2781,
+      "step": 53590
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.580575283886849e-05,
+      "loss": 0.2912,
+      "step": 53595
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5803183896263495e-05,
+      "loss": 0.2779,
+      "step": 53600
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5800614813386273e-05,
+      "loss": 0.2694,
+      "step": 53605
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.579804559027019e-05,
+      "loss": 0.2939,
+      "step": 53610
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.57954762269486e-05,
+      "loss": 0.2649,
+      "step": 53615
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.579290672345486e-05,
+      "loss": 0.2454,
+      "step": 53620
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.579033707982234e-05,
+      "loss": 0.3101,
+      "step": 53625
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.578776729608441e-05,
+      "loss": 0.2627,
+      "step": 53630
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.578519737227443e-05,
+      "loss": 0.2664,
+      "step": 53635
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.578262730842577e-05,
+      "loss": 0.2712,
+      "step": 53640
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5780057104571796e-05,
+      "loss": 0.3049,
+      "step": 53645
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5777486760745887e-05,
+      "loss": 0.2892,
+      "step": 53650
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.577491627698142e-05,
+      "loss": 0.2862,
+      "step": 53655
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5772345653311755e-05,
+      "loss": 0.2894,
+      "step": 53660
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.57697748897703e-05,
+      "loss": 0.2764,
+      "step": 53665
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.57672039863904e-05,
+      "loss": 0.2925,
+      "step": 53670
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.576463294320545e-05,
+      "loss": 0.2872,
+      "step": 53675
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.576206176024884e-05,
+      "loss": 0.2931,
+      "step": 53680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.575949043755395e-05,
+      "loss": 0.2835,
+      "step": 53685
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5756918975154174e-05,
+      "loss": 0.2891,
+      "step": 53690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.575434737308289e-05,
+      "loss": 0.2525,
+      "step": 53695
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.575177563137349e-05,
+      "loss": 0.3179,
+      "step": 53700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5749203750059365e-05,
+      "loss": 0.3078,
+      "step": 53705
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.574663172917392e-05,
+      "loss": 0.2811,
+      "step": 53710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.574405956875055e-05,
+      "loss": 0.2791,
+      "step": 53715
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.574148726882264e-05,
+      "loss": 0.3034,
+      "step": 53720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5738914829423596e-05,
+      "loss": 0.2864,
+      "step": 53725
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.573634225058682e-05,
+      "loss": 0.2926,
+      "step": 53730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5733769532345715e-05,
+      "loss": 0.2712,
+      "step": 53735
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.573119667473369e-05,
+      "loss": 0.2971,
+      "step": 53740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.572862367778415e-05,
+      "loss": 0.2853,
+      "step": 53745
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.57260505415305e-05,
+      "loss": 0.2959,
+      "step": 53750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5723477266006145e-05,
+      "loss": 0.2991,
+      "step": 53755
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.572090385124451e-05,
+      "loss": 0.3107,
+      "step": 53760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5718330297279e-05,
+      "loss": 0.2728,
+      "step": 53765
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5715756604143045e-05,
+      "loss": 0.2799,
+      "step": 53770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.571318277187005e-05,
+      "loss": 0.2949,
+      "step": 53775
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5710608800493425e-05,
+      "loss": 0.2918,
+      "step": 53780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.570803469004661e-05,
+      "loss": 0.3069,
+      "step": 53785
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.570546044056303e-05,
+      "loss": 0.3087,
+      "step": 53790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.570288605207609e-05,
+      "loss": 0.2924,
+      "step": 53795
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.570031152461923e-05,
+      "loss": 0.2727,
+      "step": 53800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5697736858225885e-05,
+      "loss": 0.3101,
+      "step": 53805
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.569516205292947e-05,
+      "loss": 0.2856,
+      "step": 53810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5692587108763424e-05,
+      "loss": 0.2822,
+      "step": 53815
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.569001202576119e-05,
+      "loss": 0.289,
+      "step": 53820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5687436803956184e-05,
+      "loss": 0.2755,
+      "step": 53825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.568486144338186e-05,
+      "loss": 0.2801,
+      "step": 53830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5682285944071645e-05,
+      "loss": 0.2956,
+      "step": 53835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.567971030605899e-05,
+      "loss": 0.2879,
+      "step": 53840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.5677134529377334e-05,
+      "loss": 0.2941,
+      "step": 53845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.567455861406013e-05,
+      "loss": 0.2889,
+      "step": 53850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.567198256014081e-05,
+      "loss": 0.2969,
+      "step": 53855
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.566940636765284e-05,
+      "loss": 0.2961,
+      "step": 53860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.566683003662964e-05,
+      "loss": 0.2817,
+      "step": 53865
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5664253567104694e-05,
+      "loss": 0.2542,
+      "step": 53870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5661676959111444e-05,
+      "loss": 0.2545,
+      "step": 53875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.565910021268334e-05,
+      "loss": 0.2826,
+      "step": 53880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.565652332785385e-05,
+      "loss": 0.2631,
+      "step": 53885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5653946304656426e-05,
+      "loss": 0.2975,
+      "step": 53890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.565136914312453e-05,
+      "loss": 0.29,
+      "step": 53895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.564879184329162e-05,
+      "loss": 0.277,
+      "step": 53900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5646214405191176e-05,
+      "loss": 0.2721,
+      "step": 53905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.564363682885665e-05,
+      "loss": 0.283,
+      "step": 53910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.564105911432151e-05,
+      "loss": 0.2778,
+      "step": 53915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.563848126161924e-05,
+      "loss": 0.2981,
+      "step": 53920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.563590327078329e-05,
+      "loss": 0.3012,
+      "step": 53925
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5633325141847144e-05,
+      "loss": 0.2836,
+      "step": 53930
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.563074687484429e-05,
+      "loss": 0.2967,
+      "step": 53935
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5628168469808184e-05,
+      "loss": 0.2996,
+      "step": 53940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.562558992677233e-05,
+      "loss": 0.2953,
+      "step": 53945
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.562301124577018e-05,
+      "loss": 0.2628,
+      "step": 53950
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5620432426835226e-05,
+      "loss": 0.2767,
+      "step": 53955
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.561785347000096e-05,
+      "loss": 0.2936,
+      "step": 53960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.561527437530088e-05,
+      "loss": 0.295,
+      "step": 53965
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.561269514276844e-05,
+      "loss": 0.2892,
+      "step": 53970
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.561011577243715e-05,
+      "loss": 0.2718,
+      "step": 53975
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5607536264340497e-05,
+      "loss": 0.2776,
+      "step": 53980
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.560495661851198e-05,
+      "loss": 0.2718,
+      "step": 53985
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5602376834985094e-05,
+      "loss": 0.2969,
+      "step": 53990
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5599796913793326e-05,
+      "loss": 0.2846,
+      "step": 53995
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.559721685497019e-05,
+      "loss": 0.2723,
+      "step": 54000
+    },
+    {
+      "epoch": 1.9,
+      "eval_loss": 0.27377253770828247,
+      "eval_runtime": 10.5385,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 54000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.559463665854916e-05,
+      "loss": 0.2901,
+      "step": 54005
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5592056324563775e-05,
+      "loss": 0.2751,
+      "step": 54010
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5589475853047505e-05,
+      "loss": 0.2912,
+      "step": 54015
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5586895244033874e-05,
+      "loss": 0.2607,
+      "step": 54020
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5584314497556385e-05,
+      "loss": 0.2925,
+      "step": 54025
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.558173361364855e-05,
+      "loss": 0.2835,
+      "step": 54030
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.557915259234387e-05,
+      "loss": 0.275,
+      "step": 54035
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5576571433675876e-05,
+      "loss": 0.2868,
+      "step": 54040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5573990137678065e-05,
+      "loss": 0.2768,
+      "step": 54045
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.557140870438397e-05,
+      "loss": 0.2995,
+      "step": 54050
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.556882713382709e-05,
+      "loss": 0.2717,
+      "step": 54055
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.556624542604096e-05,
+      "loss": 0.2686,
+      "step": 54060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5563663581059084e-05,
+      "loss": 0.2718,
+      "step": 54065
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.556108159891502e-05,
+      "loss": 0.3123,
+      "step": 54070
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.555849947964225e-05,
+      "loss": 0.2785,
+      "step": 54075
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.555591722327434e-05,
+      "loss": 0.2834,
+      "step": 54080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.555333482984479e-05,
+      "loss": 0.277,
+      "step": 54085
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.555075229938715e-05,
+      "loss": 0.3086,
+      "step": 54090
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5548169631934936e-05,
+      "loss": 0.3118,
+      "step": 54095
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5545586827521695e-05,
+      "loss": 0.2825,
+      "step": 54100
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.554300388618096e-05,
+      "loss": 0.3147,
+      "step": 54105
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5540420807946264e-05,
+      "loss": 0.2986,
+      "step": 54110
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.553783759285115e-05,
+      "loss": 0.2992,
+      "step": 54115
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.553525424092916e-05,
+      "loss": 0.2876,
+      "step": 54120
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.553267075221385e-05,
+      "loss": 0.2788,
+      "step": 54125
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.553008712673873e-05,
+      "loss": 0.2929,
+      "step": 54130
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.552750336453738e-05,
+      "loss": 0.3044,
+      "step": 54135
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.552491946564334e-05,
+      "loss": 0.2799,
+      "step": 54140
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.5522335430090153e-05,
+      "loss": 0.2941,
+      "step": 54145
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.551975125791137e-05,
+      "loss": 0.2747,
+      "step": 54150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.551716694914056e-05,
+      "loss": 0.286,
+      "step": 54155
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.551458250381126e-05,
+      "loss": 0.2917,
+      "step": 54160
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.551199792195704e-05,
+      "loss": 0.3029,
+      "step": 54165
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5509413203611445e-05,
+      "loss": 0.2845,
+      "step": 54170
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.550682834880806e-05,
+      "loss": 0.2963,
+      "step": 54175
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.550424335758042e-05,
+      "loss": 0.2871,
+      "step": 54180
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.550165822996212e-05,
+      "loss": 0.2846,
+      "step": 54185
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5499072965986694e-05,
+      "loss": 0.2951,
+      "step": 54190
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.549648756568773e-05,
+      "loss": 0.2784,
+      "step": 54195
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.549390202909879e-05,
+      "loss": 0.278,
+      "step": 54200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5491316356253446e-05,
+      "loss": 0.2768,
+      "step": 54205
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5488730547185276e-05,
+      "loss": 0.2931,
+      "step": 54210
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.548614460192785e-05,
+      "loss": 0.2693,
+      "step": 54215
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5483558520514755e-05,
+      "loss": 0.2674,
+      "step": 54220
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.548097230297954e-05,
+      "loss": 0.2962,
+      "step": 54225
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.547838594935583e-05,
+      "loss": 0.2953,
+      "step": 54230
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.547579945967718e-05,
+      "loss": 0.2914,
+      "step": 54235
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.547321283397718e-05,
+      "loss": 0.3063,
+      "step": 54240
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.54706260722894e-05,
+      "loss": 0.2954,
+      "step": 54245
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5468039174647446e-05,
+      "loss": 0.298,
+      "step": 54250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.54654521410849e-05,
+      "loss": 0.2783,
+      "step": 54255
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.546286497163537e-05,
+      "loss": 0.2827,
+      "step": 54260
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.546027766633242e-05,
+      "loss": 0.2954,
+      "step": 54265
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.545769022520966e-05,
+      "loss": 0.2752,
+      "step": 54270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5455102648300686e-05,
+      "loss": 0.2836,
+      "step": 54275
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.545251493563909e-05,
+      "loss": 0.2787,
+      "step": 54280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.544992708725848e-05,
+      "loss": 0.2801,
+      "step": 54285
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5447339103192454e-05,
+      "loss": 0.2571,
+      "step": 54290
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.544475098347462e-05,
+      "loss": 0.2798,
+      "step": 54295
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.544216272813857e-05,
+      "loss": 0.2982,
+      "step": 54300
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.543957433721792e-05,
+      "loss": 0.276,
+      "step": 54305
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.543698581074628e-05,
+      "loss": 0.2727,
+      "step": 54310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.543439714875725e-05,
+      "loss": 0.2706,
+      "step": 54315
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5431808351284455e-05,
+      "loss": 0.2784,
+      "step": 54320
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.54292194183615e-05,
+      "loss": 0.295,
+      "step": 54325
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.542663035002201e-05,
+      "loss": 0.2611,
+      "step": 54330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.542404114629959e-05,
+      "loss": 0.2928,
+      "step": 54335
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5421451807227866e-05,
+      "loss": 0.2922,
+      "step": 54340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.541886233284047e-05,
+      "loss": 0.277,
+      "step": 54345
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5416272723171e-05,
+      "loss": 0.2538,
+      "step": 54350
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5413682978253096e-05,
+      "loss": 0.2852,
+      "step": 54355
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.541109309812038e-05,
+      "loss": 0.3012,
+      "step": 54360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.540850308280649e-05,
+      "loss": 0.2755,
+      "step": 54365
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5405912932345034e-05,
+      "loss": 0.2756,
+      "step": 54370
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.540332264676967e-05,
+      "loss": 0.2885,
+      "step": 54375
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.540073222611401e-05,
+      "loss": 0.2794,
+      "step": 54380
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.53981416704117e-05,
+      "loss": 0.2606,
+      "step": 54385
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5395550979696366e-05,
+      "loss": 0.2811,
+      "step": 54390
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.539296015400166e-05,
+      "loss": 0.2918,
+      "step": 54395
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5390369193361204e-05,
+      "loss": 0.2801,
+      "step": 54400
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.538829632771053e-05,
+      "loss": 0.2677,
+      "step": 54405
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5385705124252526e-05,
+      "loss": 0.2817,
+      "step": 54410
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5383113785943e-05,
+      "loss": 0.2995,
+      "step": 54415
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.5380522312815574e-05,
+      "loss": 0.264,
+      "step": 54420
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.53779307049039e-05,
+      "loss": 0.2944,
+      "step": 54425
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.537533896224165e-05,
+      "loss": 0.2951,
+      "step": 54430
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.537274708486246e-05,
+      "loss": 0.2744,
+      "step": 54435
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5370155072799974e-05,
+      "loss": 0.3025,
+      "step": 54440
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.536756292608787e-05,
+      "loss": 0.3011,
+      "step": 54445
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.536497064475979e-05,
+      "loss": 0.2933,
+      "step": 54450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.536237822884939e-05,
+      "loss": 0.2763,
+      "step": 54455
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5359785678390346e-05,
+      "loss": 0.3083,
+      "step": 54460
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.535719299341631e-05,
+      "loss": 0.289,
+      "step": 54465
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.535460017396095e-05,
+      "loss": 0.2785,
+      "step": 54470
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.535200722005793e-05,
+      "loss": 0.2717,
+      "step": 54475
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5349414131740914e-05,
+      "loss": 0.2753,
+      "step": 54480
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5346820909043585e-05,
+      "loss": 0.2864,
+      "step": 54485
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.53442275519996e-05,
+      "loss": 0.289,
+      "step": 54490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.534163406064264e-05,
+      "loss": 0.3057,
+      "step": 54495
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.533904043500637e-05,
+      "loss": 0.293,
+      "step": 54500
+    },
+    {
+      "epoch": 1.92,
+      "eval_loss": 0.27375850081443787,
+      "eval_runtime": 10.5429,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 54500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.533644667512448e-05,
+      "loss": 0.3125,
+      "step": 54505
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5333852781030635e-05,
+      "loss": 0.3063,
+      "step": 54510
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5331258752758536e-05,
+      "loss": 0.2904,
+      "step": 54515
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5328664590341836e-05,
+      "loss": 0.3004,
+      "step": 54520
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.532607029381424e-05,
+      "loss": 0.2887,
+      "step": 54525
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.532347586320942e-05,
+      "loss": 0.2682,
+      "step": 54530
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.532088129856107e-05,
+      "loss": 0.2665,
+      "step": 54535
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.531828659990289e-05,
+      "loss": 0.3099,
+      "step": 54540
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.531569176726855e-05,
+      "loss": 0.2926,
+      "step": 54545
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5313096800691756e-05,
+      "loss": 0.2811,
+      "step": 54550
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5310501700206186e-05,
+      "loss": 0.2583,
+      "step": 54555
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5307906465845554e-05,
+      "loss": 0.314,
+      "step": 54560
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5305311097643545e-05,
+      "loss": 0.3002,
+      "step": 54565
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.530271559563387e-05,
+      "loss": 0.32,
+      "step": 54570
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.530011995985022e-05,
+      "loss": 0.2996,
+      "step": 54575
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.52975241903263e-05,
+      "loss": 0.2864,
+      "step": 54580
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.529492828709582e-05,
+      "loss": 0.2945,
+      "step": 54585
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5292332250192474e-05,
+      "loss": 0.2948,
+      "step": 54590
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.528973607964997e-05,
+      "loss": 0.3156,
+      "step": 54595
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5287139775502034e-05,
+      "loss": 0.2755,
+      "step": 54600
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.528454333778237e-05,
+      "loss": 0.2495,
+      "step": 54605
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.528194676652468e-05,
+      "loss": 0.2742,
+      "step": 54610
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.527935006176269e-05,
+      "loss": 0.2778,
+      "step": 54615
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.527675322353011e-05,
+      "loss": 0.2977,
+      "step": 54620
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.527415625186067e-05,
+      "loss": 0.2769,
+      "step": 54625
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.527155914678808e-05,
+      "loss": 0.3208,
+      "step": 54630
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.526896190834607e-05,
+      "loss": 0.2947,
+      "step": 54635
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.526636453656834e-05,
+      "loss": 0.283,
+      "step": 54640
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.526376703148864e-05,
+      "loss": 0.2712,
+      "step": 54645
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.526116939314069e-05,
+      "loss": 0.2911,
+      "step": 54650
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.525857162155822e-05,
+      "loss": 0.2716,
+      "step": 54655
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.525597371677496e-05,
+      "loss": 0.274,
+      "step": 54660
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.525337567882464e-05,
+      "loss": 0.2763,
+      "step": 54665
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5250777507740986e-05,
+      "loss": 0.3088,
+      "step": 54670
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.524817920355775e-05,
+      "loss": 0.257,
+      "step": 54675
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.524558076630865e-05,
+      "loss": 0.2821,
+      "step": 54680
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.524298219602745e-05,
+      "loss": 0.2752,
+      "step": 54685
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5240383492747874e-05,
+      "loss": 0.293,
+      "step": 54690
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.523778465650366e-05,
+      "loss": 0.2954,
+      "step": 54695
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.523518568732856e-05,
+      "loss": 0.2765,
+      "step": 54700
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5232586585256324e-05,
+      "loss": 0.2841,
+      "step": 54705
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5229987350320695e-05,
+      "loss": 0.2696,
+      "step": 54710
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.522738798255542e-05,
+      "loss": 0.291,
+      "step": 54715
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.522478848199425e-05,
+      "loss": 0.2888,
+      "step": 54720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5222188848670936e-05,
+      "loss": 0.2965,
+      "step": 54725
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.521958908261924e-05,
+      "loss": 0.2867,
+      "step": 54730
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5216989183872925e-05,
+      "loss": 0.2465,
+      "step": 54735
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.521438915246572e-05,
+      "loss": 0.2976,
+      "step": 54740
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5211788988431424e-05,
+      "loss": 0.2708,
+      "step": 54745
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.520918869180376e-05,
+      "loss": 0.2836,
+      "step": 54750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.520658826261651e-05,
+      "loss": 0.2953,
+      "step": 54755
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.520398770090345e-05,
+      "loss": 0.2719,
+      "step": 54760
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5201387006698326e-05,
+      "loss": 0.3359,
+      "step": 54765
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.519878618003493e-05,
+      "loss": 0.2837,
+      "step": 54770
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5196185220947e-05,
+      "loss": 0.2681,
+      "step": 54775
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5193584129468326e-05,
+      "loss": 0.2896,
+      "step": 54780
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.519098290563267e-05,
+      "loss": 0.3084,
+      "step": 54785
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.518838154947384e-05,
+      "loss": 0.2752,
+      "step": 54790
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.518578006102558e-05,
+      "loss": 0.2897,
+      "step": 54795
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.518317844032168e-05,
+      "loss": 0.2674,
+      "step": 54800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5180576687395915e-05,
+      "loss": 0.2786,
+      "step": 54805
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.517797480228207e-05,
+      "loss": 0.2788,
+      "step": 54810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.517537278501393e-05,
+      "loss": 0.2804,
+      "step": 54815
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.517277063562529e-05,
+      "loss": 0.2814,
+      "step": 54820
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.517016835414992e-05,
+      "loss": 0.2579,
+      "step": 54825
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5167565940621614e-05,
+      "loss": 0.2763,
+      "step": 54830
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.516496339507416e-05,
+      "loss": 0.2942,
+      "step": 54835
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5162360717541374e-05,
+      "loss": 0.2744,
+      "step": 54840
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.515975790805701e-05,
+      "loss": 0.28,
+      "step": 54845
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.51571549666549e-05,
+      "loss": 0.2785,
+      "step": 54850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5154551893368824e-05,
+      "loss": 0.2837,
+      "step": 54855
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5151948688232575e-05,
+      "loss": 0.2921,
+      "step": 54860
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5149345351279975e-05,
+      "loss": 0.2606,
+      "step": 54865
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.51467418825448e-05,
+      "loss": 0.2747,
+      "step": 54870
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5144138282060886e-05,
+      "loss": 0.2833,
+      "step": 54875
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.514153454986201e-05,
+      "loss": 0.302,
+      "step": 54880
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5138930685982e-05,
+      "loss": 0.2866,
+      "step": 54885
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.513632669045464e-05,
+      "loss": 0.2823,
+      "step": 54890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.513372256331378e-05,
+      "loss": 0.2745,
+      "step": 54895
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.513111830459319e-05,
+      "loss": 0.2888,
+      "step": 54900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.512851391432672e-05,
+      "loss": 0.2857,
+      "step": 54905
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.512590939254816e-05,
+      "loss": 0.2964,
+      "step": 54910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.512330473929135e-05,
+      "loss": 0.2619,
+      "step": 54915
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.51206999545901e-05,
+      "loss": 0.2906,
+      "step": 54920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5118095038478225e-05,
+      "loss": 0.2607,
+      "step": 54925
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.511548999098956e-05,
+      "loss": 0.2805,
+      "step": 54930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.511288481215792e-05,
+      "loss": 0.263,
+      "step": 54935
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5110279502017126e-05,
+      "loss": 0.2726,
+      "step": 54940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.510767406060103e-05,
+      "loss": 0.285,
+      "step": 54945
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5105068487943436e-05,
+      "loss": 0.2792,
+      "step": 54950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.510246278407821e-05,
+      "loss": 0.2855,
+      "step": 54955
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.509985694903913e-05,
+      "loss": 0.287,
+      "step": 54960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.509725098286009e-05,
+      "loss": 0.2961,
+      "step": 54965
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5094644885574886e-05,
+      "loss": 0.2824,
+      "step": 54970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5092038657217376e-05,
+      "loss": 0.2817,
+      "step": 54975
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.508943229782139e-05,
+      "loss": 0.2871,
+      "step": 54980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.508682580742079e-05,
+      "loss": 0.2864,
+      "step": 54985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.508421918604938e-05,
+      "loss": 0.2935,
+      "step": 54990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.508161243374105e-05,
+      "loss": 0.2825,
+      "step": 54995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5079005550529614e-05,
+      "loss": 0.2652,
+      "step": 55000
+    },
+    {
+      "epoch": 1.94,
+      "eval_loss": 0.2729741930961609,
+      "eval_runtime": 10.5572,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 9.472,
+      "step": 55000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.507639853644893e-05,
+      "loss": 0.2845,
+      "step": 55005
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.507379139153287e-05,
+      "loss": 0.2745,
+      "step": 55010
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5071184115815255e-05,
+      "loss": 0.2683,
+      "step": 55015
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.506857670932996e-05,
+      "loss": 0.3008,
+      "step": 55020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.506596917211082e-05,
+      "loss": 0.2836,
+      "step": 55025
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5063361504191714e-05,
+      "loss": 0.2907,
+      "step": 55030
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.506075370560648e-05,
+      "loss": 0.2794,
+      "step": 55035
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5058145776389e-05,
+      "loss": 0.2697,
+      "step": 55040
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.505553771657312e-05,
+      "loss": 0.2841,
+      "step": 55045
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.505292952619272e-05,
+      "loss": 0.2808,
+      "step": 55050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5050321205281636e-05,
+      "loss": 0.2969,
+      "step": 55055
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.504771275387377e-05,
+      "loss": 0.2951,
+      "step": 55060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.504510417200297e-05,
+      "loss": 0.2714,
+      "step": 55065
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.504249545970312e-05,
+      "loss": 0.2777,
+      "step": 55070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.503988661700807e-05,
+      "loss": 0.2983,
+      "step": 55075
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.503727764395172e-05,
+      "loss": 0.2766,
+      "step": 55080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.503466854056793e-05,
+      "loss": 0.2603,
+      "step": 55085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.503205930689059e-05,
+      "loss": 0.3038,
+      "step": 55090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5029449942953555e-05,
+      "loss": 0.3096,
+      "step": 55095
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5026840448790735e-05,
+      "loss": 0.2746,
+      "step": 55100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5024230824436e-05,
+      "loss": 0.2508,
+      "step": 55105
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.502162106992322e-05,
+      "loss": 0.2831,
+      "step": 55110
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.501901118528631e-05,
+      "loss": 0.2803,
+      "step": 55115
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.501640117055913e-05,
+      "loss": 0.2564,
+      "step": 55120
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.50137910257756e-05,
+      "loss": 0.262,
+      "step": 55125
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.501118075096958e-05,
+      "loss": 0.2641,
+      "step": 55130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.500857034617497e-05,
+      "loss": 0.2845,
+      "step": 55135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.500595981142568e-05,
+      "loss": 0.2981,
+      "step": 55140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.50033491467556e-05,
+      "loss": 0.2878,
+      "step": 55145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.500073835219861e-05,
+      "loss": 0.2686,
+      "step": 55150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4998127427788644e-05,
+      "loss": 0.28,
+      "step": 55155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4995516373559566e-05,
+      "loss": 0.2787,
+      "step": 55160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.49929051895453e-05,
+      "loss": 0.2695,
+      "step": 55165
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.499029387577974e-05,
+      "loss": 0.2726,
+      "step": 55170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4987682432296805e-05,
+      "loss": 0.2867,
+      "step": 55175
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.498507085913039e-05,
+      "loss": 0.2906,
+      "step": 55180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4982459156314415e-05,
+      "loss": 0.2819,
+      "step": 55185
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.497984732388278e-05,
+      "loss": 0.2734,
+      "step": 55190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.497723536186941e-05,
+      "loss": 0.2846,
+      "step": 55195
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4974623270308215e-05,
+      "loss": 0.3048,
+      "step": 55200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.497201104923311e-05,
+      "loss": 0.3095,
+      "step": 55205
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.496939869867801e-05,
+      "loss": 0.2599,
+      "step": 55210
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4966786218676836e-05,
+      "loss": 0.2875,
+      "step": 55215
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.496417360926352e-05,
+      "loss": 0.2867,
+      "step": 55220
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.496156087047197e-05,
+      "loss": 0.2812,
+      "step": 55225
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.495894800233612e-05,
+      "loss": 0.2859,
+      "step": 55230
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.495633500488989e-05,
+      "loss": 0.2807,
+      "step": 55235
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.495372187816722e-05,
+      "loss": 0.2775,
+      "step": 55240
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4951108622202024e-05,
+      "loss": 0.2604,
+      "step": 55245
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.494849523702824e-05,
+      "loss": 0.2818,
+      "step": 55250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.49458817226798e-05,
+      "loss": 0.2879,
+      "step": 55255
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.494326807919065e-05,
+      "loss": 0.281,
+      "step": 55260
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4940654306594705e-05,
+      "loss": 0.2658,
+      "step": 55265
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.493804040492592e-05,
+      "loss": 0.3026,
+      "step": 55270
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.493542637421823e-05,
+      "loss": 0.2766,
+      "step": 55275
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.4932812214505576e-05,
+      "loss": 0.3027,
+      "step": 55280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.49301979258219e-05,
+      "loss": 0.2719,
+      "step": 55285
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4927583508201155e-05,
+      "loss": 0.2899,
+      "step": 55290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.492496896167727e-05,
+      "loss": 0.2836,
+      "step": 55295
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.492235428628421e-05,
+      "loss": 0.2778,
+      "step": 55300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4919739482055914e-05,
+      "loss": 0.3118,
+      "step": 55305
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.491712454902635e-05,
+      "loss": 0.2832,
+      "step": 55310
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.491450948722945e-05,
+      "loss": 0.2933,
+      "step": 55315
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.491189429669918e-05,
+      "loss": 0.2857,
+      "step": 55320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.49092789774695e-05,
+      "loss": 0.2761,
+      "step": 55325
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4906663529574344e-05,
+      "loss": 0.2938,
+      "step": 55330
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490404795304771e-05,
+      "loss": 0.2615,
+      "step": 55335
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490143224792353e-05,
+      "loss": 0.2918,
+      "step": 55340
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.489881641423579e-05,
+      "loss": 0.3023,
+      "step": 55345
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.489620045201843e-05,
+      "loss": 0.304,
+      "step": 55350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.489358436130543e-05,
+      "loss": 0.296,
+      "step": 55355
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.489096814213076e-05,
+      "loss": 0.287,
+      "step": 55360
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.488835179452839e-05,
+      "loss": 0.2749,
+      "step": 55365
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.488573531853229e-05,
+      "loss": 0.2931,
+      "step": 55370
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4883118714176427e-05,
+      "loss": 0.2849,
+      "step": 55375
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.488050198149478e-05,
+      "loss": 0.2739,
+      "step": 55380
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.487788512052132e-05,
+      "loss": 0.2894,
+      "step": 55385
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4875268131290046e-05,
+      "loss": 0.2739,
+      "step": 55390
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4872651013834916e-05,
+      "loss": 0.2682,
+      "step": 55395
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.487003376818992e-05,
+      "loss": 0.2523,
+      "step": 55400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.486741639438904e-05,
+      "loss": 0.3135,
+      "step": 55405
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4864798892466256e-05,
+      "loss": 0.2938,
+      "step": 55410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4862181262455565e-05,
+      "loss": 0.263,
+      "step": 55415
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.485956350439095e-05,
+      "loss": 0.2899,
+      "step": 55420
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.48569456183064e-05,
+      "loss": 0.2736,
+      "step": 55425
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.485432760423591e-05,
+      "loss": 0.2928,
+      "step": 55430
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.485170946221347e-05,
+      "loss": 0.2833,
+      "step": 55435
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4849091192273064e-05,
+      "loss": 0.2792,
+      "step": 55440
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.484647279444872e-05,
+      "loss": 0.2718,
+      "step": 55445
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.48438542687744e-05,
+      "loss": 0.2887,
+      "step": 55450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.484123561528412e-05,
+      "loss": 0.274,
+      "step": 55455
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.483861683401188e-05,
+      "loss": 0.2888,
+      "step": 55460
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.483599792499169e-05,
+      "loss": 0.2933,
+      "step": 55465
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.483337888825755e-05,
+      "loss": 0.2908,
+      "step": 55470
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4830759723843464e-05,
+      "loss": 0.2973,
+      "step": 55475
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.482814043178344e-05,
+      "loss": 0.2742,
+      "step": 55480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.482552101211149e-05,
+      "loss": 0.285,
+      "step": 55485
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4822901464861633e-05,
+      "loss": 0.287,
+      "step": 55490
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.482028179006786e-05,
+      "loss": 0.2726,
+      "step": 55495
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.48176619877642e-05,
+      "loss": 0.2736,
+      "step": 55500
+    },
+    {
+      "epoch": 1.95,
+      "eval_loss": 0.27318820357322693,
+      "eval_runtime": 10.5335,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 55500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4815042057984683e-05,
+      "loss": 0.2924,
+      "step": 55505
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.481242200076331e-05,
+      "loss": 0.2876,
+      "step": 55510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4809801816134096e-05,
+      "loss": 0.2819,
+      "step": 55515
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4807181504131075e-05,
+      "loss": 0.2526,
+      "step": 55520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.480456106478826e-05,
+      "loss": 0.277,
+      "step": 55525
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.480194049813969e-05,
+      "loss": 0.2633,
+      "step": 55530
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.479931980421938e-05,
+      "loss": 0.2715,
+      "step": 55535
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.479669898306136e-05,
+      "loss": 0.2846,
+      "step": 55540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.479407803469966e-05,
+      "loss": 0.2837,
+      "step": 55545
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.47914569591683e-05,
+      "loss": 0.2764,
+      "step": 55550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.478883575650134e-05,
+      "loss": 0.2934,
+      "step": 55555
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.478621442673279e-05,
+      "loss": 0.2913,
+      "step": 55560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.47835929698967e-05,
+      "loss": 0.2786,
+      "step": 55565
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.47809713860271e-05,
+      "loss": 0.2932,
+      "step": 55570
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.477834967515803e-05,
+      "loss": 0.2587,
+      "step": 55575
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.477572783732354e-05,
+      "loss": 0.303,
+      "step": 55580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.477310587255766e-05,
+      "loss": 0.2909,
+      "step": 55585
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.477048378089444e-05,
+      "loss": 0.3092,
+      "step": 55590
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.476786156236794e-05,
+      "loss": 0.2843,
+      "step": 55595
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.476523921701218e-05,
+      "loss": 0.2992,
+      "step": 55600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4762616744861226e-05,
+      "loss": 0.2587,
+      "step": 55605
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.475999414594913e-05,
+      "loss": 0.273,
+      "step": 55610
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.475737142030995e-05,
+      "loss": 0.2752,
+      "step": 55615
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4754748567977725e-05,
+      "loss": 0.2721,
+      "step": 55620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4752125588986525e-05,
+      "loss": 0.2776,
+      "step": 55625
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.474950248337039e-05,
+      "loss": 0.2845,
+      "step": 55630
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4746879251163396e-05,
+      "loss": 0.2974,
+      "step": 55635
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4744255892399594e-05,
+      "loss": 0.2803,
+      "step": 55640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4741632407113064e-05,
+      "loss": 0.2813,
+      "step": 55645
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4739008795337845e-05,
+      "loss": 0.289,
+      "step": 55650
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.473638505710802e-05,
+      "loss": 0.2605,
+      "step": 55655
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4733761192457645e-05,
+      "loss": 0.2777,
+      "step": 55660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.473113720142081e-05,
+      "loss": 0.2803,
+      "step": 55665
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.472851308403156e-05,
+      "loss": 0.2543,
+      "step": 55670
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.472588884032398e-05,
+      "loss": 0.2648,
+      "step": 55675
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.472326447033214e-05,
+      "loss": 0.2898,
+      "step": 55680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4720639974090126e-05,
+      "loss": 0.2609,
+      "step": 55685
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4718015351632e-05,
+      "loss": 0.2882,
+      "step": 55690
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4715390602991855e-05,
+      "loss": 0.2933,
+      "step": 55695
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.471276572820377e-05,
+      "loss": 0.3122,
+      "step": 55700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.471014072730181e-05,
+      "loss": 0.2917,
+      "step": 55705
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.470751560032009e-05,
+      "loss": 0.2758,
+      "step": 55710
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4704890347292656e-05,
+      "loss": 0.2865,
+      "step": 55715
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.470226496825363e-05,
+      "loss": 0.2942,
+      "step": 55720
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.469963946323709e-05,
+      "loss": 0.2814,
+      "step": 55725
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.469701383227712e-05,
+      "loss": 0.2787,
+      "step": 55730
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.469438807540781e-05,
+      "loss": 0.2709,
+      "step": 55735
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.469176219266326e-05,
+      "loss": 0.2839,
+      "step": 55740
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.468913618407758e-05,
+      "loss": 0.2847,
+      "step": 55745
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.468651004968484e-05,
+      "loss": 0.2982,
+      "step": 55750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.468388378951915e-05,
+      "loss": 0.3017,
+      "step": 55755
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.468125740361461e-05,
+      "loss": 0.3006,
+      "step": 55760
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4678630892005334e-05,
+      "loss": 0.2897,
+      "step": 55765
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.46760042547254e-05,
+      "loss": 0.3039,
+      "step": 55770
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4673377491808935e-05,
+      "loss": 0.2681,
+      "step": 55775
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.467075060329004e-05,
+      "loss": 0.2855,
+      "step": 55780
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466812358920282e-05,
+      "loss": 0.2613,
+      "step": 55785
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466549644958138e-05,
+      "loss": 0.2904,
+      "step": 55790
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466286918445985e-05,
+      "loss": 0.2838,
+      "step": 55795
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466024179387232e-05,
+      "loss": 0.2562,
+      "step": 55800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4657614277852926e-05,
+      "loss": 0.2805,
+      "step": 55805
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.465498663643576e-05,
+      "loss": 0.2851,
+      "step": 55810
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.465235886965498e-05,
+      "loss": 0.279,
+      "step": 55815
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.464973097754466e-05,
+      "loss": 0.2746,
+      "step": 55820
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.464710296013895e-05,
+      "loss": 0.2792,
+      "step": 55825
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4644474817471964e-05,
+      "loss": 0.2919,
+      "step": 55830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4641846549577836e-05,
+      "loss": 0.2743,
+      "step": 55835
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.463921815649068e-05,
+      "loss": 0.2919,
+      "step": 55840
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4636589638244635e-05,
+      "loss": 0.263,
+      "step": 55845
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.463396099487381e-05,
+      "loss": 0.3008,
+      "step": 55850
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.463133222641236e-05,
+      "loss": 0.2541,
+      "step": 55855
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.462870333289441e-05,
+      "loss": 0.2645,
+      "step": 55860
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.462607431435409e-05,
+      "loss": 0.283,
+      "step": 55865
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.462344517082553e-05,
+      "loss": 0.3032,
+      "step": 55870
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.462081590234289e-05,
+      "loss": 0.2615,
+      "step": 55875
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4618186508940283e-05,
+      "loss": 0.2786,
+      "step": 55880
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.461555699065188e-05,
+      "loss": 0.2817,
+      "step": 55885
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.461292734751179e-05,
+      "loss": 0.2885,
+      "step": 55890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.461029757955419e-05,
+      "loss": 0.3076,
+      "step": 55895
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4607667686813196e-05,
+      "loss": 0.2632,
+      "step": 55900
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.460503766932297e-05,
+      "loss": 0.2721,
+      "step": 55905
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.460240752711766e-05,
+      "loss": 0.2715,
+      "step": 55910
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.459977726023142e-05,
+      "loss": 0.2827,
+      "step": 55915
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.45971468686984e-05,
+      "loss": 0.3024,
+      "step": 55920
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.459451635255275e-05,
+      "loss": 0.2799,
+      "step": 55925
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.459188571182863e-05,
+      "loss": 0.2953,
+      "step": 55930
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.458925494656019e-05,
+      "loss": 0.2639,
+      "step": 55935
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.45866240567816e-05,
+      "loss": 0.2732,
+      "step": 55940
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.458399304252701e-05,
+      "loss": 0.2735,
+      "step": 55945
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.458136190383059e-05,
+      "loss": 0.266,
+      "step": 55950
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4578730640726496e-05,
+      "loss": 0.2605,
+      "step": 55955
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4576099253248905e-05,
+      "loss": 0.2745,
+      "step": 55960
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.457346774143197e-05,
+      "loss": 0.2644,
+      "step": 55965
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.457083610530987e-05,
+      "loss": 0.2862,
+      "step": 55970
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4568204344916766e-05,
+      "loss": 0.2671,
+      "step": 55975
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4565572460286835e-05,
+      "loss": 0.2652,
+      "step": 55980
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.456294045145425e-05,
+      "loss": 0.2871,
+      "step": 55985
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.456030831845318e-05,
+      "loss": 0.2856,
+      "step": 55990
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.455767606131781e-05,
+      "loss": 0.3036,
+      "step": 55995
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.455504368008232e-05,
+      "loss": 0.3041,
+      "step": 56000
+    },
+    {
+      "epoch": 1.97,
+      "eval_loss": 0.27226722240448,
+      "eval_runtime": 10.5443,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 56000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4552411174780886e-05,
+      "loss": 0.285,
+      "step": 56005
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4549778545447685e-05,
+      "loss": 0.3094,
+      "step": 56010
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.45471457921169e-05,
+      "loss": 0.2706,
+      "step": 56015
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.454451291482272e-05,
+      "loss": 0.268,
+      "step": 56020
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.454187991359933e-05,
+      "loss": 0.2826,
+      "step": 56025
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.453924678848093e-05,
+      "loss": 0.2784,
+      "step": 56030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4536613539501684e-05,
+      "loss": 0.2817,
+      "step": 56035
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.453398016669579e-05,
+      "loss": 0.2681,
+      "step": 56040
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.453134667009745e-05,
+      "loss": 0.3129,
+      "step": 56045
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.452871304974086e-05,
+      "loss": 0.292,
+      "step": 56050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.452607930566022e-05,
+      "loss": 0.2589,
+      "step": 56055
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.45234454378897e-05,
+      "loss": 0.2759,
+      "step": 56060
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.452081144646353e-05,
+      "loss": 0.3165,
+      "step": 56065
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.451817733141589e-05,
+      "loss": 0.2685,
+      "step": 56070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4515543092781e-05,
+      "loss": 0.29,
+      "step": 56075
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4512908730593044e-05,
+      "loss": 0.2866,
+      "step": 56080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.451027424488624e-05,
+      "loss": 0.2695,
+      "step": 56085
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.450763963569479e-05,
+      "loss": 0.2717,
+      "step": 56090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4505004903052915e-05,
+      "loss": 0.2972,
+      "step": 56095
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4502370046994806e-05,
+      "loss": 0.2862,
+      "step": 56100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4499735067554686e-05,
+      "loss": 0.2698,
+      "step": 56105
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.449709996476677e-05,
+      "loss": 0.2861,
+      "step": 56110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.449446473866527e-05,
+      "loss": 0.2772,
+      "step": 56115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4491829389284396e-05,
+      "loss": 0.2967,
+      "step": 56120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4489193916658374e-05,
+      "loss": 0.2691,
+      "step": 56125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.448655832082143e-05,
+      "loss": 0.2875,
+      "step": 56130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.448392260180777e-05,
+      "loss": 0.3062,
+      "step": 56135
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.448128675965163e-05,
+      "loss": 0.2866,
+      "step": 56140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.447865079438723e-05,
+      "loss": 0.295,
+      "step": 56145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.447601470604879e-05,
+      "loss": 0.2987,
+      "step": 56150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.447337849467055e-05,
+      "loss": 0.2784,
+      "step": 56155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.447074216028673e-05,
+      "loss": 0.2711,
+      "step": 56160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4468105702931565e-05,
+      "loss": 0.2909,
+      "step": 56165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.446546912263929e-05,
+      "loss": 0.276,
+      "step": 56170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4462832419444135e-05,
+      "loss": 0.3056,
+      "step": 56175
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.446019559338034e-05,
+      "loss": 0.2793,
+      "step": 56180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.445755864448213e-05,
+      "loss": 0.2806,
+      "step": 56185
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.445492157278377e-05,
+      "loss": 0.288,
+      "step": 56190
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4452284378319465e-05,
+      "loss": 0.2845,
+      "step": 56195
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.444964706112349e-05,
+      "loss": 0.2675,
+      "step": 56200
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.444700962123007e-05,
+      "loss": 0.2838,
+      "step": 56205
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.444437205867346e-05,
+      "loss": 0.2862,
+      "step": 56210
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.44417343734879e-05,
+      "loss": 0.2953,
+      "step": 56215
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4439096565707647e-05,
+      "loss": 0.2968,
+      "step": 56220
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4436458635366935e-05,
+      "loss": 0.295,
+      "step": 56225
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4433820582500035e-05,
+      "loss": 0.2711,
+      "step": 56230
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4431182407141186e-05,
+      "loss": 0.2735,
+      "step": 56235
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4428544109324656e-05,
+      "loss": 0.2823,
+      "step": 56240
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.442590568908469e-05,
+      "loss": 0.2839,
+      "step": 56245
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.442326714645555e-05,
+      "loss": 0.2897,
+      "step": 56250
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.44206284814715e-05,
+      "loss": 0.2949,
+      "step": 56255
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.441798969416679e-05,
+      "loss": 0.2782,
+      "step": 56260
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.44153507845757e-05,
+      "loss": 0.2896,
+      "step": 56265
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.441271175273248e-05,
+      "loss": 0.2937,
+      "step": 56270
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.44100725986714e-05,
+      "loss": 0.2904,
+      "step": 56275
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.440743332242673e-05,
+      "loss": 0.3106,
+      "step": 56280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.440479392403274e-05,
+      "loss": 0.2914,
+      "step": 56285
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.440215440352369e-05,
+      "loss": 0.2849,
+      "step": 56290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.439951476093387e-05,
+      "loss": 0.3019,
+      "step": 56295
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4396874996297545e-05,
+      "loss": 0.2609,
+      "step": 56300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4394235109649e-05,
+      "loss": 0.3112,
+      "step": 56305
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4391595101022486e-05,
+      "loss": 0.2749,
+      "step": 56310
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.43889549704523e-05,
+      "loss": 0.272,
+      "step": 56315
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.438631471797273e-05,
+      "loss": 0.2844,
+      "step": 56320
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.438367434361805e-05,
+      "loss": 0.2786,
+      "step": 56325
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.438103384742254e-05,
+      "loss": 0.2764,
+      "step": 56330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.437839322942049e-05,
+      "loss": 0.3176,
+      "step": 56335
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.437575248964618e-05,
+      "loss": 0.2639,
+      "step": 56340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4373111628133906e-05,
+      "loss": 0.287,
+      "step": 56345
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.437047064491795e-05,
+      "loss": 0.289,
+      "step": 56350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.436782954003261e-05,
+      "loss": 0.2841,
+      "step": 56355
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.436518831351218e-05,
+      "loss": 0.2736,
+      "step": 56360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.436254696539095e-05,
+      "loss": 0.2837,
+      "step": 56365
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.435990549570321e-05,
+      "loss": 0.2807,
+      "step": 56370
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.435726390448328e-05,
+      "loss": 0.2494,
+      "step": 56375
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.435462219176543e-05,
+      "loss": 0.2607,
+      "step": 56380
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.435198035758399e-05,
+      "loss": 0.2763,
+      "step": 56385
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.434933840197324e-05,
+      "loss": 0.3139,
+      "step": 56390
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4346696324967494e-05,
+      "loss": 0.2886,
+      "step": 56395
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.434405412660106e-05,
+      "loss": 0.2772,
+      "step": 56400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.434141180690823e-05,
+      "loss": 0.2993,
+      "step": 56405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4338769365923335e-05,
+      "loss": 0.3046,
+      "step": 56410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4336126803680665e-05,
+      "loss": 0.2795,
+      "step": 56415
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.433348412021456e-05,
+      "loss": 0.2812,
+      "step": 56420
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.433084131555929e-05,
+      "loss": 0.2734,
+      "step": 56425
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.43281983897492e-05,
+      "loss": 0.2761,
+      "step": 56430
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4325555342818614e-05,
+      "loss": 0.2884,
+      "step": 56435
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.432291217480183e-05,
+      "loss": 0.2656,
+      "step": 56440
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4320268885733174e-05,
+      "loss": 0.2916,
+      "step": 56445
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.431762547564698e-05,
+      "loss": 0.2979,
+      "step": 56450
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.431498194457755e-05,
+      "loss": 0.2739,
+      "step": 56455
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4312338292559214e-05,
+      "loss": 0.2616,
+      "step": 56460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.430969451962631e-05,
+      "loss": 0.2783,
+      "step": 56465
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.430705062581316e-05,
+      "loss": 0.2749,
+      "step": 56470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4304406611154085e-05,
+      "loss": 0.2792,
+      "step": 56475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.430176247568342e-05,
+      "loss": 0.2631,
+      "step": 56480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.42991182194355e-05,
+      "loss": 0.2848,
+      "step": 56485
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4296473842444663e-05,
+      "loss": 0.2854,
+      "step": 56490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.429382934474524e-05,
+      "loss": 0.2827,
+      "step": 56495
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.429118472637157e-05,
+      "loss": 0.2898,
+      "step": 56500
+    },
+    {
+      "epoch": 1.99,
+      "eval_loss": 0.2718130946159363,
+      "eval_runtime": 10.5492,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 56500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4288539987357984e-05,
+      "loss": 0.2907,
+      "step": 56505
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.428589512773882e-05,
+      "loss": 0.2519,
+      "step": 56510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.428325014754844e-05,
+      "loss": 0.2855,
+      "step": 56515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4280605046821175e-05,
+      "loss": 0.2906,
+      "step": 56520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4277959825591366e-05,
+      "loss": 0.2636,
+      "step": 56525
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.427531448389336e-05,
+      "loss": 0.2877,
+      "step": 56530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.427266902176151e-05,
+      "loss": 0.2604,
+      "step": 56535
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4270023439230166e-05,
+      "loss": 0.2789,
+      "step": 56540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4267377736333676e-05,
+      "loss": 0.3074,
+      "step": 56545
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.42647319131064e-05,
+      "loss": 0.2577,
+      "step": 56550
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4262085969582674e-05,
+      "loss": 0.2948,
+      "step": 56555
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.425943990579687e-05,
+      "loss": 0.2721,
+      "step": 56560
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4256793721783345e-05,
+      "loss": 0.3173,
+      "step": 56565
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4254147417576445e-05,
+      "loss": 0.2765,
+      "step": 56570
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.425150099321055e-05,
+      "loss": 0.2899,
+      "step": 56575
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4248854448720004e-05,
+      "loss": 0.2773,
+      "step": 56580
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.424620778413918e-05,
+      "loss": 0.2734,
+      "step": 56585
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4243560999502445e-05,
+      "loss": 0.2944,
+      "step": 56590
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4240914094844154e-05,
+      "loss": 0.2746,
+      "step": 56595
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4238267070198686e-05,
+      "loss": 0.2693,
+      "step": 56600
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4235619925600417e-05,
+      "loss": 0.2708,
+      "step": 56605
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.42329726610837e-05,
+      "loss": 0.28,
+      "step": 56610
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4230325276682924e-05,
+      "loss": 0.2826,
+      "step": 56615
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.422767777243245e-05,
+      "loss": 0.294,
+      "step": 56620
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4225030148366664e-05,
+      "loss": 0.2748,
+      "step": 56625
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.422238240451995e-05,
+      "loss": 0.2786,
+      "step": 56630
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4219734540926665e-05,
+      "loss": 0.2295,
+      "step": 56635
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.421708655762122e-05,
+      "loss": 0.2852,
+      "step": 56640
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.421443845463797e-05,
+      "loss": 0.2934,
+      "step": 56645
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4211790232011305e-05,
+      "loss": 0.2506,
+      "step": 56650
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4209141889775615e-05,
+      "loss": 0.265,
+      "step": 56655
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.42064934279653e-05,
+      "loss": 0.3019,
+      "step": 56660
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4203844846614716e-05,
+      "loss": 0.2981,
+      "step": 56665
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4201196145758286e-05,
+      "loss": 0.2878,
+      "step": 56670
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.419854732543038e-05,
+      "loss": 0.2791,
+      "step": 56675
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.41958983856654e-05,
+      "loss": 0.2633,
+      "step": 56680
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.4193249326497744e-05,
+      "loss": 0.2844,
+      "step": 56685
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.419060014796181e-05,
+      "loss": 0.3059,
+      "step": 56690
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.418795085009198e-05,
+      "loss": 0.2673,
+      "step": 56695
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.418530143292267e-05,
+      "loss": 0.288,
+      "step": 56700
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.418265189648826e-05,
+      "loss": 0.2726,
+      "step": 56705
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.418000224082318e-05,
+      "loss": 0.2807,
+      "step": 56710
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.417735246596182e-05,
+      "loss": 0.2939,
+      "step": 56715
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.417470257193859e-05,
+      "loss": 0.297,
+      "step": 56720
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.417205255878788e-05,
+      "loss": 0.2763,
+      "step": 56725
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.416940242654412e-05,
+      "loss": 0.2568,
+      "step": 56730
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.416675217524171e-05,
+      "loss": 0.2837,
+      "step": 56735
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4164101804915064e-05,
+      "loss": 0.2868,
+      "step": 56740
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.41614513155986e-05,
+      "loss": 0.2832,
+      "step": 56745
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4158800707326723e-05,
+      "loss": 0.2483,
+      "step": 56750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4156680135084465e-05,
+      "loss": 0.2854,
+      "step": 56755
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.415402931277958e-05,
+      "loss": 0.2904,
+      "step": 56760
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.415137837161566e-05,
+      "loss": 0.2945,
+      "step": 56765
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.414872731162713e-05,
+      "loss": 0.2769,
+      "step": 56770
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4146076132848406e-05,
+      "loss": 0.2743,
+      "step": 56775
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.414342483531391e-05,
+      "loss": 0.2749,
+      "step": 56780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.414077341905806e-05,
+      "loss": 0.296,
+      "step": 56785
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.41381218841153e-05,
+      "loss": 0.2932,
+      "step": 56790
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.413547023052004e-05,
+      "loss": 0.2935,
+      "step": 56795
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4132818458306735e-05,
+      "loss": 0.2933,
+      "step": 56800
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4130166567509785e-05,
+      "loss": 0.284,
+      "step": 56805
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4127514558163656e-05,
+      "loss": 0.2524,
+      "step": 56810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.412486243030275e-05,
+      "loss": 0.3027,
+      "step": 56815
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.412221018396152e-05,
+      "loss": 0.2836,
+      "step": 56820
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4119557819174406e-05,
+      "loss": 0.2806,
+      "step": 56825
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.411690533597585e-05,
+      "loss": 0.2758,
+      "step": 56830
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4114252734400274e-05,
+      "loss": 0.28,
+      "step": 56835
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.411160001448214e-05,
+      "loss": 0.3095,
+      "step": 56840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4108947176255884e-05,
+      "loss": 0.3184,
+      "step": 56845
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4106294219755955e-05,
+      "loss": 0.2491,
+      "step": 56850
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.410364114501679e-05,
+      "loss": 0.2714,
+      "step": 56855
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.410098795207285e-05,
+      "loss": 0.2563,
+      "step": 56860
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4098334640958584e-05,
+      "loss": 0.288,
+      "step": 56865
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.409568121170843e-05,
+      "loss": 0.2888,
+      "step": 56870
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.409302766435685e-05,
+      "loss": 0.2673,
+      "step": 56875
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.40903739989383e-05,
+      "loss": 0.2644,
+      "step": 56880
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.408772021548724e-05,
+      "loss": 0.2681,
+      "step": 56885
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.408506631403811e-05,
+      "loss": 0.2682,
+      "step": 56890
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.408241229462539e-05,
+      "loss": 0.2903,
+      "step": 56895
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4079758157283526e-05,
+      "loss": 0.2452,
+      "step": 56900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.407710390204699e-05,
+      "loss": 0.272,
+      "step": 56905
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.407444952895025e-05,
+      "loss": 0.2846,
+      "step": 56910
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4071795038027755e-05,
+      "loss": 0.2791,
+      "step": 56915
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.406914042931398e-05,
+      "loss": 0.267,
+      "step": 56920
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4066485702843385e-05,
+      "loss": 0.2816,
+      "step": 56925
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.406383085865046e-05,
+      "loss": 0.2655,
+      "step": 56930
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.406117589676966e-05,
+      "loss": 0.2744,
+      "step": 56935
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.405852081723547e-05,
+      "loss": 0.2504,
+      "step": 56940
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4055865620082346e-05,
+      "loss": 0.2855,
+      "step": 56945
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.405321030534479e-05,
+      "loss": 0.2821,
+      "step": 56950
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4050554873057245e-05,
+      "loss": 0.2763,
+      "step": 56955
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.404789932325422e-05,
+      "loss": 0.2822,
+      "step": 56960
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.404524365597018e-05,
+      "loss": 0.2653,
+      "step": 56965
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.404258787123962e-05,
+      "loss": 0.2609,
+      "step": 56970
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.4039931969097006e-05,
+      "loss": 0.2755,
+      "step": 56975
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.403727594957684e-05,
+      "loss": 0.2875,
+      "step": 56980
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.40346198127136e-05,
+      "loss": 0.2825,
+      "step": 56985
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.403196355854178e-05,
+      "loss": 0.2678,
+      "step": 56990
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.402930718709586e-05,
+      "loss": 0.2585,
+      "step": 56995
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.402665069841034e-05,
+      "loss": 0.2683,
+      "step": 57000
+    },
+    {
+      "epoch": 2.01,
+      "eval_loss": 0.271958589553833,
+      "eval_runtime": 10.5499,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 57000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.402399409251971e-05,
+      "loss": 0.2793,
+      "step": 57005
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.402133736945846e-05,
+      "loss": 0.2684,
+      "step": 57010
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.401868052926108e-05,
+      "loss": 0.2791,
+      "step": 57015
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.40160235719621e-05,
+      "loss": 0.2628,
+      "step": 57020
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.401336649759598e-05,
+      "loss": 0.2844,
+      "step": 57025
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.401070930619724e-05,
+      "loss": 0.2726,
+      "step": 57030
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.400805199780037e-05,
+      "loss": 0.2613,
+      "step": 57035
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.400539457243989e-05,
+      "loss": 0.2656,
+      "step": 57040
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.4002737030150294e-05,
+      "loss": 0.2687,
+      "step": 57045
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.400007937096609e-05,
+      "loss": 0.2662,
+      "step": 57050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3997421594921795e-05,
+      "loss": 0.2857,
+      "step": 57055
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.39947637020519e-05,
+      "loss": 0.2797,
+      "step": 57060
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.399210569239093e-05,
+      "loss": 0.2614,
+      "step": 57065
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.398944756597339e-05,
+      "loss": 0.282,
+      "step": 57070
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.398678932283381e-05,
+      "loss": 0.2605,
+      "step": 57075
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.398413096300668e-05,
+      "loss": 0.2592,
+      "step": 57080
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.398147248652653e-05,
+      "loss": 0.2464,
+      "step": 57085
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3978813893427886e-05,
+      "loss": 0.2636,
+      "step": 57090
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.397615518374526e-05,
+      "loss": 0.2526,
+      "step": 57095
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.397349635751317e-05,
+      "loss": 0.2737,
+      "step": 57100
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.397083741476614e-05,
+      "loss": 0.2877,
+      "step": 57105
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.396817835553871e-05,
+      "loss": 0.2706,
+      "step": 57110
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.396551917986539e-05,
+      "loss": 0.2725,
+      "step": 57115
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.39628598877807e-05,
+      "loss": 0.2833,
+      "step": 57120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.39602004793192e-05,
+      "loss": 0.2687,
+      "step": 57125
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.395754095451539e-05,
+      "loss": 0.2735,
+      "step": 57130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.395488131340382e-05,
+      "loss": 0.2931,
+      "step": 57135
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.395222155601902e-05,
+      "loss": 0.2854,
+      "step": 57140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.394956168239551e-05,
+      "loss": 0.2831,
+      "step": 57145
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3946901692567844e-05,
+      "loss": 0.2905,
+      "step": 57150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3944241586570565e-05,
+      "loss": 0.2555,
+      "step": 57155
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.394158136443819e-05,
+      "loss": 0.2624,
+      "step": 57160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.393892102620528e-05,
+      "loss": 0.2626,
+      "step": 57165
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.393626057190638e-05,
+      "loss": 0.28,
+      "step": 57170
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.393360000157601e-05,
+      "loss": 0.2743,
+      "step": 57175
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.393093931524875e-05,
+      "loss": 0.2711,
+      "step": 57180
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.392827851295911e-05,
+      "loss": 0.2572,
+      "step": 57185
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.392561759474167e-05,
+      "loss": 0.2988,
+      "step": 57190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.392295656063096e-05,
+      "loss": 0.2747,
+      "step": 57195
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.392029541066154e-05,
+      "loss": 0.2537,
+      "step": 57200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3917634144867965e-05,
+      "loss": 0.2639,
+      "step": 57205
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3914972763284794e-05,
+      "loss": 0.2387,
+      "step": 57210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.391231126594657e-05,
+      "loss": 0.3002,
+      "step": 57215
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.390964965288786e-05,
+      "loss": 0.2646,
+      "step": 57220
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.390698792414322e-05,
+      "loss": 0.277,
+      "step": 57225
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.390432607974721e-05,
+      "loss": 0.2545,
+      "step": 57230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.390166411973439e-05,
+      "loss": 0.2576,
+      "step": 57235
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.389900204413934e-05,
+      "loss": 0.2717,
+      "step": 57240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3896339852996604e-05,
+      "loss": 0.2625,
+      "step": 57245
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3893677546340765e-05,
+      "loss": 0.2338,
+      "step": 57250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3891015124206376e-05,
+      "loss": 0.2643,
+      "step": 57255
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.388835258662802e-05,
+      "loss": 0.2568,
+      "step": 57260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.388568993364026e-05,
+      "loss": 0.2942,
+      "step": 57265
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.388302716527768e-05,
+      "loss": 0.2486,
+      "step": 57270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3880364281574836e-05,
+      "loss": 0.2601,
+      "step": 57275
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3877701282566314e-05,
+      "loss": 0.2685,
+      "step": 57280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.38750381682867e-05,
+      "loss": 0.255,
+      "step": 57285
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.387237493877056e-05,
+      "loss": 0.2693,
+      "step": 57290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.386971159405248e-05,
+      "loss": 0.2861,
+      "step": 57295
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3867048134167035e-05,
+      "loss": 0.2617,
+      "step": 57300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.386438455914882e-05,
+      "loss": 0.2772,
+      "step": 57305
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.386172086903241e-05,
+      "loss": 0.2497,
+      "step": 57310
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.38590570638524e-05,
+      "loss": 0.2822,
+      "step": 57315
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.385639314364336e-05,
+      "loss": 0.2675,
+      "step": 57320
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.38537291084399e-05,
+      "loss": 0.2926,
+      "step": 57325
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.38510649582766e-05,
+      "loss": 0.2611,
+      "step": 57330
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.384840069318807e-05,
+      "loss": 0.2586,
+      "step": 57335
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.384573631320886e-05,
+      "loss": 0.2682,
+      "step": 57340
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3843071818373617e-05,
+      "loss": 0.2768,
+      "step": 57345
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.38404072087169e-05,
+      "loss": 0.26,
+      "step": 57350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.383774248427333e-05,
+      "loss": 0.2657,
+      "step": 57355
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.38350776450775e-05,
+      "loss": 0.2644,
+      "step": 57360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3832412691164e-05,
+      "loss": 0.2567,
+      "step": 57365
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.382974762256744e-05,
+      "loss": 0.2573,
+      "step": 57370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3827082439322445e-05,
+      "loss": 0.2531,
+      "step": 57375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.382441714146358e-05,
+      "loss": 0.2777,
+      "step": 57380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.382175172902549e-05,
+      "loss": 0.2603,
+      "step": 57385
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.381908620204276e-05,
+      "loss": 0.2681,
+      "step": 57390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3816420560550004e-05,
+      "loss": 0.2705,
+      "step": 57395
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3813754804581844e-05,
+      "loss": 0.2606,
+      "step": 57400
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.381108893417288e-05,
+      "loss": 0.2762,
+      "step": 57405
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3808422949357746e-05,
+      "loss": 0.2848,
+      "step": 57410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3805756850171035e-05,
+      "loss": 0.2976,
+      "step": 57415
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.380309063664737e-05,
+      "loss": 0.2616,
+      "step": 57420
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.380042430882138e-05,
+      "loss": 0.2677,
+      "step": 57425
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.379775786672768e-05,
+      "loss": 0.2842,
+      "step": 57430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3795091310400895e-05,
+      "loss": 0.2763,
+      "step": 57435
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.379242463987564e-05,
+      "loss": 0.2363,
+      "step": 57440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.378975785518655e-05,
+      "loss": 0.2622,
+      "step": 57445
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3787090956368244e-05,
+      "loss": 0.2869,
+      "step": 57450
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3784423943455354e-05,
+      "loss": 0.2896,
+      "step": 57455
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.378175681648251e-05,
+      "loss": 0.2588,
+      "step": 57460
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3779089575484335e-05,
+      "loss": 0.249,
+      "step": 57465
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.377642222049548e-05,
+      "loss": 0.266,
+      "step": 57470
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3773754751550554e-05,
+      "loss": 0.279,
+      "step": 57475
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.377108716868421e-05,
+      "loss": 0.261,
+      "step": 57480
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.376841947193108e-05,
+      "loss": 0.2634,
+      "step": 57485
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.37657516613258e-05,
+      "loss": 0.279,
+      "step": 57490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.376308373690301e-05,
+      "loss": 0.2691,
+      "step": 57495
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.376041569869735e-05,
+      "loss": 0.2587,
+      "step": 57500
+    },
+    {
+      "epoch": 2.02,
+      "eval_loss": 0.27126699686050415,
+      "eval_runtime": 10.5272,
+      "eval_samples_per_second": 9.499,
+      "eval_steps_per_second": 9.499,
+      "step": 57500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.375774754674347e-05,
+      "loss": 0.2689,
+      "step": 57505
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.375507928107601e-05,
+      "loss": 0.2951,
+      "step": 57510
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.375241090172961e-05,
+      "loss": 0.2765,
+      "step": 57515
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3749742408738935e-05,
+      "loss": 0.2652,
+      "step": 57520
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3747073802138616e-05,
+      "loss": 0.2837,
+      "step": 57525
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.37444050819633e-05,
+      "loss": 0.2809,
+      "step": 57530
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.374173624824765e-05,
+      "loss": 0.2927,
+      "step": 57535
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.373906730102632e-05,
+      "loss": 0.2504,
+      "step": 57540
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.373639824033396e-05,
+      "loss": 0.2715,
+      "step": 57545
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.3733729066205225e-05,
+      "loss": 0.2624,
+      "step": 57550
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.373105977867478e-05,
+      "loss": 0.2941,
+      "step": 57555
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3728390377777263e-05,
+      "loss": 0.2496,
+      "step": 57560
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.372572086354736e-05,
+      "loss": 0.2969,
+      "step": 57565
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.372305123601972e-05,
+      "loss": 0.3003,
+      "step": 57570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.372038149522901e-05,
+      "loss": 0.2764,
+      "step": 57575
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.371771164120989e-05,
+      "loss": 0.2714,
+      "step": 57580
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3715041673997035e-05,
+      "loss": 0.2711,
+      "step": 57585
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.37123715936251e-05,
+      "loss": 0.2781,
+      "step": 57590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3709701400128765e-05,
+      "loss": 0.2579,
+      "step": 57595
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.370703109354269e-05,
+      "loss": 0.2673,
+      "step": 57600
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.370436067390157e-05,
+      "loss": 0.2843,
+      "step": 57605
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3701690141240056e-05,
+      "loss": 0.2998,
+      "step": 57610
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.369901949559283e-05,
+      "loss": 0.2908,
+      "step": 57615
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.369634873699457e-05,
+      "loss": 0.276,
+      "step": 57620
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3693677865479945e-05,
+      "loss": 0.2981,
+      "step": 57625
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.369100688108365e-05,
+      "loss": 0.2664,
+      "step": 57630
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3688335783840355e-05,
+      "loss": 0.2672,
+      "step": 57635
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.368566457378474e-05,
+      "loss": 0.2837,
+      "step": 57640
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.36829932509515e-05,
+      "loss": 0.26,
+      "step": 57645
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.368032181537532e-05,
+      "loss": 0.2865,
+      "step": 57650
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3677650267090875e-05,
+      "loss": 0.3003,
+      "step": 57655
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3674978606132866e-05,
+      "loss": 0.2764,
+      "step": 57660
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3672306832535974e-05,
+      "loss": 0.2719,
+      "step": 57665
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.366963494633489e-05,
+      "loss": 0.269,
+      "step": 57670
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.366696294756431e-05,
+      "loss": 0.2546,
+      "step": 57675
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3664290836258934e-05,
+      "loss": 0.2789,
+      "step": 57680
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3661618612453446e-05,
+      "loss": 0.2519,
+      "step": 57685
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3658946276182556e-05,
+      "loss": 0.2899,
+      "step": 57690
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3656273827480944e-05,
+      "loss": 0.2752,
+      "step": 57695
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3653601266383326e-05,
+      "loss": 0.2817,
+      "step": 57700
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3650928592924395e-05,
+      "loss": 0.2655,
+      "step": 57705
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3648255807138866e-05,
+      "loss": 0.2676,
+      "step": 57710
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.364558290906143e-05,
+      "loss": 0.2963,
+      "step": 57715
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.364290989872679e-05,
+      "loss": 0.2762,
+      "step": 57720
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.364023677616967e-05,
+      "loss": 0.2767,
+      "step": 57725
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.363756354142478e-05,
+      "loss": 0.2809,
+      "step": 57730
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.36348901945268e-05,
+      "loss": 0.2566,
+      "step": 57735
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.363221673551047e-05,
+      "loss": 0.285,
+      "step": 57740
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3629543164410496e-05,
+      "loss": 0.247,
+      "step": 57745
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3626869481261586e-05,
+      "loss": 0.2778,
+      "step": 57750
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3624195686098465e-05,
+      "loss": 0.2722,
+      "step": 57755
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.362152177895584e-05,
+      "loss": 0.2861,
+      "step": 57760
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3618847759868445e-05,
+      "loss": 0.2812,
+      "step": 57765
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.361617362887099e-05,
+      "loss": 0.252,
+      "step": 57770
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3613499385998196e-05,
+      "loss": 0.2673,
+      "step": 57775
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3610825031284784e-05,
+      "loss": 0.2857,
+      "step": 57780
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.36081505647655e-05,
+      "loss": 0.2488,
+      "step": 57785
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.360547598647504e-05,
+      "loss": 0.2621,
+      "step": 57790
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3602801296448154e-05,
+      "loss": 0.2739,
+      "step": 57795
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3600126494719555e-05,
+      "loss": 0.269,
+      "step": 57800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.359745158132399e-05,
+      "loss": 0.2596,
+      "step": 57805
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.359477655629617e-05,
+      "loss": 0.3004,
+      "step": 57810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.359210141967085e-05,
+      "loss": 0.2488,
+      "step": 57815
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.3589426171482754e-05,
+      "loss": 0.2864,
+      "step": 57820
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.358675081176662e-05,
+      "loss": 0.2883,
+      "step": 57825
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.358407534055718e-05,
+      "loss": 0.2591,
+      "step": 57830
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.358139975788918e-05,
+      "loss": 0.2733,
+      "step": 57835
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.357872406379736e-05,
+      "loss": 0.2702,
+      "step": 57840
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.357604825831647e-05,
+      "loss": 0.2588,
+      "step": 57845
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.357337234148124e-05,
+      "loss": 0.2755,
+      "step": 57850
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.357069631332642e-05,
+      "loss": 0.2558,
+      "step": 57855
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.356802017388675e-05,
+      "loss": 0.2694,
+      "step": 57860
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.356534392319699e-05,
+      "loss": 0.2742,
+      "step": 57865
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.356266756129189e-05,
+      "loss": 0.2663,
+      "step": 57870
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.355999108820619e-05,
+      "loss": 0.2779,
+      "step": 57875
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3557314503974646e-05,
+      "loss": 0.3263,
+      "step": 57880
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.355463780863201e-05,
+      "loss": 0.2674,
+      "step": 57885
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.355196100221304e-05,
+      "loss": 0.2731,
+      "step": 57890
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3549284084752494e-05,
+      "loss": 0.2736,
+      "step": 57895
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.354660705628514e-05,
+      "loss": 0.2806,
+      "step": 57900
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.354392991684571e-05,
+      "loss": 0.2514,
+      "step": 57905
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.354125266646899e-05,
+      "loss": 0.2987,
+      "step": 57910
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.353857530518972e-05,
+      "loss": 0.2642,
+      "step": 57915
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.353589783304269e-05,
+      "loss": 0.258,
+      "step": 57920
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3533220250062646e-05,
+      "loss": 0.2822,
+      "step": 57925
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3530542556284363e-05,
+      "loss": 0.2591,
+      "step": 57930
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.352786475174261e-05,
+      "loss": 0.2674,
+      "step": 57935
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3525186836472144e-05,
+      "loss": 0.2903,
+      "step": 57940
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3522508810507746e-05,
+      "loss": 0.2626,
+      "step": 57945
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.35198306738842e-05,
+      "loss": 0.2595,
+      "step": 57950
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.351715242663626e-05,
+      "loss": 0.2622,
+      "step": 57955
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.351447406879871e-05,
+      "loss": 0.2506,
+      "step": 57960
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.351179560040632e-05,
+      "loss": 0.2667,
+      "step": 57965
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.350911702149388e-05,
+      "loss": 0.2623,
+      "step": 57970
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.350643833209616e-05,
+      "loss": 0.285,
+      "step": 57975
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.350375953224794e-05,
+      "loss": 0.2561,
+      "step": 57980
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3501080621984013e-05,
+      "loss": 0.2873,
+      "step": 57985
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3498401601339156e-05,
+      "loss": 0.2427,
+      "step": 57990
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.349572247034816e-05,
+      "loss": 0.2907,
+      "step": 57995
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.34930432290458e-05,
+      "loss": 0.3012,
+      "step": 58000
+    },
+    {
+      "epoch": 2.04,
+      "eval_loss": 0.2711896598339081,
+      "eval_runtime": 10.5364,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 58000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.349036387746687e-05,
+      "loss": 0.2664,
+      "step": 58005
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.348768441564616e-05,
+      "loss": 0.2721,
+      "step": 58010
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.348500484361846e-05,
+      "loss": 0.2644,
+      "step": 58015
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.348232516141857e-05,
+      "loss": 0.2656,
+      "step": 58020
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.347964536908128e-05,
+      "loss": 0.2686,
+      "step": 58025
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.347696546664137e-05,
+      "loss": 0.2754,
+      "step": 58030
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.347428545413366e-05,
+      "loss": 0.2692,
+      "step": 58035
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3471605331592935e-05,
+      "loss": 0.2756,
+      "step": 58040
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.346892509905401e-05,
+      "loss": 0.2624,
+      "step": 58045
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3466244756551656e-05,
+      "loss": 0.2684,
+      "step": 58050
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.34635643041207e-05,
+      "loss": 0.2734,
+      "step": 58055
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.346088374179593e-05,
+      "loss": 0.2706,
+      "step": 58060
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.345820306961218e-05,
+      "loss": 0.2608,
+      "step": 58065
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.345552228760422e-05,
+      "loss": 0.2507,
+      "step": 58070
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3452841395806887e-05,
+      "loss": 0.2956,
+      "step": 58075
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.345016039425497e-05,
+      "loss": 0.2717,
+      "step": 58080
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.34474792829833e-05,
+      "loss": 0.2607,
+      "step": 58085
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3444798062026674e-05,
+      "loss": 0.2719,
+      "step": 58090
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.344211673141991e-05,
+      "loss": 0.2629,
+      "step": 58095
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3439435291197824e-05,
+      "loss": 0.2764,
+      "step": 58100
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.343675374139523e-05,
+      "loss": 0.2804,
+      "step": 58105
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.343407208204695e-05,
+      "loss": 0.2788,
+      "step": 58110
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.34313903131878e-05,
+      "loss": 0.2597,
+      "step": 58115
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.342870843485261e-05,
+      "loss": 0.2737,
+      "step": 58120
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.3426026447076186e-05,
+      "loss": 0.2646,
+      "step": 58125
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.342334434989337e-05,
+      "loss": 0.2729,
+      "step": 58130
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3420662143338974e-05,
+      "loss": 0.2923,
+      "step": 58135
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.341797982744783e-05,
+      "loss": 0.2681,
+      "step": 58140
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3415297402254764e-05,
+      "loss": 0.2529,
+      "step": 58145
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.341261486779462e-05,
+      "loss": 0.2796,
+      "step": 58150
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3409932224102205e-05,
+      "loss": 0.2909,
+      "step": 58155
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.340724947121236e-05,
+      "loss": 0.2871,
+      "step": 58160
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.340456660915992e-05,
+      "loss": 0.2733,
+      "step": 58165
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.340188363797973e-05,
+      "loss": 0.2731,
+      "step": 58170
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.339920055770661e-05,
+      "loss": 0.2587,
+      "step": 58175
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.339651736837542e-05,
+      "loss": 0.2766,
+      "step": 58180
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3393834070020966e-05,
+      "loss": 0.2629,
+      "step": 58185
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.339115066267812e-05,
+      "loss": 0.2745,
+      "step": 58190
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.338846714638171e-05,
+      "loss": 0.2617,
+      "step": 58195
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.338578352116659e-05,
+      "loss": 0.2845,
+      "step": 58200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.338309978706758e-05,
+      "loss": 0.2641,
+      "step": 58205
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.338041594411956e-05,
+      "loss": 0.3062,
+      "step": 58210
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3377731992357355e-05,
+      "loss": 0.2742,
+      "step": 58215
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3375047931815816e-05,
+      "loss": 0.2501,
+      "step": 58220
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3372363762529815e-05,
+      "loss": 0.2379,
+      "step": 58225
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3369679484534175e-05,
+      "loss": 0.2645,
+      "step": 58230
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.336699509786377e-05,
+      "loss": 0.2771,
+      "step": 58235
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.336431060255344e-05,
+      "loss": 0.2872,
+      "step": 58240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.336162599863805e-05,
+      "loss": 0.2668,
+      "step": 58245
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3358941286152465e-05,
+      "loss": 0.2605,
+      "step": 58250
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.335625646513153e-05,
+      "loss": 0.2445,
+      "step": 58255
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.335357153561011e-05,
+      "loss": 0.2861,
+      "step": 58260
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.335088649762308e-05,
+      "loss": 0.2569,
+      "step": 58265
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3348201351205274e-05,
+      "loss": 0.2772,
+      "step": 58270
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.334551609639159e-05,
+      "loss": 0.2614,
+      "step": 58275
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.334283073321687e-05,
+      "loss": 0.2743,
+      "step": 58280
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3340145261716e-05,
+      "loss": 0.3164,
+      "step": 58285
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3337459681923835e-05,
+      "loss": 0.3013,
+      "step": 58290
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3334773993875254e-05,
+      "loss": 0.2843,
+      "step": 58295
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3332088197605115e-05,
+      "loss": 0.2746,
+      "step": 58300
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.332940229314832e-05,
+      "loss": 0.2719,
+      "step": 58305
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.33267162805397e-05,
+      "loss": 0.2672,
+      "step": 58310
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.332403015981418e-05,
+      "loss": 0.2518,
+      "step": 58315
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.332134393100661e-05,
+      "loss": 0.2824,
+      "step": 58320
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.331865759415186e-05,
+      "loss": 0.2766,
+      "step": 58325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.331597114928484e-05,
+      "loss": 0.263,
+      "step": 58330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.33132845964404e-05,
+      "loss": 0.2864,
+      "step": 58335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.331059793565345e-05,
+      "loss": 0.3015,
+      "step": 58340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.330791116695886e-05,
+      "loss": 0.2648,
+      "step": 58345
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.330522429039152e-05,
+      "loss": 0.2498,
+      "step": 58350
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.330253730598631e-05,
+      "loss": 0.2872,
+      "step": 58355
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3299850213778125e-05,
+      "loss": 0.281,
+      "step": 58360
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.329716301380185e-05,
+      "loss": 0.2943,
+      "step": 58365
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.32944757060924e-05,
+      "loss": 0.2771,
+      "step": 58370
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.329178829068463e-05,
+      "loss": 0.277,
+      "step": 58375
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3289100767613464e-05,
+      "loss": 0.2565,
+      "step": 58380
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.328641313691378e-05,
+      "loss": 0.2783,
+      "step": 58385
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.328372539862049e-05,
+      "loss": 0.2651,
+      "step": 58390
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.328103755276848e-05,
+      "loss": 0.2727,
+      "step": 58395
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3278349599392664e-05,
+      "loss": 0.2875,
+      "step": 58400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.3275661538527926e-05,
+      "loss": 0.2729,
+      "step": 58405
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.327297337020917e-05,
+      "loss": 0.2701,
+      "step": 58410
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.327028509447132e-05,
+      "loss": 0.2621,
+      "step": 58415
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3267596711349256e-05,
+      "loss": 0.2715,
+      "step": 58420
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3264908220877904e-05,
+      "loss": 0.2651,
+      "step": 58425
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3262219623092174e-05,
+      "loss": 0.2773,
+      "step": 58430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.325953091802696e-05,
+      "loss": 0.2623,
+      "step": 58435
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.325684210571717e-05,
+      "loss": 0.267,
+      "step": 58440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.325415318619773e-05,
+      "loss": 0.2918,
+      "step": 58445
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.325146415950356e-05,
+      "loss": 0.2827,
+      "step": 58450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.324877502566957e-05,
+      "loss": 0.2751,
+      "step": 58455
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.324608578473066e-05,
+      "loss": 0.251,
+      "step": 58460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.324339643672176e-05,
+      "loss": 0.3062,
+      "step": 58465
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.324070698167779e-05,
+      "loss": 0.259,
+      "step": 58470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3238017419633684e-05,
+      "loss": 0.2959,
+      "step": 58475
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.323532775062434e-05,
+      "loss": 0.2834,
+      "step": 58480
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.32326379746847e-05,
+      "loss": 0.2654,
+      "step": 58485
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3229948091849674e-05,
+      "loss": 0.2705,
+      "step": 58490
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.32272581021542e-05,
+      "loss": 0.2969,
+      "step": 58495
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3224568005633195e-05,
+      "loss": 0.2682,
+      "step": 58500
+    },
+    {
+      "epoch": 2.06,
+      "eval_loss": 0.2701320946216583,
+      "eval_runtime": 10.5303,
+      "eval_samples_per_second": 9.496,
+      "eval_steps_per_second": 9.496,
+      "step": 58500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.32218778023216e-05,
+      "loss": 0.2425,
+      "step": 58505
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.321918749225434e-05,
+      "loss": 0.2702,
+      "step": 58510
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.321649707546635e-05,
+      "loss": 0.2658,
+      "step": 58515
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.321380655199255e-05,
+      "loss": 0.2684,
+      "step": 58520
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3211115921867894e-05,
+      "loss": 0.2719,
+      "step": 58525
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3208425185127305e-05,
+      "loss": 0.2789,
+      "step": 58530
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3205734341805725e-05,
+      "loss": 0.2787,
+      "step": 58535
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3203043391938094e-05,
+      "loss": 0.2594,
+      "step": 58540
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.320035233555935e-05,
+      "loss": 0.2918,
+      "step": 58545
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.319766117270444e-05,
+      "loss": 0.2422,
+      "step": 58550
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3194969903408294e-05,
+      "loss": 0.281,
+      "step": 58555
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.319227852770587e-05,
+      "loss": 0.2878,
+      "step": 58560
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.31895870456321e-05,
+      "loss": 0.2678,
+      "step": 58565
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.318689545722195e-05,
+      "loss": 0.2968,
+      "step": 58570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.318420376251035e-05,
+      "loss": 0.2631,
+      "step": 58575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.318151196153226e-05,
+      "loss": 0.2765,
+      "step": 58580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.317882005432263e-05,
+      "loss": 0.2809,
+      "step": 58585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.317612804091641e-05,
+      "loss": 0.2923,
+      "step": 58590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.317343592134856e-05,
+      "loss": 0.2739,
+      "step": 58595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3170743695654026e-05,
+      "loss": 0.2769,
+      "step": 58600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3168051363867774e-05,
+      "loss": 0.2654,
+      "step": 58605
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.316535892602476e-05,
+      "loss": 0.2636,
+      "step": 58610
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3162666382159935e-05,
+      "loss": 0.264,
+      "step": 58615
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.315997373230828e-05,
+      "loss": 0.3044,
+      "step": 58620
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3157280976504726e-05,
+      "loss": 0.2707,
+      "step": 58625
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.315458811478426e-05,
+      "loss": 0.315,
+      "step": 58630
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.315189514718184e-05,
+      "loss": 0.2899,
+      "step": 58635
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3149202073732446e-05,
+      "loss": 0.2767,
+      "step": 58640
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.314650889447102e-05,
+      "loss": 0.2539,
+      "step": 58645
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.314381560943255e-05,
+      "loss": 0.2922,
+      "step": 58650
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3141122218652e-05,
+      "loss": 0.2856,
+      "step": 58655
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.313842872216434e-05,
+      "loss": 0.2456,
+      "step": 58660
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.313573512000455e-05,
+      "loss": 0.2928,
+      "step": 58665
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3133041412207596e-05,
+      "loss": 0.2632,
+      "step": 58670
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.313034759880847e-05,
+      "loss": 0.2838,
+      "step": 58675
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.312765367984213e-05,
+      "loss": 0.2953,
+      "step": 58680
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3124959655343564e-05,
+      "loss": 0.264,
+      "step": 58685
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.3122265525347754e-05,
+      "loss": 0.2863,
+      "step": 58690
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3119571289889675e-05,
+      "loss": 0.2702,
+      "step": 58695
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.311687694900432e-05,
+      "loss": 0.303,
+      "step": 58700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3114182502726664e-05,
+      "loss": 0.2653,
+      "step": 58705
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.31114879510917e-05,
+      "loss": 0.2366,
+      "step": 58710
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3108793294134405e-05,
+      "loss": 0.2524,
+      "step": 58715
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.310609853188977e-05,
+      "loss": 0.2888,
+      "step": 58720
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3103403664392797e-05,
+      "loss": 0.2645,
+      "step": 58725
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.310070869167846e-05,
+      "loss": 0.2613,
+      "step": 58730
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.309801361378177e-05,
+      "loss": 0.2458,
+      "step": 58735
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.309531843073769e-05,
+      "loss": 0.2817,
+      "step": 58740
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.309262314258125e-05,
+      "loss": 0.269,
+      "step": 58745
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3089927749347426e-05,
+      "loss": 0.2837,
+      "step": 58750
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.308723225107123e-05,
+      "loss": 0.2785,
+      "step": 58755
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.308453664778765e-05,
+      "loss": 0.2643,
+      "step": 58760
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3081840939531684e-05,
+      "loss": 0.2622,
+      "step": 58765
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.307914512633834e-05,
+      "loss": 0.2705,
+      "step": 58770
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3076449208242625e-05,
+      "loss": 0.2742,
+      "step": 58775
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.307375318527954e-05,
+      "loss": 0.2669,
+      "step": 58780
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.307105705748409e-05,
+      "loss": 0.2481,
+      "step": 58785
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.306836082489128e-05,
+      "loss": 0.2822,
+      "step": 58790
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3065664487536124e-05,
+      "loss": 0.2952,
+      "step": 58795
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.306296804545362e-05,
+      "loss": 0.248,
+      "step": 58800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.306027149867881e-05,
+      "loss": 0.2751,
+      "step": 58805
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.305757484724667e-05,
+      "loss": 0.2852,
+      "step": 58810
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3054878091192245e-05,
+      "loss": 0.2712,
+      "step": 58815
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.305218123055052e-05,
+      "loss": 0.2791,
+      "step": 58820
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3049484265356536e-05,
+      "loss": 0.2696,
+      "step": 58825
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3046787195645306e-05,
+      "loss": 0.2566,
+      "step": 58830
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3044090021451845e-05,
+      "loss": 0.2605,
+      "step": 58835
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3041392742811184e-05,
+      "loss": 0.2699,
+      "step": 58840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3038695359758324e-05,
+      "loss": 0.2714,
+      "step": 58845
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3035997872328304e-05,
+      "loss": 0.2648,
+      "step": 58850
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.303330028055616e-05,
+      "loss": 0.274,
+      "step": 58855
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.30306025844769e-05,
+      "loss": 0.2708,
+      "step": 58860
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.302790478412555e-05,
+      "loss": 0.2962,
+      "step": 58865
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3025206879537166e-05,
+      "loss": 0.2773,
+      "step": 58870
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.302250887074674e-05,
+      "loss": 0.2597,
+      "step": 58875
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3019810757789325e-05,
+      "loss": 0.2545,
+      "step": 58880
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.301711254069996e-05,
+      "loss": 0.2814,
+      "step": 58885
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3014414219513676e-05,
+      "loss": 0.2782,
+      "step": 58890
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3011715794265494e-05,
+      "loss": 0.2859,
+      "step": 58895
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.300901726499046e-05,
+      "loss": 0.2637,
+      "step": 58900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.300631863172363e-05,
+      "loss": 0.2925,
+      "step": 58905
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.300361989450002e-05,
+      "loss": 0.2558,
+      "step": 58910
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.3000921053354675e-05,
+      "loss": 0.2598,
+      "step": 58915
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.299822210832265e-05,
+      "loss": 0.2496,
+      "step": 58920
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.299552305943899e-05,
+      "loss": 0.2827,
+      "step": 58925
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.299282390673872e-05,
+      "loss": 0.2558,
+      "step": 58930
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.29901246502569e-05,
+      "loss": 0.2689,
+      "step": 58935
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.298742529002857e-05,
+      "loss": 0.282,
+      "step": 58940
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.298472582608881e-05,
+      "loss": 0.2818,
+      "step": 58945
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.298202625847263e-05,
+      "loss": 0.2714,
+      "step": 58950
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.297932658721511e-05,
+      "loss": 0.2718,
+      "step": 58955
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2976626812351275e-05,
+      "loss": 0.2591,
+      "step": 58960
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.297392693391621e-05,
+      "loss": 0.277,
+      "step": 58965
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2971226951944955e-05,
+      "loss": 0.26,
+      "step": 58970
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.2968526866472576e-05,
+      "loss": 0.269,
+      "step": 58975
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2965826677534126e-05,
+      "loss": 0.2814,
+      "step": 58980
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.296312638516466e-05,
+      "loss": 0.3001,
+      "step": 58985
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.296042598939925e-05,
+      "loss": 0.2824,
+      "step": 58990
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.295772549027296e-05,
+      "loss": 0.256,
+      "step": 58995
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.295502488782084e-05,
+      "loss": 0.2646,
+      "step": 59000
+    },
+    {
+      "epoch": 2.08,
+      "eval_loss": 0.27035436034202576,
+      "eval_runtime": 10.5421,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 59000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.295232418207798e-05,
+      "loss": 0.275,
+      "step": 59005
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2949623373079406e-05,
+      "loss": 0.2762,
+      "step": 59010
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.294692246086023e-05,
+      "loss": 0.2561,
+      "step": 59015
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2944221445455494e-05,
+      "loss": 0.2895,
+      "step": 59020
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.294152032690029e-05,
+      "loss": 0.2872,
+      "step": 59025
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.293881910522967e-05,
+      "loss": 0.2779,
+      "step": 59030
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.293611778047871e-05,
+      "loss": 0.2808,
+      "step": 59035
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.293341635268249e-05,
+      "loss": 0.2849,
+      "step": 59040
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2930714821876094e-05,
+      "loss": 0.242,
+      "step": 59045
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.29280131880946e-05,
+      "loss": 0.2622,
+      "step": 59050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.292531145137307e-05,
+      "loss": 0.2645,
+      "step": 59055
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2922609611746595e-05,
+      "loss": 0.2838,
+      "step": 59060
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.291990766925026e-05,
+      "loss": 0.2518,
+      "step": 59065
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2917205623919134e-05,
+      "loss": 0.2829,
+      "step": 59070
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.291450347578832e-05,
+      "loss": 0.2837,
+      "step": 59075
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2911801224892894e-05,
+      "loss": 0.2655,
+      "step": 59080
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2909098871267936e-05,
+      "loss": 0.2522,
+      "step": 59085
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.290639641494855e-05,
+      "loss": 0.2826,
+      "step": 59090
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.290369385596981e-05,
+      "loss": 0.2664,
+      "step": 59095
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.290099119436681e-05,
+      "loss": 0.2599,
+      "step": 59100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.289828843017465e-05,
+      "loss": 0.2667,
+      "step": 59105
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.289558556342843e-05,
+      "loss": 0.2931,
+      "step": 59110
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.289288259416322e-05,
+      "loss": 0.2535,
+      "step": 59115
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.289017952241414e-05,
+      "loss": 0.265,
+      "step": 59120
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2887476348216275e-05,
+      "loss": 0.2608,
+      "step": 59125
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.288477307160474e-05,
+      "loss": 0.269,
+      "step": 59130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.28820696926146e-05,
+      "loss": 0.2513,
+      "step": 59135
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2879366211281e-05,
+      "loss": 0.2632,
+      "step": 59140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.287666262763901e-05,
+      "loss": 0.2539,
+      "step": 59145
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2873958941723745e-05,
+      "loss": 0.26,
+      "step": 59150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.287125515357032e-05,
+      "loss": 0.2569,
+      "step": 59155
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.286855126321383e-05,
+      "loss": 0.2446,
+      "step": 59160
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.28658472706894e-05,
+      "loss": 0.2793,
+      "step": 59165
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.286314317603211e-05,
+      "loss": 0.276,
+      "step": 59170
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2860438979277096e-05,
+      "loss": 0.294,
+      "step": 59175
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2857734680459455e-05,
+      "loss": 0.2766,
+      "step": 59180
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2855030279614316e-05,
+      "loss": 0.2468,
+      "step": 59185
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.285232577677679e-05,
+      "loss": 0.2683,
+      "step": 59190
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.284962117198198e-05,
+      "loss": 0.266,
+      "step": 59195
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2846916465265016e-05,
+      "loss": 0.2587,
+      "step": 59200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.284421165666101e-05,
+      "loss": 0.2774,
+      "step": 59205
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.284150674620508e-05,
+      "loss": 0.2789,
+      "step": 59210
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.283880173393236e-05,
+      "loss": 0.2867,
+      "step": 59215
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.283609661987797e-05,
+      "loss": 0.2893,
+      "step": 59220
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2833391404077025e-05,
+      "loss": 0.3004,
+      "step": 59225
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.283068608656465e-05,
+      "loss": 0.2544,
+      "step": 59230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.282798066737598e-05,
+      "loss": 0.2583,
+      "step": 59235
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2825275146546136e-05,
+      "loss": 0.2954,
+      "step": 59240
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.282256952411026e-05,
+      "loss": 0.2731,
+      "step": 59245
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2819863800103466e-05,
+      "loss": 0.2743,
+      "step": 59250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2817157974560904e-05,
+      "loss": 0.2589,
+      "step": 59255
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.2814452047517685e-05,
+      "loss": 0.2905,
+      "step": 59260
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2811746019008966e-05,
+      "loss": 0.2651,
+      "step": 59265
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.280903988906986e-05,
+      "loss": 0.2655,
+      "step": 59270
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.280633365773553e-05,
+      "loss": 0.2618,
+      "step": 59275
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.28036273250411e-05,
+      "loss": 0.2809,
+      "step": 59280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2800920891021714e-05,
+      "loss": 0.2727,
+      "step": 59285
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.27982143557125e-05,
+      "loss": 0.2796,
+      "step": 59290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2795507719148625e-05,
+      "loss": 0.2735,
+      "step": 59295
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.279280098136521e-05,
+      "loss": 0.246,
+      "step": 59300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.279009414239742e-05,
+      "loss": 0.2973,
+      "step": 59305
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.278738720228039e-05,
+      "loss": 0.2841,
+      "step": 59310
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.278468016104926e-05,
+      "loss": 0.2841,
+      "step": 59315
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.278197301873919e-05,
+      "loss": 0.2635,
+      "step": 59320
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2779265775385335e-05,
+      "loss": 0.2695,
+      "step": 59325
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.277655843102284e-05,
+      "loss": 0.2893,
+      "step": 59330
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.277385098568685e-05,
+      "loss": 0.2398,
+      "step": 59335
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.277114343941254e-05,
+      "loss": 0.2809,
+      "step": 59340
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2768435792235044e-05,
+      "loss": 0.2598,
+      "step": 59345
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.276572804418954e-05,
+      "loss": 0.2785,
+      "step": 59350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.276302019531116e-05,
+      "loss": 0.2738,
+      "step": 59355
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2760312245635094e-05,
+      "loss": 0.293,
+      "step": 59360
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2757604195196477e-05,
+      "loss": 0.2736,
+      "step": 59365
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.275489604403048e-05,
+      "loss": 0.2656,
+      "step": 59370
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2752187792172276e-05,
+      "loss": 0.2904,
+      "step": 59375
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.274947943965703e-05,
+      "loss": 0.2758,
+      "step": 59380
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.274677098651989e-05,
+      "loss": 0.2808,
+      "step": 59385
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.274406243279603e-05,
+      "loss": 0.3196,
+      "step": 59390
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2741353778520636e-05,
+      "loss": 0.2627,
+      "step": 59395
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.273864502372885e-05,
+      "loss": 0.2666,
+      "step": 59400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2735936168455864e-05,
+      "loss": 0.2909,
+      "step": 59405
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.273322721273685e-05,
+      "loss": 0.2701,
+      "step": 59410
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.273051815660698e-05,
+      "loss": 0.266,
+      "step": 59415
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2727809000101413e-05,
+      "loss": 0.2632,
+      "step": 59420
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2725099743255345e-05,
+      "loss": 0.2612,
+      "step": 59425
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2722390386103946e-05,
+      "loss": 0.2551,
+      "step": 59430
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.271968092868241e-05,
+      "loss": 0.2569,
+      "step": 59435
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.271697137102588e-05,
+      "loss": 0.2754,
+      "step": 59440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.271426171316959e-05,
+      "loss": 0.2608,
+      "step": 59445
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.271155195514868e-05,
+      "loss": 0.2745,
+      "step": 59450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2708842096998346e-05,
+      "loss": 0.2517,
+      "step": 59455
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2706132138753784e-05,
+      "loss": 0.2879,
+      "step": 59460
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.270342208045018e-05,
+      "loss": 0.2594,
+      "step": 59465
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.270071192212271e-05,
+      "loss": 0.2515,
+      "step": 59470
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2698001663806575e-05,
+      "loss": 0.2823,
+      "step": 59475
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.269529130553695e-05,
+      "loss": 0.2801,
+      "step": 59480
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.269258084734905e-05,
+      "loss": 0.2701,
+      "step": 59485
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.268987028927806e-05,
+      "loss": 0.2805,
+      "step": 59490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2687159631359164e-05,
+      "loss": 0.2796,
+      "step": 59495
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.268444887362757e-05,
+      "loss": 0.2773,
+      "step": 59500
+    },
+    {
+      "epoch": 2.09,
+      "eval_loss": 0.26948508620262146,
+      "eval_runtime": 10.5443,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 59500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.268173801611847e-05,
+      "loss": 0.2598,
+      "step": 59505
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.267902705886706e-05,
+      "loss": 0.2798,
+      "step": 59510
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.267631600190855e-05,
+      "loss": 0.2612,
+      "step": 59515
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.267360484527814e-05,
+      "loss": 0.2736,
+      "step": 59520
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2670893589011025e-05,
+      "loss": 0.2493,
+      "step": 59525
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.2668182233142414e-05,
+      "loss": 0.257,
+      "step": 59530
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.266547077770751e-05,
+      "loss": 0.2727,
+      "step": 59535
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.266275922274151e-05,
+      "loss": 0.2915,
+      "step": 59540
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.266004756827963e-05,
+      "loss": 0.2943,
+      "step": 59545
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.26573358143571e-05,
+      "loss": 0.2857,
+      "step": 59550
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.26546239610091e-05,
+      "loss": 0.2579,
+      "step": 59555
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.265191200827085e-05,
+      "loss": 0.2391,
+      "step": 59560
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2649199956177564e-05,
+      "loss": 0.2757,
+      "step": 59565
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2646487804764456e-05,
+      "loss": 0.2885,
+      "step": 59570
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2643775554066754e-05,
+      "loss": 0.2694,
+      "step": 59575
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.264106320411966e-05,
+      "loss": 0.2738,
+      "step": 59580
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.263835075495839e-05,
+      "loss": 0.2896,
+      "step": 59585
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.263563820661817e-05,
+      "loss": 0.2797,
+      "step": 59590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.263292555913422e-05,
+      "loss": 0.2776,
+      "step": 59595
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.263021281254177e-05,
+      "loss": 0.272,
+      "step": 59600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.262749996687603e-05,
+      "loss": 0.2816,
+      "step": 59605
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.262478702217223e-05,
+      "loss": 0.2527,
+      "step": 59610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.262207397846559e-05,
+      "loss": 0.2852,
+      "step": 59615
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.261936083579135e-05,
+      "loss": 0.2905,
+      "step": 59620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2616647594184726e-05,
+      "loss": 0.2726,
+      "step": 59625
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2613934253680956e-05,
+      "loss": 0.2932,
+      "step": 59630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.261122081431527e-05,
+      "loss": 0.2569,
+      "step": 59635
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.260850727612289e-05,
+      "loss": 0.2814,
+      "step": 59640
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.260633637443745e-05,
+      "loss": 0.2669,
+      "step": 59645
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.260362265844583e-05,
+      "loss": 0.2827,
+      "step": 59650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2600908843726175e-05,
+      "loss": 0.2866,
+      "step": 59655
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2598194930313734e-05,
+      "loss": 0.2778,
+      "step": 59660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2595480918243734e-05,
+      "loss": 0.2808,
+      "step": 59665
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.259276680755143e-05,
+      "loss": 0.2708,
+      "step": 59670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.259005259827205e-05,
+      "loss": 0.2914,
+      "step": 59675
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2587338290440824e-05,
+      "loss": 0.2822,
+      "step": 59680
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2584623884093027e-05,
+      "loss": 0.2654,
+      "step": 59685
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.258190937926387e-05,
+      "loss": 0.2899,
+      "step": 59690
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.257919477598863e-05,
+      "loss": 0.2628,
+      "step": 59695
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2576480074302544e-05,
+      "loss": 0.2657,
+      "step": 59700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2573765274240856e-05,
+      "loss": 0.2717,
+      "step": 59705
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2571050375838806e-05,
+      "loss": 0.2832,
+      "step": 59710
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.256833537913166e-05,
+      "loss": 0.2799,
+      "step": 59715
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.256562028415467e-05,
+      "loss": 0.2696,
+      "step": 59720
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2562905090943095e-05,
+      "loss": 0.2867,
+      "step": 59725
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.256018979953217e-05,
+      "loss": 0.2534,
+      "step": 59730
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.255747440995717e-05,
+      "loss": 0.2905,
+      "step": 59735
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2554758922253336e-05,
+      "loss": 0.2662,
+      "step": 59740
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.255204333645595e-05,
+      "loss": 0.2624,
+      "step": 59745
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2549327652600245e-05,
+      "loss": 0.284,
+      "step": 59750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2546611870721506e-05,
+      "loss": 0.2606,
+      "step": 59755
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.254389599085498e-05,
+      "loss": 0.2445,
+      "step": 59760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.254118001303593e-05,
+      "loss": 0.2853,
+      "step": 59765
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2538463937299635e-05,
+      "loss": 0.2689,
+      "step": 59770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.253574776368135e-05,
+      "loss": 0.2578,
+      "step": 59775
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.253303149221635e-05,
+      "loss": 0.2666,
+      "step": 59780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.25303151229399e-05,
+      "loss": 0.2814,
+      "step": 59785
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.252759865588726e-05,
+      "loss": 0.2801,
+      "step": 59790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.252488209109373e-05,
+      "loss": 0.2776,
+      "step": 59795
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.252216542859456e-05,
+      "loss": 0.2657,
+      "step": 59800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.251944866842502e-05,
+      "loss": 0.2865,
+      "step": 59805
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.25167318106204e-05,
+      "loss": 0.2521,
+      "step": 59810
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.251401485521598e-05,
+      "loss": 0.2784,
+      "step": 59815
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.251129780224702e-05,
+      "loss": 0.2967,
+      "step": 59820
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.2508580651748804e-05,
+      "loss": 0.2709,
+      "step": 59825
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.250586340375663e-05,
+      "loss": 0.2599,
+      "step": 59830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.250314605830575e-05,
+      "loss": 0.2538,
+      "step": 59835
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2500428615431474e-05,
+      "loss": 0.2799,
+      "step": 59840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.249771107516907e-05,
+      "loss": 0.2776,
+      "step": 59845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2494993437553836e-05,
+      "loss": 0.2709,
+      "step": 59850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2492275702621045e-05,
+      "loss": 0.247,
+      "step": 59855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2489557870406e-05,
+      "loss": 0.2678,
+      "step": 59860
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2486839940943967e-05,
+      "loss": 0.2833,
+      "step": 59865
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.248412191427027e-05,
+      "loss": 0.3118,
+      "step": 59870
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2481403790420164e-05,
+      "loss": 0.2482,
+      "step": 59875
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2478685569428975e-05,
+      "loss": 0.2719,
+      "step": 59880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.247596725133197e-05,
+      "loss": 0.2679,
+      "step": 59885
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.247324883616447e-05,
+      "loss": 0.3003,
+      "step": 59890
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.247053032396175e-05,
+      "loss": 0.268,
+      "step": 59895
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2467811714759114e-05,
+      "loss": 0.2698,
+      "step": 59900
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.246509300859187e-05,
+      "loss": 0.2986,
+      "step": 59905
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2462374205495314e-05,
+      "loss": 0.255,
+      "step": 59910
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.245965530550475e-05,
+      "loss": 0.2731,
+      "step": 59915
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2456936308655475e-05,
+      "loss": 0.2569,
+      "step": 59920
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.245421721498279e-05,
+      "loss": 0.2728,
+      "step": 59925
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.245149802452201e-05,
+      "loss": 0.2508,
+      "step": 59930
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.244877873730844e-05,
+      "loss": 0.2521,
+      "step": 59935
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.244605935337739e-05,
+      "loss": 0.297,
+      "step": 59940
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.244333987276417e-05,
+      "loss": 0.2943,
+      "step": 59945
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2440620295504076e-05,
+      "loss": 0.2698,
+      "step": 59950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2437900621632435e-05,
+      "loss": 0.2422,
+      "step": 59955
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.243518085118456e-05,
+      "loss": 0.2802,
+      "step": 59960
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.243246098419576e-05,
+      "loss": 0.2635,
+      "step": 59965
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.242974102070134e-05,
+      "loss": 0.2706,
+      "step": 59970
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2427020960736645e-05,
+      "loss": 0.2721,
+      "step": 59975
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.242430080433698e-05,
+      "loss": 0.2893,
+      "step": 59980
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.242158055153764e-05,
+      "loss": 0.2645,
+      "step": 59985
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.241886020237399e-05,
+      "loss": 0.2629,
+      "step": 59990
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.241613975688132e-05,
+      "loss": 0.2837,
+      "step": 59995
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2413419215094966e-05,
+      "loss": 0.2808,
+      "step": 60000
+    },
+    {
+      "epoch": 2.11,
+      "eval_loss": 0.2701481282711029,
+      "eval_runtime": 10.5458,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 60000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.241069857705025e-05,
+      "loss": 0.2705,
+      "step": 60005
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.240797784278249e-05,
+      "loss": 0.2748,
+      "step": 60010
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.240525701232703e-05,
+      "loss": 0.2734,
+      "step": 60015
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.240253608571918e-05,
+      "loss": 0.2739,
+      "step": 60020
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.239981506299428e-05,
+      "loss": 0.2664,
+      "step": 60025
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2397093944187656e-05,
+      "loss": 0.2603,
+      "step": 60030
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.239437272933464e-05,
+      "loss": 0.2468,
+      "step": 60035
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.239165141847057e-05,
+      "loss": 0.286,
+      "step": 60040
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.238893001163078e-05,
+      "loss": 0.2758,
+      "step": 60045
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2386208508850605e-05,
+      "loss": 0.288,
+      "step": 60050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.238348691016537e-05,
+      "loss": 0.2894,
+      "step": 60055
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.238076521561044e-05,
+      "loss": 0.2633,
+      "step": 60060
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2378043425221116e-05,
+      "loss": 0.2684,
+      "step": 60065
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.237532153903278e-05,
+      "loss": 0.3006,
+      "step": 60070
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.237259955708074e-05,
+      "loss": 0.2576,
+      "step": 60075
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.236987747940037e-05,
+      "loss": 0.2727,
+      "step": 60080
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.236715530602698e-05,
+      "loss": 0.2846,
+      "step": 60085
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.236443303699595e-05,
+      "loss": 0.2586,
+      "step": 60090
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.236171067234259e-05,
+      "loss": 0.2887,
+      "step": 60095
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.235898821210229e-05,
+      "loss": 0.2896,
+      "step": 60100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.2356265656310365e-05,
+      "loss": 0.2694,
+      "step": 60105
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.235354300500218e-05,
+      "loss": 0.25,
+      "step": 60110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.235082025821309e-05,
+      "loss": 0.2701,
+      "step": 60115
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.234809741597844e-05,
+      "loss": 0.293,
+      "step": 60120
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.234537447833358e-05,
+      "loss": 0.2771,
+      "step": 60125
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.234265144531389e-05,
+      "loss": 0.282,
+      "step": 60130
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2339928316954696e-05,
+      "loss": 0.2647,
+      "step": 60135
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2337205093291375e-05,
+      "loss": 0.282,
+      "step": 60140
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.233448177435928e-05,
+      "loss": 0.2892,
+      "step": 60145
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2331758360193774e-05,
+      "loss": 0.2863,
+      "step": 60150
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.232903485083022e-05,
+      "loss": 0.2757,
+      "step": 60155
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2326311246303967e-05,
+      "loss": 0.2936,
+      "step": 60160
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2323587546650406e-05,
+      "loss": 0.254,
+      "step": 60165
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2320863751904875e-05,
+      "loss": 0.2743,
+      "step": 60170
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.231813986210276e-05,
+      "loss": 0.2725,
+      "step": 60175
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.231541587727942e-05,
+      "loss": 0.2518,
+      "step": 60180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.231269179747023e-05,
+      "loss": 0.2639,
+      "step": 60185
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.230996762271055e-05,
+      "loss": 0.2875,
+      "step": 60190
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.230724335303576e-05,
+      "loss": 0.2559,
+      "step": 60195
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2304518988481234e-05,
+      "loss": 0.2703,
+      "step": 60200
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.230179452908234e-05,
+      "loss": 0.272,
+      "step": 60205
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.229906997487445e-05,
+      "loss": 0.2638,
+      "step": 60210
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.229634532589296e-05,
+      "loss": 0.2637,
+      "step": 60215
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2293620582173226e-05,
+      "loss": 0.2767,
+      "step": 60220
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2290895743750634e-05,
+      "loss": 0.2669,
+      "step": 60225
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2288170810660565e-05,
+      "loss": 0.2701,
+      "step": 60230
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.228544578293841e-05,
+      "loss": 0.2598,
+      "step": 60235
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.228272066061953e-05,
+      "loss": 0.2803,
+      "step": 60240
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.227999544373933e-05,
+      "loss": 0.2956,
+      "step": 60245
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2277270132333185e-05,
+      "loss": 0.249,
+      "step": 60250
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.227454472643649e-05,
+      "loss": 0.259,
+      "step": 60255
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.227181922608462e-05,
+      "loss": 0.2769,
+      "step": 60260
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.226909363131296e-05,
+      "loss": 0.2997,
+      "step": 60265
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.226636794215692e-05,
+      "loss": 0.2831,
+      "step": 60270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.226364215865188e-05,
+      "loss": 0.2585,
+      "step": 60275
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2260916280833244e-05,
+      "loss": 0.2614,
+      "step": 60280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2258190308736375e-05,
+      "loss": 0.2785,
+      "step": 60285
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.22554642423967e-05,
+      "loss": 0.2941,
+      "step": 60290
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.22527380818496e-05,
+      "loss": 0.2816,
+      "step": 60295
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.225001182713048e-05,
+      "loss": 0.2667,
+      "step": 60300
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.224728547827473e-05,
+      "loss": 0.2592,
+      "step": 60305
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.224455903531775e-05,
+      "loss": 0.2858,
+      "step": 60310
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.224183249829495e-05,
+      "loss": 0.2857,
+      "step": 60315
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.223910586724173e-05,
+      "loss": 0.268,
+      "step": 60320
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.223637914219349e-05,
+      "loss": 0.275,
+      "step": 60325
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2233652323185636e-05,
+      "loss": 0.2762,
+      "step": 60330
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2230925410253576e-05,
+      "loss": 0.2881,
+      "step": 60335
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2228198403432705e-05,
+      "loss": 0.2852,
+      "step": 60340
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.222547130275845e-05,
+      "loss": 0.2704,
+      "step": 60345
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.222274410826621e-05,
+      "loss": 0.2745,
+      "step": 60350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.22200168199914e-05,
+      "loss": 0.2804,
+      "step": 60355
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2217289437969435e-05,
+      "loss": 0.2782,
+      "step": 60360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.22145619622357e-05,
+      "loss": 0.2681,
+      "step": 60365
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.221183439282566e-05,
+      "loss": 0.2567,
+      "step": 60370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2209106729774686e-05,
+      "loss": 0.2655,
+      "step": 60375
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.220637897311822e-05,
+      "loss": 0.2645,
+      "step": 60380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2203651122891664e-05,
+      "loss": 0.2739,
+      "step": 60385
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2200923179130455e-05,
+      "loss": 0.282,
+      "step": 60390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.2198195141869995e-05,
+      "loss": 0.2717,
+      "step": 60395
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.219546701114572e-05,
+      "loss": 0.251,
+      "step": 60400
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.219273878699305e-05,
+      "loss": 0.2748,
+      "step": 60405
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.219001046944741e-05,
+      "loss": 0.2541,
+      "step": 60410
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2187282058544214e-05,
+      "loss": 0.2729,
+      "step": 60415
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2184553554318894e-05,
+      "loss": 0.2877,
+      "step": 60420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2181824956806896e-05,
+      "loss": 0.2689,
+      "step": 60425
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.217909626604362e-05,
+      "loss": 0.2677,
+      "step": 60430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.217636748206452e-05,
+      "loss": 0.2678,
+      "step": 60435
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.217363860490501e-05,
+      "loss": 0.2515,
+      "step": 60440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.217090963460053e-05,
+      "loss": 0.2667,
+      "step": 60445
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.216818057118652e-05,
+      "loss": 0.2902,
+      "step": 60450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.216545141469841e-05,
+      "loss": 0.2829,
+      "step": 60455
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2162722165171624e-05,
+      "loss": 0.2775,
+      "step": 60460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2159992822641626e-05,
+      "loss": 0.2978,
+      "step": 60465
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.215726338714383e-05,
+      "loss": 0.2594,
+      "step": 60470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.215453385871369e-05,
+      "loss": 0.282,
+      "step": 60475
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.215180423738664e-05,
+      "loss": 0.2855,
+      "step": 60480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.214907452319813e-05,
+      "loss": 0.268,
+      "step": 60485
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.21463447161836e-05,
+      "loss": 0.2403,
+      "step": 60490
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2143614816378486e-05,
+      "loss": 0.2631,
+      "step": 60495
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2140884823818246e-05,
+      "loss": 0.2759,
+      "step": 60500
+    },
+    {
+      "epoch": 2.13,
+      "eval_loss": 0.26962408423423767,
+      "eval_runtime": 10.5459,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 60500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.213815473853832e-05,
+      "loss": 0.2721,
+      "step": 60505
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.213542456057416e-05,
+      "loss": 0.2745,
+      "step": 60510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.213269428996121e-05,
+      "loss": 0.2793,
+      "step": 60515
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.212996392673494e-05,
+      "loss": 0.2722,
+      "step": 60520
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.212723347093077e-05,
+      "loss": 0.2784,
+      "step": 60525
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.212450292258418e-05,
+      "loss": 0.2722,
+      "step": 60530
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2121772281730603e-05,
+      "loss": 0.2757,
+      "step": 60535
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2119041548405526e-05,
+      "loss": 0.2651,
+      "step": 60540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2116310722644374e-05,
+      "loss": 0.2571,
+      "step": 60545
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.211357980448262e-05,
+      "loss": 0.2568,
+      "step": 60550
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.211084879395572e-05,
+      "loss": 0.2969,
+      "step": 60555
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.210811769109913e-05,
+      "loss": 0.2799,
+      "step": 60560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.210538649594832e-05,
+      "loss": 0.2843,
+      "step": 60565
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.210265520853875e-05,
+      "loss": 0.2762,
+      "step": 60570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2099923828905883e-05,
+      "loss": 0.2606,
+      "step": 60575
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.209719235708519e-05,
+      "loss": 0.2838,
+      "step": 60580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.209446079311211e-05,
+      "loss": 0.2753,
+      "step": 60585
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.209172913702215e-05,
+      "loss": 0.2683,
+      "step": 60590
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.208899738885076e-05,
+      "loss": 0.2761,
+      "step": 60595
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.208626554863341e-05,
+      "loss": 0.2663,
+      "step": 60600
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.208353361640557e-05,
+      "loss": 0.2608,
+      "step": 60605
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2080801592202715e-05,
+      "loss": 0.2925,
+      "step": 60610
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2078069476060315e-05,
+      "loss": 0.2767,
+      "step": 60615
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.207533726801385e-05,
+      "loss": 0.2809,
+      "step": 60620
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.207260496809879e-05,
+      "loss": 0.2835,
+      "step": 60625
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.206987257635062e-05,
+      "loss": 0.2855,
+      "step": 60630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.206714009280481e-05,
+      "loss": 0.2797,
+      "step": 60635
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.206440751749685e-05,
+      "loss": 0.2827,
+      "step": 60640
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.206167485046221e-05,
+      "loss": 0.2775,
+      "step": 60645
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2058942091736366e-05,
+      "loss": 0.2738,
+      "step": 60650
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2056209241354826e-05,
+      "loss": 0.2301,
+      "step": 60655
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.205347629935305e-05,
+      "loss": 0.3128,
+      "step": 60660
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.2050743265766534e-05,
+      "loss": 0.2745,
+      "step": 60665
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.204801014063077e-05,
+      "loss": 0.2572,
+      "step": 60670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.204527692398124e-05,
+      "loss": 0.2781,
+      "step": 60675
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.204254361585342e-05,
+      "loss": 0.2621,
+      "step": 60680
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.203981021628282e-05,
+      "loss": 0.2497,
+      "step": 60685
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.203707672530493e-05,
+      "loss": 0.2819,
+      "step": 60690
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2034343142955225e-05,
+      "loss": 0.249,
+      "step": 60695
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.203160946926921e-05,
+      "loss": 0.2888,
+      "step": 60700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.202887570428239e-05,
+      "loss": 0.2805,
+      "step": 60705
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2026141848030244e-05,
+      "loss": 0.2742,
+      "step": 60710
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.202340790054828e-05,
+      "loss": 0.2432,
+      "step": 60715
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.202067386187199e-05,
+      "loss": 0.2833,
+      "step": 60720
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.201793973203688e-05,
+      "loss": 0.267,
+      "step": 60725
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.201520551107844e-05,
+      "loss": 0.2622,
+      "step": 60730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.201247119903219e-05,
+      "loss": 0.2818,
+      "step": 60735
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2009736795933616e-05,
+      "loss": 0.2603,
+      "step": 60740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.200700230181823e-05,
+      "loss": 0.2846,
+      "step": 60745
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.2004267716721534e-05,
+      "loss": 0.2779,
+      "step": 60750
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.200153304067904e-05,
+      "loss": 0.2703,
+      "step": 60755
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.199879827372626e-05,
+      "loss": 0.2681,
+      "step": 60760
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.199606341589869e-05,
+      "loss": 0.276,
+      "step": 60765
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.199332846723185e-05,
+      "loss": 0.2633,
+      "step": 60770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.199059342776125e-05,
+      "loss": 0.2834,
+      "step": 60775
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.19878582975224e-05,
+      "loss": 0.2831,
+      "step": 60780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.198512307655081e-05,
+      "loss": 0.2631,
+      "step": 60785
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1982387764882005e-05,
+      "loss": 0.279,
+      "step": 60790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.19796523625515e-05,
+      "loss": 0.2773,
+      "step": 60795
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.197691686959481e-05,
+      "loss": 0.2691,
+      "step": 60800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1974181286047435e-05,
+      "loss": 0.2634,
+      "step": 60805
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.197144561194493e-05,
+      "loss": 0.2702,
+      "step": 60810
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.196870984732279e-05,
+      "loss": 0.2642,
+      "step": 60815
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1965973992216554e-05,
+      "loss": 0.2564,
+      "step": 60820
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.196323804666173e-05,
+      "loss": 0.2799,
+      "step": 60825
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1960502010693844e-05,
+      "loss": 0.2689,
+      "step": 60830
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1957765884348427e-05,
+      "loss": 0.271,
+      "step": 60835
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1955029667661016e-05,
+      "loss": 0.2715,
+      "step": 60840
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.195229336066712e-05,
+      "loss": 0.277,
+      "step": 60845
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.194955696340228e-05,
+      "loss": 0.2632,
+      "step": 60850
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1946820475902024e-05,
+      "loss": 0.2803,
+      "step": 60855
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1944083898201874e-05,
+      "loss": 0.2787,
+      "step": 60860
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.194134723033738e-05,
+      "loss": 0.2935,
+      "step": 60865
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.193861047234406e-05,
+      "loss": 0.279,
+      "step": 60870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.193587362425746e-05,
+      "loss": 0.2632,
+      "step": 60875
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.193313668611311e-05,
+      "loss": 0.2805,
+      "step": 60880
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.193039965794655e-05,
+      "loss": 0.2754,
+      "step": 60885
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.192766253979332e-05,
+      "loss": 0.259,
+      "step": 60890
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.192492533168896e-05,
+      "loss": 0.2614,
+      "step": 60895
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1922188033668995e-05,
+      "loss": 0.2569,
+      "step": 60900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1919450645769e-05,
+      "loss": 0.2703,
+      "step": 60905
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1916713168024484e-05,
+      "loss": 0.2464,
+      "step": 60910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1913975600471006e-05,
+      "loss": 0.2377,
+      "step": 60915
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.191123794314411e-05,
+      "loss": 0.2752,
+      "step": 60920
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.190850019607935e-05,
+      "loss": 0.2745,
+      "step": 60925
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.190576235931227e-05,
+      "loss": 0.2692,
+      "step": 60930
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.190302443287841e-05,
+      "loss": 0.2722,
+      "step": 60935
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.190028641681334e-05,
+      "loss": 0.28,
+      "step": 60940
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.189754831115258e-05,
+      "loss": 0.2689,
+      "step": 60945
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.189481011593171e-05,
+      "loss": 0.2435,
+      "step": 60950
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.1892071831186275e-05,
+      "loss": 0.2531,
+      "step": 60955
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.188933345695183e-05,
+      "loss": 0.2654,
+      "step": 60960
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.188659499326393e-05,
+      "loss": 0.2704,
+      "step": 60965
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1883856440158124e-05,
+      "loss": 0.3151,
+      "step": 60970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.188111779766998e-05,
+      "loss": 0.2528,
+      "step": 60975
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.187837906583507e-05,
+      "loss": 0.2974,
+      "step": 60980
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.187564024468893e-05,
+      "loss": 0.2805,
+      "step": 60985
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.187290133426714e-05,
+      "loss": 0.2844,
+      "step": 60990
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.187016233460525e-05,
+      "loss": 0.2758,
+      "step": 60995
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.186742324573883e-05,
+      "loss": 0.2762,
+      "step": 61000
+    },
+    {
+      "epoch": 2.15,
+      "eval_loss": 0.2689875662326813,
+      "eval_runtime": 10.5446,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 61000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.186468406770344e-05,
+      "loss": 0.2781,
+      "step": 61005
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.186194480053467e-05,
+      "loss": 0.2567,
+      "step": 61010
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.185920544426805e-05,
+      "loss": 0.2522,
+      "step": 61015
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.185646599893918e-05,
+      "loss": 0.2552,
+      "step": 61020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.18537264645836e-05,
+      "loss": 0.2444,
+      "step": 61025
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.185098684123691e-05,
+      "loss": 0.2853,
+      "step": 61030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.184824712893467e-05,
+      "loss": 0.3003,
+      "step": 61035
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.184550732771246e-05,
+      "loss": 0.2714,
+      "step": 61040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.184276743760584e-05,
+      "loss": 0.2848,
+      "step": 61045
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.18400274586504e-05,
+      "loss": 0.2724,
+      "step": 61050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.183728739088171e-05,
+      "loss": 0.2825,
+      "step": 61055
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.183454723433535e-05,
+      "loss": 0.2541,
+      "step": 61060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.183180698904689e-05,
+      "loss": 0.2586,
+      "step": 61065
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.182906665505193e-05,
+      "loss": 0.2811,
+      "step": 61070
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.182632623238604e-05,
+      "loss": 0.2755,
+      "step": 61075
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.182358572108479e-05,
+      "loss": 0.2842,
+      "step": 61080
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1820845121183786e-05,
+      "loss": 0.2524,
+      "step": 61085
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1818104432718596e-05,
+      "loss": 0.2618,
+      "step": 61090
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.181536365572483e-05,
+      "loss": 0.2663,
+      "step": 61095
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.181262279023804e-05,
+      "loss": 0.2534,
+      "step": 61100
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.180988183629384e-05,
+      "loss": 0.259,
+      "step": 61105
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.180714079392781e-05,
+      "loss": 0.2751,
+      "step": 61110
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.180439966317555e-05,
+      "loss": 0.2816,
+      "step": 61115
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.180165844407264e-05,
+      "loss": 0.2554,
+      "step": 61120
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1798917136654676e-05,
+      "loss": 0.2584,
+      "step": 61125
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.179617574095725e-05,
+      "loss": 0.2674,
+      "step": 61130
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1793434257015965e-05,
+      "loss": 0.272,
+      "step": 61135
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1790692684866414e-05,
+      "loss": 0.2681,
+      "step": 61140
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1787951024544196e-05,
+      "loss": 0.2902,
+      "step": 61145
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.178520927608491e-05,
+      "loss": 0.2537,
+      "step": 61150
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.178246743952414e-05,
+      "loss": 0.289,
+      "step": 61155
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.177972551489751e-05,
+      "loss": 0.2592,
+      "step": 61160
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.177698350224061e-05,
+      "loss": 0.2744,
+      "step": 61165
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1774241401589044e-05,
+      "loss": 0.2709,
+      "step": 61170
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.177149921297842e-05,
+      "loss": 0.2811,
+      "step": 61175
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.176875693644434e-05,
+      "loss": 0.2822,
+      "step": 61180
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1766014572022415e-05,
+      "loss": 0.278,
+      "step": 61185
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1763272119748244e-05,
+      "loss": 0.2664,
+      "step": 61190
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.176052957965745e-05,
+      "loss": 0.2495,
+      "step": 61195
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.175778695178563e-05,
+      "loss": 0.2562,
+      "step": 61200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.17550442361684e-05,
+      "loss": 0.2734,
+      "step": 61205
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.175230143284137e-05,
+      "loss": 0.2879,
+      "step": 61210
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.174955854184016e-05,
+      "loss": 0.264,
+      "step": 61215
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.174681556320038e-05,
+      "loss": 0.2764,
+      "step": 61220
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.174407249695764e-05,
+      "loss": 0.2588,
+      "step": 61225
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1741329343147575e-05,
+      "loss": 0.2544,
+      "step": 61230
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.173858610180578e-05,
+      "loss": 0.2865,
+      "step": 61235
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1735842772967895e-05,
+      "loss": 0.2596,
+      "step": 61240
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1733099356669514e-05,
+      "loss": 0.2852,
+      "step": 61245
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.1730355852946295e-05,
+      "loss": 0.2792,
+      "step": 61250
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1727612261833826e-05,
+      "loss": 0.2798,
+      "step": 61255
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1724868583367756e-05,
+      "loss": 0.277,
+      "step": 61260
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.172212481758369e-05,
+      "loss": 0.2579,
+      "step": 61265
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.171938096451727e-05,
+      "loss": 0.256,
+      "step": 61270
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.171663702420412e-05,
+      "loss": 0.2693,
+      "step": 61275
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.171389299667986e-05,
+      "loss": 0.2638,
+      "step": 61280
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.171114888198012e-05,
+      "loss": 0.2672,
+      "step": 61285
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.170840468014054e-05,
+      "loss": 0.2712,
+      "step": 61290
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.170566039119675e-05,
+      "loss": 0.277,
+      "step": 61295
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.170291601518436e-05,
+      "loss": 0.2836,
+      "step": 61300
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1700171552139036e-05,
+      "loss": 0.2654,
+      "step": 61305
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16974270020964e-05,
+      "loss": 0.2694,
+      "step": 61310
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1694682365092096e-05,
+      "loss": 0.249,
+      "step": 61315
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.169193764116175e-05,
+      "loss": 0.2703,
+      "step": 61320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1689192830340984e-05,
+      "loss": 0.2533,
+      "step": 61325
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1686447932665475e-05,
+      "loss": 0.2801,
+      "step": 61330
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1683702948170845e-05,
+      "loss": 0.2747,
+      "step": 61335
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1680957876892735e-05,
+      "loss": 0.2789,
+      "step": 61340
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16782127188668e-05,
+      "loss": 0.2675,
+      "step": 61345
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.167546747412866e-05,
+      "loss": 0.2792,
+      "step": 61350
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.167272214271398e-05,
+      "loss": 0.2758,
+      "step": 61355
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16699767246584e-05,
+      "loss": 0.3046,
+      "step": 61360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.166723121999757e-05,
+      "loss": 0.285,
+      "step": 61365
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1664485628767126e-05,
+      "loss": 0.2992,
+      "step": 61370
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.166173995100275e-05,
+      "loss": 0.3017,
+      "step": 61375
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.165899418674005e-05,
+      "loss": 0.3009,
+      "step": 61380
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1656248336014706e-05,
+      "loss": 0.2688,
+      "step": 61385
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1653502398862356e-05,
+      "loss": 0.297,
+      "step": 61390
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.165075637531867e-05,
+      "loss": 0.2663,
+      "step": 61395
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16480102654193e-05,
+      "loss": 0.2722,
+      "step": 61400
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1645264069199896e-05,
+      "loss": 0.2626,
+      "step": 61405
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.164251778669611e-05,
+      "loss": 0.2683,
+      "step": 61410
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1639771417943616e-05,
+      "loss": 0.2736,
+      "step": 61415
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1637024962978064e-05,
+      "loss": 0.272,
+      "step": 61420
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1634278421835116e-05,
+      "loss": 0.2703,
+      "step": 61425
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1631531794550435e-05,
+      "loss": 0.2385,
+      "step": 61430
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.162878508115968e-05,
+      "loss": 0.2926,
+      "step": 61435
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.162603828169852e-05,
+      "loss": 0.2872,
+      "step": 61440
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.162329139620262e-05,
+      "loss": 0.2672,
+      "step": 61445
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.162054442470765e-05,
+      "loss": 0.2541,
+      "step": 61450
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1617797367249274e-05,
+      "loss": 0.2762,
+      "step": 61455
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.161505022386316e-05,
+      "loss": 0.2837,
+      "step": 61460
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1612302994584976e-05,
+      "loss": 0.2656,
+      "step": 61465
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.16095556794504e-05,
+      "loss": 0.2893,
+      "step": 61470
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1606808278495094e-05,
+      "loss": 0.2888,
+      "step": 61475
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.160406079175474e-05,
+      "loss": 0.2957,
+      "step": 61480
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1601313219265e-05,
+      "loss": 0.2729,
+      "step": 61485
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.159856556106157e-05,
+      "loss": 0.2943,
+      "step": 61490
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1595817817180104e-05,
+      "loss": 0.2943,
+      "step": 61495
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.15930699876563e-05,
+      "loss": 0.2764,
+      "step": 61500
+    },
+    {
+      "epoch": 2.16,
+      "eval_loss": 0.26903456449508667,
+      "eval_runtime": 10.5402,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 61500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.159032207252582e-05,
+      "loss": 0.2633,
+      "step": 61505
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.1587574071824367e-05,
+      "loss": 0.2699,
+      "step": 61510
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.158482598558759e-05,
+      "loss": 0.265,
+      "step": 61515
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.158207781385119e-05,
+      "loss": 0.2584,
+      "step": 61520
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.157932955665085e-05,
+      "loss": 0.2815,
+      "step": 61525
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.157658121402225e-05,
+      "loss": 0.2744,
+      "step": 61530
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.157383278600108e-05,
+      "loss": 0.2673,
+      "step": 61535
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.157108427262302e-05,
+      "loss": 0.2736,
+      "step": 61540
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1568335673923756e-05,
+      "loss": 0.2542,
+      "step": 61545
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1565586989938996e-05,
+      "loss": 0.2861,
+      "step": 61550
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.15628382207044e-05,
+      "loss": 0.2817,
+      "step": 61555
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1560089366255684e-05,
+      "loss": 0.2694,
+      "step": 61560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.155734042662853e-05,
+      "loss": 0.2562,
+      "step": 61565
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1554591401858634e-05,
+      "loss": 0.2769,
+      "step": 61570
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.155184229198168e-05,
+      "loss": 0.2685,
+      "step": 61575
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.154909309703338e-05,
+      "loss": 0.2839,
+      "step": 61580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.154634381704942e-05,
+      "loss": 0.2619,
+      "step": 61585
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.15435944520655e-05,
+      "loss": 0.2865,
+      "step": 61590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.154084500211732e-05,
+      "loss": 0.2774,
+      "step": 61595
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1538095467240574e-05,
+      "loss": 0.2769,
+      "step": 61600
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.153534584747097e-05,
+      "loss": 0.2741,
+      "step": 61605
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1532596142844205e-05,
+      "loss": 0.2868,
+      "step": 61610
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1529846353395986e-05,
+      "loss": 0.2889,
+      "step": 61615
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.152709647916202e-05,
+      "loss": 0.2689,
+      "step": 61620
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.152434652017801e-05,
+      "loss": 0.2735,
+      "step": 61625
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.152159647647964e-05,
+      "loss": 0.2756,
+      "step": 61630
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.151884634810266e-05,
+      "loss": 0.2433,
+      "step": 61635
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.151609613508274e-05,
+      "loss": 0.2631,
+      "step": 61640
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.151334583745562e-05,
+      "loss": 0.2759,
+      "step": 61645
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.151059545525698e-05,
+      "loss": 0.2369,
+      "step": 61650
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.150784498852256e-05,
+      "loss": 0.2596,
+      "step": 61655
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.150509443728806e-05,
+      "loss": 0.286,
+      "step": 61660
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.15023438015892e-05,
+      "loss": 0.28,
+      "step": 61665
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.149959308146168e-05,
+      "loss": 0.2702,
+      "step": 61670
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.149684227694123e-05,
+      "loss": 0.2794,
+      "step": 61675
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1494091388063573e-05,
+      "loss": 0.278,
+      "step": 61680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.149134041486441e-05,
+      "loss": 0.2791,
+      "step": 61685
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1488589357379476e-05,
+      "loss": 0.2698,
+      "step": 61690
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.148583821564448e-05,
+      "loss": 0.2683,
+      "step": 61695
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.148308698969516e-05,
+      "loss": 0.2948,
+      "step": 61700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.148033567956722e-05,
+      "loss": 0.2722,
+      "step": 61705
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1477584285296385e-05,
+      "loss": 0.2688,
+      "step": 61710
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1474832806918395e-05,
+      "loss": 0.2561,
+      "step": 61715
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.147208124446897e-05,
+      "loss": 0.257,
+      "step": 61720
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1469329597983834e-05,
+      "loss": 0.249,
+      "step": 61725
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.146657786749871e-05,
+      "loss": 0.256,
+      "step": 61730
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.146382605304935e-05,
+      "loss": 0.2751,
+      "step": 61735
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.146107415467145e-05,
+      "loss": 0.2642,
+      "step": 61740
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1458322172400766e-05,
+      "loss": 0.2778,
+      "step": 61745
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.145557010627303e-05,
+      "loss": 0.2913,
+      "step": 61750
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1452817956323967e-05,
+      "loss": 0.2724,
+      "step": 61755
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1450065722589316e-05,
+      "loss": 0.2712,
+      "step": 61760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1447313405104805e-05,
+      "loss": 0.2703,
+      "step": 61765
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.144456100390619e-05,
+      "loss": 0.2594,
+      "step": 61770
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.144180851902919e-05,
+      "loss": 0.2719,
+      "step": 61775
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.143905595050956e-05,
+      "loss": 0.2843,
+      "step": 61780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1436303298383027e-05,
+      "loss": 0.2774,
+      "step": 61785
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.1433550562685335e-05,
+      "loss": 0.2669,
+      "step": 61790
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.143079774345223e-05,
+      "loss": 0.2658,
+      "step": 61795
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.142804484071945e-05,
+      "loss": 0.2773,
+      "step": 61800
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.142529185452275e-05,
+      "loss": 0.2765,
+      "step": 61805
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.142253878489787e-05,
+      "loss": 0.2758,
+      "step": 61810
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.141978563188056e-05,
+      "loss": 0.2643,
+      "step": 61815
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.141703239550656e-05,
+      "loss": 0.2776,
+      "step": 61820
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.141427907581162e-05,
+      "loss": 0.2638,
+      "step": 61825
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.141152567283149e-05,
+      "loss": 0.2818,
+      "step": 61830
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.140877218660194e-05,
+      "loss": 0.2655,
+      "step": 61835
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1406018617158695e-05,
+      "loss": 0.2551,
+      "step": 61840
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1403264964537525e-05,
+      "loss": 0.2567,
+      "step": 61845
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.140051122877417e-05,
+      "loss": 0.2728,
+      "step": 61850
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1397757409904405e-05,
+      "loss": 0.2919,
+      "step": 61855
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.139500350796397e-05,
+      "loss": 0.2607,
+      "step": 61860
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1392249522988636e-05,
+      "loss": 0.2678,
+      "step": 61865
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1389495455014154e-05,
+      "loss": 0.2838,
+      "step": 61870
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.138674130407628e-05,
+      "loss": 0.265,
+      "step": 61875
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.138398707021077e-05,
+      "loss": 0.2934,
+      "step": 61880
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1381232753453416e-05,
+      "loss": 0.2583,
+      "step": 61885
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.137847835383995e-05,
+      "loss": 0.2492,
+      "step": 61890
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1375723871406146e-05,
+      "loss": 0.2666,
+      "step": 61895
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.137296930618777e-05,
+      "loss": 0.2915,
+      "step": 61900
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1370214658220574e-05,
+      "loss": 0.2795,
+      "step": 61905
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1367459927540354e-05,
+      "loss": 0.2567,
+      "step": 61910
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1364705114182855e-05,
+      "loss": 0.2675,
+      "step": 61915
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.136195021818387e-05,
+      "loss": 0.2815,
+      "step": 61920
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1359195239579145e-05,
+      "loss": 0.2679,
+      "step": 61925
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.135644017840445e-05,
+      "loss": 0.2546,
+      "step": 61930
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.135368503469558e-05,
+      "loss": 0.2836,
+      "step": 61935
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.13509298084883e-05,
+      "loss": 0.2724,
+      "step": 61940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1348174499818375e-05,
+      "loss": 0.2578,
+      "step": 61945
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.134541910872159e-05,
+      "loss": 0.2742,
+      "step": 61950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1342663635233713e-05,
+      "loss": 0.267,
+      "step": 61955
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.133990807939053e-05,
+      "loss": 0.2694,
+      "step": 61960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.133715244122783e-05,
+      "loss": 0.2745,
+      "step": 61965
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.133439672078138e-05,
+      "loss": 0.2624,
+      "step": 61970
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1331640918086956e-05,
+      "loss": 0.2828,
+      "step": 61975
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.132888503318036e-05,
+      "loss": 0.2696,
+      "step": 61980
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.132612906609734e-05,
+      "loss": 0.2563,
+      "step": 61985
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1323373016873715e-05,
+      "loss": 0.2596,
+      "step": 61990
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.132061688554526e-05,
+      "loss": 0.2638,
+      "step": 61995
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.131786067214777e-05,
+      "loss": 0.29,
+      "step": 62000
+    },
+    {
+      "epoch": 2.18,
+      "eval_loss": 0.2678808569908142,
+      "eval_runtime": 10.5389,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 62000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1315104376717e-05,
+      "loss": 0.3023,
+      "step": 62005
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1312347999288777e-05,
+      "loss": 0.2886,
+      "step": 62010
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1309591539898866e-05,
+      "loss": 0.2419,
+      "step": 62015
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.130683499858308e-05,
+      "loss": 0.2869,
+      "step": 62020
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.130407837537719e-05,
+      "loss": 0.2772,
+      "step": 62025
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1301321670316994e-05,
+      "loss": 0.2674,
+      "step": 62030
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1298564883438295e-05,
+      "loss": 0.2838,
+      "step": 62035
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1295808014776884e-05,
+      "loss": 0.2683,
+      "step": 62040
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.129305106436855e-05,
+      "loss": 0.2963,
+      "step": 62045
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.129029403224909e-05,
+      "loss": 0.2604,
+      "step": 62050
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.128753691845432e-05,
+      "loss": 0.2684,
+      "step": 62055
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.128477972302001e-05,
+      "loss": 0.2561,
+      "step": 62060
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.128202244598199e-05,
+      "loss": 0.2732,
+      "step": 62065
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1279265087376044e-05,
+      "loss": 0.2673,
+      "step": 62070
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1276507647237984e-05,
+      "loss": 0.2787,
+      "step": 62075
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1273750125603604e-05,
+      "loss": 0.2969,
+      "step": 62080
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.127099252250872e-05,
+      "loss": 0.2773,
+      "step": 62085
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.126823483798912e-05,
+      "loss": 0.251,
+      "step": 62090
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.126547707208063e-05,
+      "loss": 0.27,
+      "step": 62095
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.1262719224819044e-05,
+      "loss": 0.2743,
+      "step": 62100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.125996129624018e-05,
+      "loss": 0.2802,
+      "step": 62105
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1257203286379845e-05,
+      "loss": 0.2797,
+      "step": 62110
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1254445195273844e-05,
+      "loss": 0.2787,
+      "step": 62115
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1251687022958e-05,
+      "loss": 0.266,
+      "step": 62120
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.124892876946811e-05,
+      "loss": 0.2657,
+      "step": 62125
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.124617043484001e-05,
+      "loss": 0.2578,
+      "step": 62130
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1243412019109496e-05,
+      "loss": 0.2822,
+      "step": 62135
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.124065352231239e-05,
+      "loss": 0.2852,
+      "step": 62140
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.123789494448451e-05,
+      "loss": 0.2712,
+      "step": 62145
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.123513628566168e-05,
+      "loss": 0.273,
+      "step": 62150
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.123237754587971e-05,
+      "loss": 0.2482,
+      "step": 62155
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1229618725174436e-05,
+      "loss": 0.2561,
+      "step": 62160
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.122685982358165e-05,
+      "loss": 0.2584,
+      "step": 62165
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1224100841137217e-05,
+      "loss": 0.2677,
+      "step": 62170
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.122134177787692e-05,
+      "loss": 0.2588,
+      "step": 62175
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.12185826338366e-05,
+      "loss": 0.2661,
+      "step": 62180
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.121582340905209e-05,
+      "loss": 0.2692,
+      "step": 62185
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1213064103559206e-05,
+      "loss": 0.2429,
+      "step": 62190
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1210304717393775e-05,
+      "loss": 0.2833,
+      "step": 62195
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.120754525059163e-05,
+      "loss": 0.2604,
+      "step": 62200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.12047857031886e-05,
+      "loss": 0.2753,
+      "step": 62205
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1202026075220525e-05,
+      "loss": 0.2935,
+      "step": 62210
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1199266366723224e-05,
+      "loss": 0.257,
+      "step": 62215
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1196506577732545e-05,
+      "loss": 0.2837,
+      "step": 62220
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1193746708284294e-05,
+      "loss": 0.2723,
+      "step": 62225
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.119098675841433e-05,
+      "loss": 0.2625,
+      "step": 62230
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1188226728158485e-05,
+      "loss": 0.2771,
+      "step": 62235
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1185466617552594e-05,
+      "loss": 0.2816,
+      "step": 62240
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1182706426632496e-05,
+      "loss": 0.2625,
+      "step": 62245
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.117994615543403e-05,
+      "loss": 0.2642,
+      "step": 62250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.117718580399304e-05,
+      "loss": 0.2604,
+      "step": 62255
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.117442537234535e-05,
+      "loss": 0.2451,
+      "step": 62260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1171664860526826e-05,
+      "loss": 0.2944,
+      "step": 62265
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1168904268573294e-05,
+      "loss": 0.2968,
+      "step": 62270
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.116614359652061e-05,
+      "loss": 0.2615,
+      "step": 62275
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.116338284440462e-05,
+      "loss": 0.2496,
+      "step": 62280
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.116062201226116e-05,
+      "loss": 0.2591,
+      "step": 62285
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.115786110012608e-05,
+      "loss": 0.265,
+      "step": 62290
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.115510010803524e-05,
+      "loss": 0.2671,
+      "step": 62295
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.115233903602448e-05,
+      "loss": 0.2622,
+      "step": 62300
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.114957788412965e-05,
+      "loss": 0.266,
+      "step": 62305
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.11468166523866e-05,
+      "loss": 0.3061,
+      "step": 62310
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.114405534083119e-05,
+      "loss": 0.2594,
+      "step": 62315
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.114129394949927e-05,
+      "loss": 0.2714,
+      "step": 62320
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.113853247842669e-05,
+      "loss": 0.285,
+      "step": 62325
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.113577092764932e-05,
+      "loss": 0.2718,
+      "step": 62330
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.113300929720302e-05,
+      "loss": 0.2715,
+      "step": 62335
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.113024758712361e-05,
+      "loss": 0.2728,
+      "step": 62340
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.112748579744699e-05,
+      "loss": 0.264,
+      "step": 62345
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1124723928209e-05,
+      "loss": 0.2649,
+      "step": 62350
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1121961979445506e-05,
+      "loss": 0.2562,
+      "step": 62355
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.111919995119237e-05,
+      "loss": 0.2866,
+      "step": 62360
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1116437843485465e-05,
+      "loss": 0.2639,
+      "step": 62365
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.111367565636063e-05,
+      "loss": 0.2655,
+      "step": 62370
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.111091338985376e-05,
+      "loss": 0.2735,
+      "step": 62375
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.110815104400069e-05,
+      "loss": 0.2723,
+      "step": 62380
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.1105388618837314e-05,
+      "loss": 0.3042,
+      "step": 62385
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.11026261143995e-05,
+      "loss": 0.2662,
+      "step": 62390
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.109986353072309e-05,
+      "loss": 0.2803,
+      "step": 62395
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.109710086784398e-05,
+      "loss": 0.28,
+      "step": 62400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1094338125798035e-05,
+      "loss": 0.2713,
+      "step": 62405
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1091575304621126e-05,
+      "loss": 0.2577,
+      "step": 62410
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.108881240434912e-05,
+      "loss": 0.2785,
+      "step": 62415
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.108604942501791e-05,
+      "loss": 0.2456,
+      "step": 62420
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.108328636666335e-05,
+      "loss": 0.2876,
+      "step": 62425
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.108052322932132e-05,
+      "loss": 0.2638,
+      "step": 62430
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.107776001302771e-05,
+      "loss": 0.2657,
+      "step": 62435
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.10749967178184e-05,
+      "loss": 0.2973,
+      "step": 62440
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.107223334372925e-05,
+      "loss": 0.2855,
+      "step": 62445
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.106946989079616e-05,
+      "loss": 0.2855,
+      "step": 62450
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1066706359054984e-05,
+      "loss": 0.2701,
+      "step": 62455
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1063942748541644e-05,
+      "loss": 0.2738,
+      "step": 62460
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1061179059292e-05,
+      "loss": 0.2961,
+      "step": 62465
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1058415291341944e-05,
+      "loss": 0.2745,
+      "step": 62470
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1055651444727345e-05,
+      "loss": 0.2487,
+      "step": 62475
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.105288751948411e-05,
+      "loss": 0.2876,
+      "step": 62480
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.105012351564812e-05,
+      "loss": 0.2709,
+      "step": 62485
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1047359433255263e-05,
+      "loss": 0.2986,
+      "step": 62490
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1044595272341425e-05,
+      "loss": 0.2757,
+      "step": 62495
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.104183103294251e-05,
+      "loss": 0.2839,
+      "step": 62500
+    },
+    {
+      "epoch": 2.2,
+      "eval_loss": 0.2684471011161804,
+      "eval_runtime": 10.5326,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 62500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.103906671509439e-05,
+      "loss": 0.2537,
+      "step": 62505
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.103630231883297e-05,
+      "loss": 0.2632,
+      "step": 62510
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1033537844194155e-05,
+      "loss": 0.2969,
+      "step": 62515
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.103077329121382e-05,
+      "loss": 0.2819,
+      "step": 62520
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1028008659927866e-05,
+      "loss": 0.237,
+      "step": 62525
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.102524395037219e-05,
+      "loss": 0.2558,
+      "step": 62530
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1022479162582695e-05,
+      "loss": 0.2682,
+      "step": 62535
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.101971429659528e-05,
+      "loss": 0.249,
+      "step": 62540
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.101694935244584e-05,
+      "loss": 0.2778,
+      "step": 62545
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.101418433017027e-05,
+      "loss": 0.2924,
+      "step": 62550
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.10114192298045e-05,
+      "loss": 0.2784,
+      "step": 62555
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.10086540513844e-05,
+      "loss": 0.2703,
+      "step": 62560
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1005888794945884e-05,
+      "loss": 0.285,
+      "step": 62565
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.1003123460524865e-05,
+      "loss": 0.2675,
+      "step": 62570
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.100035804815724e-05,
+      "loss": 0.2688,
+      "step": 62575
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0997592557878926e-05,
+      "loss": 0.2583,
+      "step": 62580
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.099482698972582e-05,
+      "loss": 0.2913,
+      "step": 62585
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0992061343733835e-05,
+      "loss": 0.308,
+      "step": 62590
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.098929561993889e-05,
+      "loss": 0.274,
+      "step": 62595
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.098652981837688e-05,
+      "loss": 0.2655,
+      "step": 62600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.098376393908373e-05,
+      "loss": 0.2737,
+      "step": 62605
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.098099798209535e-05,
+      "loss": 0.2755,
+      "step": 62610
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.097823194744765e-05,
+      "loss": 0.2636,
+      "step": 62615
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0975465835176545e-05,
+      "loss": 0.2543,
+      "step": 62620
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.097269964531796e-05,
+      "loss": 0.2555,
+      "step": 62625
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.096993337790781e-05,
+      "loss": 0.2465,
+      "step": 62630
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.0967167032982e-05,
+      "loss": 0.2686,
+      "step": 62635
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.096440061057647e-05,
+      "loss": 0.2759,
+      "step": 62640
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.096163411072711e-05,
+      "loss": 0.2765,
+      "step": 62645
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.095886753346988e-05,
+      "loss": 0.2789,
+      "step": 62650
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.095610087884067e-05,
+      "loss": 0.2703,
+      "step": 62655
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.095333414687542e-05,
+      "loss": 0.2798,
+      "step": 62660
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.095056733761005e-05,
+      "loss": 0.274,
+      "step": 62665
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.094780045108048e-05,
+      "loss": 0.2811,
+      "step": 62670
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.094503348732264e-05,
+      "loss": 0.2785,
+      "step": 62675
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.094226644637246e-05,
+      "loss": 0.2748,
+      "step": 62680
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.093949932826587e-05,
+      "loss": 0.2676,
+      "step": 62685
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0936732133038795e-05,
+      "loss": 0.2652,
+      "step": 62690
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0933964860727167e-05,
+      "loss": 0.2779,
+      "step": 62695
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0931197511366914e-05,
+      "loss": 0.2713,
+      "step": 62700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0928430084993965e-05,
+      "loss": 0.2618,
+      "step": 62705
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.092566258164427e-05,
+      "loss": 0.2765,
+      "step": 62710
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0922895001353736e-05,
+      "loss": 0.2803,
+      "step": 62715
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0920127344158326e-05,
+      "loss": 0.2767,
+      "step": 62720
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.091735961009396e-05,
+      "loss": 0.2781,
+      "step": 62725
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.091459179919657e-05,
+      "loss": 0.2569,
+      "step": 62730
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0911823911502116e-05,
+      "loss": 0.2774,
+      "step": 62735
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.090905594704652e-05,
+      "loss": 0.2852,
+      "step": 62740
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.090628790586573e-05,
+      "loss": 0.2657,
+      "step": 62745
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0903519787995684e-05,
+      "loss": 0.2806,
+      "step": 62750
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.090075159347232e-05,
+      "loss": 0.2882,
+      "step": 62755
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0897983322331586e-05,
+      "loss": 0.2767,
+      "step": 62760
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0895214974609425e-05,
+      "loss": 0.2908,
+      "step": 62765
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0892446550341783e-05,
+      "loss": 0.2871,
+      "step": 62770
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.088967804956461e-05,
+      "loss": 0.2766,
+      "step": 62775
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.088690947231385e-05,
+      "loss": 0.2849,
+      "step": 62780
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.088414081862544e-05,
+      "loss": 0.2766,
+      "step": 62785
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.088137208853535e-05,
+      "loss": 0.252,
+      "step": 62790
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0878603282079516e-05,
+      "loss": 0.2417,
+      "step": 62795
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.087583439929389e-05,
+      "loss": 0.2696,
+      "step": 62800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.087306544021443e-05,
+      "loss": 0.2914,
+      "step": 62805
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.087029640487708e-05,
+      "loss": 0.2581,
+      "step": 62810
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.08675272933178e-05,
+      "loss": 0.2741,
+      "step": 62815
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.086475810557255e-05,
+      "loss": 0.2921,
+      "step": 62820
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.086198884167728e-05,
+      "loss": 0.2578,
+      "step": 62825
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0859219501667944e-05,
+      "loss": 0.2567,
+      "step": 62830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.085645008558051e-05,
+      "loss": 0.2587,
+      "step": 62835
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.085368059345093e-05,
+      "loss": 0.2745,
+      "step": 62840
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.085091102531516e-05,
+      "loss": 0.2569,
+      "step": 62845
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.084814138120917e-05,
+      "loss": 0.279,
+      "step": 62850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.084537166116892e-05,
+      "loss": 0.2621,
+      "step": 62855
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.084260186523038e-05,
+      "loss": 0.2597,
+      "step": 62860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.083983199342949e-05,
+      "loss": 0.2743,
+      "step": 62865
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0837062045802235e-05,
+      "loss": 0.2986,
+      "step": 62870
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.083429202238457e-05,
+      "loss": 0.2834,
+      "step": 62875
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0831521923212484e-05,
+      "loss": 0.2812,
+      "step": 62880
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0828751748321914e-05,
+      "loss": 0.2694,
+      "step": 62885
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.082598149774884e-05,
+      "loss": 0.2614,
+      "step": 62890
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.082321117152925e-05,
+      "loss": 0.2899,
+      "step": 62895
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.08204407696991e-05,
+      "loss": 0.2703,
+      "step": 62900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.081767029229435e-05,
+      "loss": 0.2616,
+      "step": 62905
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0814899739350996e-05,
+      "loss": 0.2762,
+      "step": 62910
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0812129110905e-05,
+      "loss": 0.2624,
+      "step": 62915
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.080935840699234e-05,
+      "loss": 0.2833,
+      "step": 62920
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.080658762764898e-05,
+      "loss": 0.2786,
+      "step": 62925
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.080381677291091e-05,
+      "loss": 0.2778,
+      "step": 62930
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.080104584281411e-05,
+      "loss": 0.27,
+      "step": 62935
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0798274837394554e-05,
+      "loss": 0.2539,
+      "step": 62940
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.079550375668821e-05,
+      "loss": 0.2617,
+      "step": 62945
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.0792732600731075e-05,
+      "loss": 0.2757,
+      "step": 62950
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.078996136955912e-05,
+      "loss": 0.2698,
+      "step": 62955
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.078719006320834e-05,
+      "loss": 0.2779,
+      "step": 62960
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.078441868171471e-05,
+      "loss": 0.2606,
+      "step": 62965
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.078164722511421e-05,
+      "loss": 0.2839,
+      "step": 62970
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.077887569344283e-05,
+      "loss": 0.2678,
+      "step": 62975
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0776104086736556e-05,
+      "loss": 0.2597,
+      "step": 62980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.077333240503139e-05,
+      "loss": 0.2734,
+      "step": 62985
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.07705606483633e-05,
+      "loss": 0.2606,
+      "step": 62990
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0767788816768285e-05,
+      "loss": 0.2728,
+      "step": 62995
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.076501691028233e-05,
+      "loss": 0.2611,
+      "step": 63000
+    },
+    {
+      "epoch": 2.22,
+      "eval_loss": 0.2684405446052551,
+      "eval_runtime": 10.5301,
+      "eval_samples_per_second": 9.497,
+      "eval_steps_per_second": 9.497,
+      "step": 63000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.076224492894143e-05,
+      "loss": 0.2829,
+      "step": 63005
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0759472872781576e-05,
+      "loss": 0.2912,
+      "step": 63010
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.075670074183877e-05,
+      "loss": 0.2739,
+      "step": 63015
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0753928536148986e-05,
+      "loss": 0.2833,
+      "step": 63020
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.075115625574824e-05,
+      "loss": 0.2689,
+      "step": 63025
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.074838390067252e-05,
+      "loss": 0.2743,
+      "step": 63030
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.074561147095782e-05,
+      "loss": 0.2613,
+      "step": 63035
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.074283896664014e-05,
+      "loss": 0.2802,
+      "step": 63040
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0740066387755496e-05,
+      "loss": 0.2777,
+      "step": 63045
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.073729373433986e-05,
+      "loss": 0.2518,
+      "step": 63050
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.073452100642924e-05,
+      "loss": 0.2616,
+      "step": 63055
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.073174820405965e-05,
+      "loss": 0.277,
+      "step": 63060
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.07289753272671e-05,
+      "loss": 0.3092,
+      "step": 63065
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.072620237608756e-05,
+      "loss": 0.2681,
+      "step": 63070
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.072342935055707e-05,
+      "loss": 0.2587,
+      "step": 63075
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0720656250711614e-05,
+      "loss": 0.2689,
+      "step": 63080
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.07178830765872e-05,
+      "loss": 0.2684,
+      "step": 63085
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0715109828219856e-05,
+      "loss": 0.2542,
+      "step": 63090
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0712336505645573e-05,
+      "loss": 0.2734,
+      "step": 63095
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.070956310890037e-05,
+      "loss": 0.2789,
+      "step": 63100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.070678963802025e-05,
+      "loss": 0.2828,
+      "step": 63105
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.070401609304122e-05,
+      "loss": 0.3018,
+      "step": 63110
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0701242473999304e-05,
+      "loss": 0.2615,
+      "step": 63115
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.069846878093052e-05,
+      "loss": 0.2628,
+      "step": 63120
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.069569501387086e-05,
+      "loss": 0.2609,
+      "step": 63125
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.069292117285637e-05,
+      "loss": 0.2709,
+      "step": 63130
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.069014725792304e-05,
+      "loss": 0.294,
+      "step": 63135
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0687373269106894e-05,
+      "loss": 0.2662,
+      "step": 63140
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.068459920644396e-05,
+      "loss": 0.289,
+      "step": 63145
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.068182506997026e-05,
+      "loss": 0.2689,
+      "step": 63150
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.067905085972179e-05,
+      "loss": 0.2564,
+      "step": 63155
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.06762765757346e-05,
+      "loss": 0.2651,
+      "step": 63160
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0673502218044694e-05,
+      "loss": 0.2759,
+      "step": 63165
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.067072778668809e-05,
+      "loss": 0.2634,
+      "step": 63170
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0667953281700834e-05,
+      "loss": 0.2703,
+      "step": 63175
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0665178703118945e-05,
+      "loss": 0.261,
+      "step": 63180
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.066240405097844e-05,
+      "loss": 0.257,
+      "step": 63185
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.065962932531534e-05,
+      "loss": 0.2566,
+      "step": 63190
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0656854526165695e-05,
+      "loss": 0.2697,
+      "step": 63195
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.065407965356552e-05,
+      "loss": 0.2739,
+      "step": 63200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0651304707550846e-05,
+      "loss": 0.2766,
+      "step": 63205
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0648529688157704e-05,
+      "loss": 0.2808,
+      "step": 63210
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.064575459542213e-05,
+      "loss": 0.2638,
+      "step": 63215
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0642979429380144e-05,
+      "loss": 0.2747,
+      "step": 63220
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.064020419006779e-05,
+      "loss": 0.2562,
+      "step": 63225
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.0637428877521115e-05,
+      "loss": 0.2634,
+      "step": 63230
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.063465349177613e-05,
+      "loss": 0.2749,
+      "step": 63235
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.063187803286889e-05,
+      "loss": 0.2611,
+      "step": 63240
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.062910250083542e-05,
+      "loss": 0.274,
+      "step": 63245
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.062632689571176e-05,
+      "loss": 0.2638,
+      "step": 63250
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.062355121753396e-05,
+      "loss": 0.2689,
+      "step": 63255
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.062077546633805e-05,
+      "loss": 0.2442,
+      "step": 63260
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.061799964216008e-05,
+      "loss": 0.278,
+      "step": 63265
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.061522374503608e-05,
+      "loss": 0.2693,
+      "step": 63270
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.06124477750021e-05,
+      "loss": 0.2743,
+      "step": 63275
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.060967173209419e-05,
+      "loss": 0.2704,
+      "step": 63280
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.060689561634838e-05,
+      "loss": 0.25,
+      "step": 63285
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.060411942780074e-05,
+      "loss": 0.2897,
+      "step": 63290
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.06013431664873e-05,
+      "loss": 0.2894,
+      "step": 63295
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.05985668324441e-05,
+      "loss": 0.2574,
+      "step": 63300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.059579042570721e-05,
+      "loss": 0.2719,
+      "step": 63305
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.059301394631266e-05,
+      "loss": 0.2507,
+      "step": 63310
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.059023739429651e-05,
+      "loss": 0.2601,
+      "step": 63315
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.058746076969482e-05,
+      "loss": 0.2822,
+      "step": 63320
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.058468407254363e-05,
+      "loss": 0.2661,
+      "step": 63325
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.058190730287899e-05,
+      "loss": 0.2667,
+      "step": 63330
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0579130460736974e-05,
+      "loss": 0.2643,
+      "step": 63335
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.057635354615362e-05,
+      "loss": 0.2804,
+      "step": 63340
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.057357655916499e-05,
+      "loss": 0.2649,
+      "step": 63345
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0570799499807145e-05,
+      "loss": 0.2675,
+      "step": 63350
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.056802236811614e-05,
+      "loss": 0.2487,
+      "step": 63355
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.056524516412803e-05,
+      "loss": 0.2895,
+      "step": 63360
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0562467887878876e-05,
+      "loss": 0.2858,
+      "step": 63365
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.055969053940475e-05,
+      "loss": 0.2805,
+      "step": 63370
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.05569131187417e-05,
+      "loss": 0.2549,
+      "step": 63375
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0554135625925806e-05,
+      "loss": 0.2794,
+      "step": 63380
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.055135806099311e-05,
+      "loss": 0.2655,
+      "step": 63385
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.054858042397969e-05,
+      "loss": 0.303,
+      "step": 63390
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.054580271492161e-05,
+      "loss": 0.2728,
+      "step": 63395
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0543024933854944e-05,
+      "loss": 0.2585,
+      "step": 63400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.054024708081575e-05,
+      "loss": 0.2828,
+      "step": 63405
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0537469155840095e-05,
+      "loss": 0.2673,
+      "step": 63410
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.053469115896405e-05,
+      "loss": 0.2639,
+      "step": 63415
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0531913090223693e-05,
+      "loss": 0.2824,
+      "step": 63420
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.052913494965509e-05,
+      "loss": 0.2922,
+      "step": 63425
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0526356737294306e-05,
+      "loss": 0.2759,
+      "step": 63430
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.052357845317743e-05,
+      "loss": 0.2859,
+      "step": 63435
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0520800097340526e-05,
+      "loss": 0.2588,
+      "step": 63440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.051802166981967e-05,
+      "loss": 0.2786,
+      "step": 63445
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.051524317065093e-05,
+      "loss": 0.2616,
+      "step": 63450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.051246459987041e-05,
+      "loss": 0.2702,
+      "step": 63455
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.050968595751416e-05,
+      "loss": 0.2691,
+      "step": 63460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.050690724361827e-05,
+      "loss": 0.2864,
+      "step": 63465
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0504128458218815e-05,
+      "loss": 0.2618,
+      "step": 63470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0501349601351887e-05,
+      "loss": 0.2828,
+      "step": 63475
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0498570673053557e-05,
+      "loss": 0.2864,
+      "step": 63480
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.049579167335991e-05,
+      "loss": 0.2753,
+      "step": 63485
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.049301260230703e-05,
+      "loss": 0.2759,
+      "step": 63490
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0490233459931e-05,
+      "loss": 0.2643,
+      "step": 63495
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0487454246267904e-05,
+      "loss": 0.2826,
+      "step": 63500
+    },
+    {
+      "epoch": 2.23,
+      "eval_loss": 0.26817554235458374,
+      "eval_runtime": 10.5266,
+      "eval_samples_per_second": 9.5,
+      "eval_steps_per_second": 9.5,
+      "step": 63500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.048467496135384e-05,
+      "loss": 0.2642,
+      "step": 63505
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.048189560522488e-05,
+      "loss": 0.2709,
+      "step": 63510
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.047911617791712e-05,
+      "loss": 0.2787,
+      "step": 63515
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.0476336679466645e-05,
+      "loss": 0.2723,
+      "step": 63520
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.047355710990955e-05,
+      "loss": 0.2784,
+      "step": 63525
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0470777469281918e-05,
+      "loss": 0.2844,
+      "step": 63530
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0467997757619853e-05,
+      "loss": 0.2933,
+      "step": 63535
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0465217974959436e-05,
+      "loss": 0.2765,
+      "step": 63540
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0462438121336767e-05,
+      "loss": 0.291,
+      "step": 63545
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0459658196787937e-05,
+      "loss": 0.2678,
+      "step": 63550
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.045687820134904e-05,
+      "loss": 0.2615,
+      "step": 63555
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0454098135056186e-05,
+      "loss": 0.2461,
+      "step": 63560
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0451317997945454e-05,
+      "loss": 0.2817,
+      "step": 63565
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0448537790052962e-05,
+      "loss": 0.264,
+      "step": 63570
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0445757511414785e-05,
+      "loss": 0.2779,
+      "step": 63575
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.044297716206704e-05,
+      "loss": 0.2688,
+      "step": 63580
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.044019674204582e-05,
+      "loss": 0.2604,
+      "step": 63585
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0437416251387234e-05,
+      "loss": 0.2821,
+      "step": 63590
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0434635690127382e-05,
+      "loss": 0.2489,
+      "step": 63595
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.043185505830237e-05,
+      "loss": 0.2696,
+      "step": 63600
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0429074355948295e-05,
+      "loss": 0.2913,
+      "step": 63605
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0426293583101267e-05,
+      "loss": 0.2934,
+      "step": 63610
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0423512739797387e-05,
+      "loss": 0.2659,
+      "step": 63615
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0420731826072785e-05,
+      "loss": 0.2681,
+      "step": 63620
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.041795084196354e-05,
+      "loss": 0.2589,
+      "step": 63625
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0415169787505777e-05,
+      "loss": 0.2642,
+      "step": 63630
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.04123886627356e-05,
+      "loss": 0.2544,
+      "step": 63635
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0409607467689127e-05,
+      "loss": 0.2651,
+      "step": 63640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0406826202402466e-05,
+      "loss": 0.2479,
+      "step": 63645
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0404044866911736e-05,
+      "loss": 0.2829,
+      "step": 63650
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0401263461253033e-05,
+      "loss": 0.2997,
+      "step": 63655
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.039848198546249e-05,
+      "loss": 0.2897,
+      "step": 63660
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.039570043957621e-05,
+      "loss": 0.278,
+      "step": 63665
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.039291882363032e-05,
+      "loss": 0.2656,
+      "step": 63670
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0390137137660933e-05,
+      "loss": 0.2906,
+      "step": 63675
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.038735538170417e-05,
+      "loss": 0.28,
+      "step": 63680
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.038457355579615e-05,
+      "loss": 0.2634,
+      "step": 63685
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.038179165997298e-05,
+      "loss": 0.2563,
+      "step": 63690
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.03790096942708e-05,
+      "loss": 0.2743,
+      "step": 63695
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.037622765872572e-05,
+      "loss": 0.2518,
+      "step": 63700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0373445553373874e-05,
+      "loss": 0.2593,
+      "step": 63705
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0370663378251374e-05,
+      "loss": 0.2665,
+      "step": 63710
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.036788113339435e-05,
+      "loss": 0.2536,
+      "step": 63715
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0365098818838923e-05,
+      "loss": 0.2749,
+      "step": 63720
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.036231643462123e-05,
+      "loss": 0.2811,
+      "step": 63725
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.035953398077739e-05,
+      "loss": 0.2634,
+      "step": 63730
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0356751457343534e-05,
+      "loss": 0.2931,
+      "step": 63735
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.035396886435579e-05,
+      "loss": 0.2714,
+      "step": 63740
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0351186201850286e-05,
+      "loss": 0.2481,
+      "step": 63745
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0348403469863156e-05,
+      "loss": 0.2698,
+      "step": 63750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0345620668430535e-05,
+      "loss": 0.2541,
+      "step": 63755
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0342837797588552e-05,
+      "loss": 0.2757,
+      "step": 63760
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0340054857373344e-05,
+      "loss": 0.255,
+      "step": 63765
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.033727184782103e-05,
+      "loss": 0.2832,
+      "step": 63770
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.033448876896777e-05,
+      "loss": 0.2753,
+      "step": 63775
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.033170562084969e-05,
+      "loss": 0.2628,
+      "step": 63780
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.032892240350292e-05,
+      "loss": 0.277,
+      "step": 63785
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.032613911696361e-05,
+      "loss": 0.3037,
+      "step": 63790
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.0323355761267892e-05,
+      "loss": 0.2693,
+      "step": 63795
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.03205723364519e-05,
+      "loss": 0.2608,
+      "step": 63800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.031778884255179e-05,
+      "loss": 0.2782,
+      "step": 63805
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0315005279603696e-05,
+      "loss": 0.2889,
+      "step": 63810
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.031222164764376e-05,
+      "loss": 0.2652,
+      "step": 63815
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.030943794670813e-05,
+      "loss": 0.2428,
+      "step": 63820
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.030665417683295e-05,
+      "loss": 0.2678,
+      "step": 63825
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0303870338054354e-05,
+      "loss": 0.2648,
+      "step": 63830
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0301086430408498e-05,
+      "loss": 0.2691,
+      "step": 63835
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0298302453931533e-05,
+      "loss": 0.2628,
+      "step": 63840
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.02955184086596e-05,
+      "loss": 0.2702,
+      "step": 63845
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.029273429462885e-05,
+      "loss": 0.2707,
+      "step": 63850
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.028995011187543e-05,
+      "loss": 0.29,
+      "step": 63855
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.02871658604355e-05,
+      "loss": 0.2597,
+      "step": 63860
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0284381540345203e-05,
+      "loss": 0.2668,
+      "step": 63865
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.02815971516407e-05,
+      "loss": 0.2517,
+      "step": 63870
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0278812694358127e-05,
+      "loss": 0.2788,
+      "step": 63875
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.027602816853366e-05,
+      "loss": 0.2658,
+      "step": 63880
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0273243574203436e-05,
+      "loss": 0.2614,
+      "step": 63885
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0270458911403625e-05,
+      "loss": 0.2683,
+      "step": 63890
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.026767418017038e-05,
+      "loss": 0.2708,
+      "step": 63895
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.026488938053985e-05,
+      "loss": 0.272,
+      "step": 63900
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0262104512548212e-05,
+      "loss": 0.2595,
+      "step": 63905
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0259319576231603e-05,
+      "loss": 0.2552,
+      "step": 63910
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0256534571626206e-05,
+      "loss": 0.2814,
+      "step": 63915
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0253749498768163e-05,
+      "loss": 0.2735,
+      "step": 63920
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0250964357693652e-05,
+      "loss": 0.2509,
+      "step": 63925
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0248179148438826e-05,
+      "loss": 0.2908,
+      "step": 63930
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.024539387103985e-05,
+      "loss": 0.2475,
+      "step": 63935
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.024260852553289e-05,
+      "loss": 0.2848,
+      "step": 63940
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0239823111954124e-05,
+      "loss": 0.258,
+      "step": 63945
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.02370376303397e-05,
+      "loss": 0.2805,
+      "step": 63950
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0234252080725796e-05,
+      "loss": 0.2801,
+      "step": 63955
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0231466463148577e-05,
+      "loss": 0.2646,
+      "step": 63960
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0228680777644214e-05,
+      "loss": 0.2717,
+      "step": 63965
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0225895024248874e-05,
+      "loss": 0.2827,
+      "step": 63970
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0223109202998734e-05,
+      "loss": 0.2582,
+      "step": 63975
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0220323313929965e-05,
+      "loss": 0.2712,
+      "step": 63980
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.021753735707874e-05,
+      "loss": 0.2704,
+      "step": 63985
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0214751332481218e-05,
+      "loss": 0.2845,
+      "step": 63990
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.02119652401736e-05,
+      "loss": 0.2768,
+      "step": 63995
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.020917908019204e-05,
+      "loss": 0.2669,
+      "step": 64000
+    },
+    {
+      "epoch": 2.25,
+      "eval_loss": 0.26856276392936707,
+      "eval_runtime": 10.5405,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 64000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0206392852572733e-05,
+      "loss": 0.2684,
+      "step": 64005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0203606557351836e-05,
+      "loss": 0.2997,
+      "step": 64010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0200820194565537e-05,
+      "loss": 0.297,
+      "step": 64015
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0198033764250023e-05,
+      "loss": 0.2825,
+      "step": 64020
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0195247266441458e-05,
+      "loss": 0.2744,
+      "step": 64025
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.019246070117604e-05,
+      "loss": 0.2528,
+      "step": 64030
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.018967406848993e-05,
+      "loss": 0.2631,
+      "step": 64035
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0186887368419337e-05,
+      "loss": 0.2773,
+      "step": 64040
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.018410060100042e-05,
+      "loss": 0.2583,
+      "step": 64045
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0181313766269376e-05,
+      "loss": 0.2694,
+      "step": 64050
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.017852686426238e-05,
+      "loss": 0.2807,
+      "step": 64055
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0175739895015636e-05,
+      "loss": 0.2918,
+      "step": 64060
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0172952858565317e-05,
+      "loss": 0.3113,
+      "step": 64065
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.017016575494761e-05,
+      "loss": 0.2758,
+      "step": 64070
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0167378584198714e-05,
+      "loss": 0.2863,
+      "step": 64075
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.016459134635481e-05,
+      "loss": 0.2638,
+      "step": 64080
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0161804041452092e-05,
+      "loss": 0.2724,
+      "step": 64085
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0159016669526752e-05,
+      "loss": 0.2684,
+      "step": 64090
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0156229230614973e-05,
+      "loss": 0.2513,
+      "step": 64095
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.015344172475296e-05,
+      "loss": 0.2973,
+      "step": 64100
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0150654151976905e-05,
+      "loss": 0.2869,
+      "step": 64105
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0147866512322996e-05,
+      "loss": 0.2659,
+      "step": 64110
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0145078805827437e-05,
+      "loss": 0.3028,
+      "step": 64115
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.014229103252642e-05,
+      "loss": 0.2946,
+      "step": 64120
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.013950319245614e-05,
+      "loss": 0.276,
+      "step": 64125
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0136715285652795e-05,
+      "loss": 0.2682,
+      "step": 64130
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0133927312152593e-05,
+      "loss": 0.2826,
+      "step": 64135
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0131139271991726e-05,
+      "loss": 0.2862,
+      "step": 64140
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0128351165206404e-05,
+      "loss": 0.2731,
+      "step": 64145
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.012556299183281e-05,
+      "loss": 0.2835,
+      "step": 64150
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.012277475190716e-05,
+      "loss": 0.2742,
+      "step": 64155
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0119986445465663e-05,
+      "loss": 0.268,
+      "step": 64160
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0117198072544518e-05,
+      "loss": 0.2489,
+      "step": 64165
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.011440963317992e-05,
+      "loss": 0.2575,
+      "step": 64170
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.011162112740809e-05,
+      "loss": 0.268,
+      "step": 64175
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0108832555265225e-05,
+      "loss": 0.272,
+      "step": 64180
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0106043916787534e-05,
+      "loss": 0.2692,
+      "step": 64185
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.010325521201123e-05,
+      "loss": 0.2692,
+      "step": 64190
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0100466440972523e-05,
+      "loss": 0.2626,
+      "step": 64195
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0097677603707614e-05,
+      "loss": 0.2806,
+      "step": 64200
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.009488870025273e-05,
+      "loss": 0.2461,
+      "step": 64205
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0092099730644064e-05,
+      "loss": 0.2769,
+      "step": 64210
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0089310694917845e-05,
+      "loss": 0.2555,
+      "step": 64215
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008652159311028e-05,
+      "loss": 0.279,
+      "step": 64220
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0083732425257583e-05,
+      "loss": 0.2507,
+      "step": 64225
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008094319139597e-05,
+      "loss": 0.2701,
+      "step": 64230
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.007815389156166e-05,
+      "loss": 0.2702,
+      "step": 64235
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0075364525790865e-05,
+      "loss": 0.2656,
+      "step": 64240
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0072575094119813e-05,
+      "loss": 0.281,
+      "step": 64245
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.006978559658471e-05,
+      "loss": 0.2743,
+      "step": 64250
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0066996033221785e-05,
+      "loss": 0.2509,
+      "step": 64255
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0064206404067258e-05,
+      "loss": 0.2651,
+      "step": 64260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0061416709157343e-05,
+      "loss": 0.2743,
+      "step": 64265
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.005862694852827e-05,
+      "loss": 0.262,
+      "step": 64270
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.005583712221626e-05,
+      "loss": 0.2638,
+      "step": 64275
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0053047230257535e-05,
+      "loss": 0.2714,
+      "step": 64280
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0050257272688327e-05,
+      "loss": 0.2561,
+      "step": 64285
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0047467249544854e-05,
+      "loss": 0.2601,
+      "step": 64290
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0044677160863338e-05,
+      "loss": 0.2726,
+      "step": 64295
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0041887006680026e-05,
+      "loss": 0.2747,
+      "step": 64300
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0039096787031123e-05,
+      "loss": 0.2926,
+      "step": 64305
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0036306501952882e-05,
+      "loss": 0.2767,
+      "step": 64310
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0033516151481512e-05,
+      "loss": 0.291,
+      "step": 64315
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0030725735653246e-05,
+      "loss": 0.2831,
+      "step": 64320
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.002793525450433e-05,
+      "loss": 0.248,
+      "step": 64325
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.002514470807099e-05,
+      "loss": 0.2657,
+      "step": 64330
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.002235409638945e-05,
+      "loss": 0.2686,
+      "step": 64335
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.001956341949596e-05,
+      "loss": 0.2589,
+      "step": 64340
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.001677267742674e-05,
+      "loss": 0.2523,
+      "step": 64345
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0013981870218034e-05,
+      "loss": 0.2848,
+      "step": 64350
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.001119099790608e-05,
+      "loss": 0.2721,
+      "step": 64355
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0008400060527113e-05,
+      "loss": 0.2901,
+      "step": 64360
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0005609058117374e-05,
+      "loss": 0.2814,
+      "step": 64365
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0002817990713096e-05,
+      "loss": 0.256,
+      "step": 64370
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0000026858350522e-05,
+      "loss": 0.2928,
+      "step": 64375
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.99972356610659e-05,
+      "loss": 0.2758,
+      "step": 64380
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9994444398895464e-05,
+      "loss": 0.2717,
+      "step": 64385
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.999165307187546e-05,
+      "loss": 0.2795,
+      "step": 64390
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.998886168004213e-05,
+      "loss": 0.263,
+      "step": 64395
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.998607022343172e-05,
+      "loss": 0.2792,
+      "step": 64400
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9983278702080468e-05,
+      "loss": 0.2497,
+      "step": 64405
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.998048711602463e-05,
+      "loss": 0.2719,
+      "step": 64410
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9977695465300455e-05,
+      "loss": 0.2633,
+      "step": 64415
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.997490374994418e-05,
+      "loss": 0.2752,
+      "step": 64420
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9972111969992055e-05,
+      "loss": 0.3034,
+      "step": 64425
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.996932012548033e-05,
+      "loss": 0.2559,
+      "step": 64430
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9966528216445265e-05,
+      "loss": 0.2829,
+      "step": 64435
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9963736242923107e-05,
+      "loss": 0.2869,
+      "step": 64440
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9960944204950105e-05,
+      "loss": 0.2867,
+      "step": 64445
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9958152102562503e-05,
+      "loss": 0.2968,
+      "step": 64450
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9955359935796568e-05,
+      "loss": 0.2643,
+      "step": 64455
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.995256770468855e-05,
+      "loss": 0.2917,
+      "step": 64460
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9949775409274706e-05,
+      "loss": 0.2657,
+      "step": 64465
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.994698304959129e-05,
+      "loss": 0.2644,
+      "step": 64470
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.994419062567456e-05,
+      "loss": 0.2726,
+      "step": 64475
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9941398137560772e-05,
+      "loss": 0.272,
+      "step": 64480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9938605585286182e-05,
+      "loss": 0.2778,
+      "step": 64485
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9935812968887055e-05,
+      "loss": 0.2645,
+      "step": 64490
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.993302028839965e-05,
+      "loss": 0.2834,
+      "step": 64495
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.993022754386023e-05,
+      "loss": 0.2824,
+      "step": 64500
+    },
+    {
+      "epoch": 2.27,
+      "eval_loss": 0.26770463585853577,
+      "eval_runtime": 10.5455,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 64500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.992743473530505e-05,
+      "loss": 0.2888,
+      "step": 64505
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9924641862770376e-05,
+      "loss": 0.2819,
+      "step": 64510
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.992184892629248e-05,
+      "loss": 0.2553,
+      "step": 64515
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.991905592590762e-05,
+      "loss": 0.2743,
+      "step": 64520
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.991626286165205e-05,
+      "loss": 0.2775,
+      "step": 64525
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.991346973356206e-05,
+      "loss": 0.2794,
+      "step": 64530
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9910676541673893e-05,
+      "loss": 0.2578,
+      "step": 64535
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.990788328602383e-05,
+      "loss": 0.2791,
+      "step": 64540
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9905089966648137e-05,
+      "loss": 0.2512,
+      "step": 64545
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.990229658358309e-05,
+      "loss": 0.2719,
+      "step": 64550
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.989950313686494e-05,
+      "loss": 0.2566,
+      "step": 64555
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9896709626529986e-05,
+      "loss": 0.2486,
+      "step": 64560
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9893916052614474e-05,
+      "loss": 0.2557,
+      "step": 64565
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.989112241515469e-05,
+      "loss": 0.2585,
+      "step": 64570
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9888328714186898e-05,
+      "loss": 0.294,
+      "step": 64575
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.988553494974739e-05,
+      "loss": 0.2827,
+      "step": 64580
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9882741121872425e-05,
+      "loss": 0.2649,
+      "step": 64585
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9879947230598283e-05,
+      "loss": 0.2765,
+      "step": 64590
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9877153275961244e-05,
+      "loss": 0.2723,
+      "step": 64595
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.987435925799758e-05,
+      "loss": 0.2668,
+      "step": 64600
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9871565176743583e-05,
+      "loss": 0.2791,
+      "step": 64605
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9868771032235514e-05,
+      "loss": 0.2842,
+      "step": 64610
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9865976824509673e-05,
+      "loss": 0.2497,
+      "step": 64615
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.986318255360232e-05,
+      "loss": 0.2682,
+      "step": 64620
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9860388219549744e-05,
+      "loss": 0.2814,
+      "step": 64625
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9857593822388237e-05,
+      "loss": 0.286,
+      "step": 64630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9854799362154074e-05,
+      "loss": 0.2544,
+      "step": 64635
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.985200483888354e-05,
+      "loss": 0.2455,
+      "step": 64640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9849210252612926e-05,
+      "loss": 0.2825,
+      "step": 64645
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9846415603378496e-05,
+      "loss": 0.2637,
+      "step": 64650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9843620891216566e-05,
+      "loss": 0.2724,
+      "step": 64655
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.984082611616341e-05,
+      "loss": 0.2574,
+      "step": 64660
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.983803127825532e-05,
+      "loss": 0.2862,
+      "step": 64665
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9835236377528575e-05,
+      "loss": 0.2622,
+      "step": 64670
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9832441414019474e-05,
+      "loss": 0.2929,
+      "step": 64675
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.982964638776431e-05,
+      "loss": 0.2792,
+      "step": 64680
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.982685129879937e-05,
+      "loss": 0.2724,
+      "step": 64685
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9824056147160944e-05,
+      "loss": 0.2722,
+      "step": 64690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.982126093288533e-05,
+      "loss": 0.2795,
+      "step": 64695
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.981846565600882e-05,
+      "loss": 0.2508,
+      "step": 64700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9815670316567706e-05,
+      "loss": 0.2602,
+      "step": 64705
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9812874914598288e-05,
+      "loss": 0.2481,
+      "step": 64710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9810079450136863e-05,
+      "loss": 0.2904,
+      "step": 64715
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9807283923219724e-05,
+      "loss": 0.2735,
+      "step": 64720
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9804488333883172e-05,
+      "loss": 0.2911,
+      "step": 64725
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9801692682163508e-05,
+      "loss": 0.2698,
+      "step": 64730
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9798896968097025e-05,
+      "loss": 0.2831,
+      "step": 64735
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9796101191720033e-05,
+      "loss": 0.2828,
+      "step": 64740
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9793305353068816e-05,
+      "loss": 0.2739,
+      "step": 64745
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9790509452179706e-05,
+      "loss": 0.2611,
+      "step": 64750
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9787713489088975e-05,
+      "loss": 0.2712,
+      "step": 64755
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.978491746383294e-05,
+      "loss": 0.2625,
+      "step": 64760
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.97821213764479e-05,
+      "loss": 0.3069,
+      "step": 64765
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9779325226970183e-05,
+      "loss": 0.2856,
+      "step": 64770
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9776529015436065e-05,
+      "loss": 0.2727,
+      "step": 64775
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.977373274188187e-05,
+      "loss": 0.2896,
+      "step": 64780
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9770936406343895e-05,
+      "loss": 0.2778,
+      "step": 64785
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9768140008858464e-05,
+      "loss": 0.2569,
+      "step": 64790
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9765343549461866e-05,
+      "loss": 0.2996,
+      "step": 64795
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9762547028190436e-05,
+      "loss": 0.2896,
+      "step": 64800
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9759750445080465e-05,
+      "loss": 0.2651,
+      "step": 64805
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9756953800168262e-05,
+      "loss": 0.2642,
+      "step": 64810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.975415709349016e-05,
+      "loss": 0.2653,
+      "step": 64815
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.975136032508246e-05,
+      "loss": 0.2717,
+      "step": 64820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.974856349498148e-05,
+      "loss": 0.2564,
+      "step": 64825
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.974576660322353e-05,
+      "loss": 0.2669,
+      "step": 64830
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9742969649844926e-05,
+      "loss": 0.2428,
+      "step": 64835
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9740172634881986e-05,
+      "loss": 0.291,
+      "step": 64840
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9737375558371026e-05,
+      "loss": 0.2643,
+      "step": 64845
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.973457842034837e-05,
+      "loss": 0.2804,
+      "step": 64850
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.973178122085034e-05,
+      "loss": 0.2886,
+      "step": 64855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9728983959913243e-05,
+      "loss": 0.2789,
+      "step": 64860
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9726186637573404e-05,
+      "loss": 0.2683,
+      "step": 64865
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9723389253867146e-05,
+      "loss": 0.2639,
+      "step": 64870
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9720591808830796e-05,
+      "loss": 0.2614,
+      "step": 64875
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.971779430250067e-05,
+      "loss": 0.2755,
+      "step": 64880
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.97149967349131e-05,
+      "loss": 0.3162,
+      "step": 64885
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9712199106104392e-05,
+      "loss": 0.2874,
+      "step": 64890
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9709401416110895e-05,
+      "loss": 0.2674,
+      "step": 64895
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9706603664968913e-05,
+      "loss": 0.26,
+      "step": 64900
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9703805852714796e-05,
+      "loss": 0.2534,
+      "step": 64905
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9701007979384855e-05,
+      "loss": 0.294,
+      "step": 64910
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9698210045015424e-05,
+      "loss": 0.2643,
+      "step": 64915
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9695971653595856e-05,
+      "loss": 0.2674,
+      "step": 64920
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9693173609446895e-05,
+      "loss": 0.2858,
+      "step": 64925
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9690375504360162e-05,
+      "loss": 0.2574,
+      "step": 64930
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9687577338371997e-05,
+      "loss": 0.2608,
+      "step": 64935
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9684779111518728e-05,
+      "loss": 0.2728,
+      "step": 64940
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9681980823836696e-05,
+      "loss": 0.2791,
+      "step": 64945
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.967918247536222e-05,
+      "loss": 0.2715,
+      "step": 64950
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9676384066131646e-05,
+      "loss": 0.2619,
+      "step": 64955
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9673585596181303e-05,
+      "loss": 0.2621,
+      "step": 64960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9670787065547538e-05,
+      "loss": 0.2665,
+      "step": 64965
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9667988474266678e-05,
+      "loss": 0.2658,
+      "step": 64970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.966518982237507e-05,
+      "loss": 0.2694,
+      "step": 64975
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9662391109909038e-05,
+      "loss": 0.2698,
+      "step": 64980
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9659592336904933e-05,
+      "loss": 0.279,
+      "step": 64985
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9656793503399095e-05,
+      "loss": 0.2622,
+      "step": 64990
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.965399460942786e-05,
+      "loss": 0.2711,
+      "step": 64995
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.965119565502757e-05,
+      "loss": 0.2372,
+      "step": 65000
+    },
+    {
+      "epoch": 2.29,
+      "eval_loss": 0.2671353220939636,
+      "eval_runtime": 10.5464,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 65000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9648396640234578e-05,
+      "loss": 0.2522,
+      "step": 65005
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9645597565085208e-05,
+      "loss": 0.2762,
+      "step": 65010
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.964279842961582e-05,
+      "loss": 0.2681,
+      "step": 65015
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9639999233862752e-05,
+      "loss": 0.2478,
+      "step": 65020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.963719997786235e-05,
+      "loss": 0.2876,
+      "step": 65025
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9634400661650973e-05,
+      "loss": 0.2809,
+      "step": 65030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9631601285264947e-05,
+      "loss": 0.2557,
+      "step": 65035
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9628801848740633e-05,
+      "loss": 0.2618,
+      "step": 65040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.962600235211438e-05,
+      "loss": 0.2638,
+      "step": 65045
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9623202795422532e-05,
+      "loss": 0.2836,
+      "step": 65050
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9620403178701445e-05,
+      "loss": 0.261,
+      "step": 65055
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9617603501987473e-05,
+      "loss": 0.2787,
+      "step": 65060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9614803765316956e-05,
+      "loss": 0.2619,
+      "step": 65065
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9612003968726254e-05,
+      "loss": 0.2937,
+      "step": 65070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9609204112251715e-05,
+      "loss": 0.2591,
+      "step": 65075
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.960640419592971e-05,
+      "loss": 0.2509,
+      "step": 65080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9603604219796573e-05,
+      "loss": 0.2657,
+      "step": 65085
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9600804183888675e-05,
+      "loss": 0.2633,
+      "step": 65090
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9598004088242355e-05,
+      "loss": 0.2567,
+      "step": 65095
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9595203932893996e-05,
+      "loss": 0.2692,
+      "step": 65100
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9592403717879935e-05,
+      "loss": 0.2629,
+      "step": 65105
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.958960344323654e-05,
+      "loss": 0.2529,
+      "step": 65110
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.958680310900017e-05,
+      "loss": 0.278,
+      "step": 65115
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9584002715207182e-05,
+      "loss": 0.2807,
+      "step": 65120
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9581202261893937e-05,
+      "loss": 0.2419,
+      "step": 65125
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9578401749096806e-05,
+      "loss": 0.2967,
+      "step": 65130
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9575601176852146e-05,
+      "loss": 0.2738,
+      "step": 65135
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9572800545196317e-05,
+      "loss": 0.2837,
+      "step": 65140
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9569999854165687e-05,
+      "loss": 0.2811,
+      "step": 65145
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.956719910379662e-05,
+      "loss": 0.2877,
+      "step": 65150
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9564398294125485e-05,
+      "loss": 0.2804,
+      "step": 65155
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9561597425188642e-05,
+      "loss": 0.2756,
+      "step": 65160
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.955879649702247e-05,
+      "loss": 0.2892,
+      "step": 65165
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.955599550966332e-05,
+      "loss": 0.2629,
+      "step": 65170
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.955319446314757e-05,
+      "loss": 0.2763,
+      "step": 65175
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9550393357511596e-05,
+      "loss": 0.2644,
+      "step": 65180
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9547592192791766e-05,
+      "loss": 0.2685,
+      "step": 65185
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.954479096902445e-05,
+      "loss": 0.2634,
+      "step": 65190
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.954198968624602e-05,
+      "loss": 0.304,
+      "step": 65195
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9539188344492837e-05,
+      "loss": 0.2807,
+      "step": 65200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9536386943801287e-05,
+      "loss": 0.3002,
+      "step": 65205
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9533585484207743e-05,
+      "loss": 0.253,
+      "step": 65210
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.953078396574858e-05,
+      "loss": 0.295,
+      "step": 65215
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9527982388460175e-05,
+      "loss": 0.2554,
+      "step": 65220
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.9525180752378904e-05,
+      "loss": 0.274,
+      "step": 65225
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.952237905754114e-05,
+      "loss": 0.275,
+      "step": 65230
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9519577303983264e-05,
+      "loss": 0.2911,
+      "step": 65235
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9516775491741654e-05,
+      "loss": 0.2803,
+      "step": 65240
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.95139736208527e-05,
+      "loss": 0.2559,
+      "step": 65245
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9511171691352764e-05,
+      "loss": 0.2758,
+      "step": 65250
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.950836970327823e-05,
+      "loss": 0.276,
+      "step": 65255
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9505567656665495e-05,
+      "loss": 0.2704,
+      "step": 65260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.950276555155093e-05,
+      "loss": 0.2763,
+      "step": 65265
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.949996338797093e-05,
+      "loss": 0.2608,
+      "step": 65270
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.949716116596187e-05,
+      "loss": 0.2825,
+      "step": 65275
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9494358885560138e-05,
+      "loss": 0.2638,
+      "step": 65280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9491556546802106e-05,
+      "loss": 0.2725,
+      "step": 65285
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9488754149724175e-05,
+      "loss": 0.2593,
+      "step": 65290
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9485951694362734e-05,
+      "loss": 0.2668,
+      "step": 65295
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9483149180754168e-05,
+      "loss": 0.2634,
+      "step": 65300
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.948034660893486e-05,
+      "loss": 0.2597,
+      "step": 65305
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9477543978941213e-05,
+      "loss": 0.2492,
+      "step": 65310
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9474741290809592e-05,
+      "loss": 0.2786,
+      "step": 65315
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9471938544576417e-05,
+      "loss": 0.2849,
+      "step": 65320
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9469135740278066e-05,
+      "loss": 0.279,
+      "step": 65325
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.946633287795093e-05,
+      "loss": 0.268,
+      "step": 65330
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9463529957631404e-05,
+      "loss": 0.2755,
+      "step": 65335
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.946072697935588e-05,
+      "loss": 0.2858,
+      "step": 65340
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.945792394316076e-05,
+      "loss": 0.2783,
+      "step": 65345
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9455120849082442e-05,
+      "loss": 0.2666,
+      "step": 65350
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9452317697157307e-05,
+      "loss": 0.2716,
+      "step": 65355
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9449514487421764e-05,
+      "loss": 0.2684,
+      "step": 65360
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9446711219912203e-05,
+      "loss": 0.2683,
+      "step": 65365
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9443907894665036e-05,
+      "loss": 0.2721,
+      "step": 65370
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9441104511716645e-05,
+      "loss": 0.2548,
+      "step": 65375
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.943830107110344e-05,
+      "loss": 0.2763,
+      "step": 65380
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9435497572861824e-05,
+      "loss": 0.2818,
+      "step": 65385
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9432694017028195e-05,
+      "loss": 0.2857,
+      "step": 65390
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9429890403638955e-05,
+      "loss": 0.2841,
+      "step": 65395
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9427086732730506e-05,
+      "loss": 0.279,
+      "step": 65400
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.942428300433926e-05,
+      "loss": 0.2693,
+      "step": 65405
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9421479218501607e-05,
+      "loss": 0.2909,
+      "step": 65410
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9418675375253968e-05,
+      "loss": 0.2845,
+      "step": 65415
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9415871474632735e-05,
+      "loss": 0.2763,
+      "step": 65420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.941306751667432e-05,
+      "loss": 0.2548,
+      "step": 65425
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.941026350141513e-05,
+      "loss": 0.2905,
+      "step": 65430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9407459428891583e-05,
+      "loss": 0.2622,
+      "step": 65435
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.940465529914007e-05,
+      "loss": 0.2889,
+      "step": 65440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9401851112197022e-05,
+      "loss": 0.2797,
+      "step": 65445
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9399046868098833e-05,
+      "loss": 0.2415,
+      "step": 65450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9396242566881916e-05,
+      "loss": 0.2487,
+      "step": 65455
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.939343820858269e-05,
+      "loss": 0.2523,
+      "step": 65460
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9390633793237566e-05,
+      "loss": 0.2733,
+      "step": 65465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9387829320882954e-05,
+      "loss": 0.2728,
+      "step": 65470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9385024791555267e-05,
+      "loss": 0.2709,
+      "step": 65475
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9382220205290923e-05,
+      "loss": 0.2774,
+      "step": 65480
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9379415562126344e-05,
+      "loss": 0.2885,
+      "step": 65485
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9376610862097936e-05,
+      "loss": 0.2733,
+      "step": 65490
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.937380610524212e-05,
+      "loss": 0.2483,
+      "step": 65495
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9371001291595314e-05,
+      "loss": 0.2939,
+      "step": 65500
+    },
+    {
+      "epoch": 2.3,
+      "eval_loss": 0.26703858375549316,
+      "eval_runtime": 10.541,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 65500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.936819642119394e-05,
+      "loss": 0.287,
+      "step": 65505
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9365391494074407e-05,
+      "loss": 0.2491,
+      "step": 65510
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.9362586510273143e-05,
+      "loss": 0.2859,
+      "step": 65515
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.935978146982658e-05,
+      "loss": 0.263,
+      "step": 65520
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9356976372771117e-05,
+      "loss": 0.2789,
+      "step": 65525
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9354171219143193e-05,
+      "loss": 0.2859,
+      "step": 65530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9351366008979215e-05,
+      "loss": 0.2607,
+      "step": 65535
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.934856074231563e-05,
+      "loss": 0.2644,
+      "step": 65540
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9345755419188843e-05,
+      "loss": 0.289,
+      "step": 65545
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9342950039635287e-05,
+      "loss": 0.287,
+      "step": 65550
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9340144603691387e-05,
+      "loss": 0.2628,
+      "step": 65555
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.933733911139357e-05,
+      "loss": 0.2871,
+      "step": 65560
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.933453356277826e-05,
+      "loss": 0.2694,
+      "step": 65565
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9331727957881892e-05,
+      "loss": 0.2994,
+      "step": 65570
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.932892229674089e-05,
+      "loss": 0.2825,
+      "step": 65575
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.932611657939169e-05,
+      "loss": 0.2706,
+      "step": 65580
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.932331080587071e-05,
+      "loss": 0.2541,
+      "step": 65585
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9320504976214385e-05,
+      "loss": 0.2803,
+      "step": 65590
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9317699090459154e-05,
+      "loss": 0.2711,
+      "step": 65595
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9314893148641455e-05,
+      "loss": 0.2808,
+      "step": 65600
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9312087150797697e-05,
+      "loss": 0.2655,
+      "step": 65605
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9309281096964336e-05,
+      "loss": 0.2863,
+      "step": 65610
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.93064749871778e-05,
+      "loss": 0.2395,
+      "step": 65615
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9303668821474522e-05,
+      "loss": 0.2728,
+      "step": 65620
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9300862599890944e-05,
+      "loss": 0.2665,
+      "step": 65625
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.92980563224635e-05,
+      "loss": 0.2773,
+      "step": 65630
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9295249989228628e-05,
+      "loss": 0.2719,
+      "step": 65635
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9292443600222764e-05,
+      "loss": 0.2502,
+      "step": 65640
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9289637155482346e-05,
+      "loss": 0.2657,
+      "step": 65645
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9286830655043817e-05,
+      "loss": 0.2493,
+      "step": 65650
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9284024098943618e-05,
+      "loss": 0.2523,
+      "step": 65655
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9281217487218193e-05,
+      "loss": 0.2934,
+      "step": 65660
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9278410819903978e-05,
+      "loss": 0.2807,
+      "step": 65665
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9275604097037418e-05,
+      "loss": 0.3029,
+      "step": 65670
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9272797318654954e-05,
+      "loss": 0.2729,
+      "step": 65675
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.926999048479303e-05,
+      "loss": 0.284,
+      "step": 65680
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9267183595488106e-05,
+      "loss": 0.2547,
+      "step": 65685
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9264376650776604e-05,
+      "loss": 0.2649,
+      "step": 65690
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9261569650694986e-05,
+      "loss": 0.2765,
+      "step": 65695
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9258762595279693e-05,
+      "loss": 0.2556,
+      "step": 65700
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9255955484567176e-05,
+      "loss": 0.3017,
+      "step": 65705
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.925314831859388e-05,
+      "loss": 0.2805,
+      "step": 65710
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9250341097396262e-05,
+      "loss": 0.2693,
+      "step": 65715
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9247533821010758e-05,
+      "loss": 0.2638,
+      "step": 65720
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9244726489473827e-05,
+      "loss": 0.2623,
+      "step": 65725
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.924191910282192e-05,
+      "loss": 0.2724,
+      "step": 65730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9239111661091488e-05,
+      "loss": 0.2673,
+      "step": 65735
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9236304164318995e-05,
+      "loss": 0.2617,
+      "step": 65740
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9233496612540874e-05,
+      "loss": 0.2669,
+      "step": 65745
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9230689005793598e-05,
+      "loss": 0.2578,
+      "step": 65750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.92278813441136e-05,
+      "loss": 0.2796,
+      "step": 65755
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9225073627537358e-05,
+      "loss": 0.2737,
+      "step": 65760
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9222265856101317e-05,
+      "loss": 0.2745,
+      "step": 65765
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9219458029841945e-05,
+      "loss": 0.2825,
+      "step": 65770
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9216650148795683e-05,
+      "loss": 0.2576,
+      "step": 65775
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9213842212998994e-05,
+      "loss": 0.2439,
+      "step": 65780
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9211034222488343e-05,
+      "loss": 0.2586,
+      "step": 65785
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9208226177300195e-05,
+      "loss": 0.2714,
+      "step": 65790
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.9205418077470998e-05,
+      "loss": 0.2634,
+      "step": 65795
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9202609923037222e-05,
+      "loss": 0.2927,
+      "step": 65800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9199801714035324e-05,
+      "loss": 0.27,
+      "step": 65805
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9196993450501764e-05,
+      "loss": 0.2663,
+      "step": 65810
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9194185132473013e-05,
+      "loss": 0.2596,
+      "step": 65815
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9191376759985534e-05,
+      "loss": 0.268,
+      "step": 65820
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9188568333075787e-05,
+      "loss": 0.2997,
+      "step": 65825
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9185759851780247e-05,
+      "loss": 0.2841,
+      "step": 65830
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9182951316135367e-05,
+      "loss": 0.2822,
+      "step": 65835
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9180142726177616e-05,
+      "loss": 0.2858,
+      "step": 65840
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.917733408194348e-05,
+      "loss": 0.2727,
+      "step": 65845
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9174525383469408e-05,
+      "loss": 0.2686,
+      "step": 65850
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9171716630791878e-05,
+      "loss": 0.2368,
+      "step": 65855
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.916890782394735e-05,
+      "loss": 0.2818,
+      "step": 65860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9166098962972306e-05,
+      "loss": 0.256,
+      "step": 65865
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9163290047903208e-05,
+      "loss": 0.2594,
+      "step": 65870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9160481078776535e-05,
+      "loss": 0.259,
+      "step": 65875
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.915767205562876e-05,
+      "loss": 0.278,
+      "step": 65880
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9154862978496354e-05,
+      "loss": 0.2681,
+      "step": 65885
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9152053847415782e-05,
+      "loss": 0.2755,
+      "step": 65890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9149244662423535e-05,
+      "loss": 0.2759,
+      "step": 65895
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.914643542355608e-05,
+      "loss": 0.2689,
+      "step": 65900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9143626130849895e-05,
+      "loss": 0.2655,
+      "step": 65905
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9140816784341445e-05,
+      "loss": 0.2749,
+      "step": 65910
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.913800738406723e-05,
+      "loss": 0.277,
+      "step": 65915
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9135197930063708e-05,
+      "loss": 0.2265,
+      "step": 65920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9132388422367375e-05,
+      "loss": 0.2685,
+      "step": 65925
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.912957886101469e-05,
+      "loss": 0.2863,
+      "step": 65930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.912676924604216e-05,
+      "loss": 0.2608,
+      "step": 65935
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.912395957748624e-05,
+      "loss": 0.2767,
+      "step": 65940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9121149855383417e-05,
+      "loss": 0.2627,
+      "step": 65945
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.911834007977019e-05,
+      "loss": 0.2609,
+      "step": 65950
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.911553025068302e-05,
+      "loss": 0.3032,
+      "step": 65955
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.911272036815842e-05,
+      "loss": 0.303,
+      "step": 65960
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.910991043223284e-05,
+      "loss": 0.2768,
+      "step": 65965
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9107100442942793e-05,
+      "loss": 0.247,
+      "step": 65970
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.910429040032474e-05,
+      "loss": 0.2768,
+      "step": 65975
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9101480304415196e-05,
+      "loss": 0.2787,
+      "step": 65980
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9098670155250624e-05,
+      "loss": 0.2531,
+      "step": 65985
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9095859952867527e-05,
+      "loss": 0.253,
+      "step": 65990
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9093049697302382e-05,
+      "loss": 0.2818,
+      "step": 65995
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9090239388591684e-05,
+      "loss": 0.2416,
+      "step": 66000
+    },
+    {
+      "epoch": 2.32,
+      "eval_loss": 0.2667825520038605,
+      "eval_runtime": 10.5519,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 66000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9087429026771924e-05,
+      "loss": 0.2977,
+      "step": 66005
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9084618611879598e-05,
+      "loss": 0.2883,
+      "step": 66010
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9081808143951185e-05,
+      "loss": 0.2677,
+      "step": 66015
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.907899762302319e-05,
+      "loss": 0.2862,
+      "step": 66020
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.907618704913209e-05,
+      "loss": 0.2749,
+      "step": 66025
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.907337642231439e-05,
+      "loss": 0.2824,
+      "step": 66030
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9070565742606582e-05,
+      "loss": 0.2703,
+      "step": 66035
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9067755010045167e-05,
+      "loss": 0.2743,
+      "step": 66040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9064944224666623e-05,
+      "loss": 0.2623,
+      "step": 66045
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9062133386507467e-05,
+      "loss": 0.246,
+      "step": 66050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9059322495604173e-05,
+      "loss": 0.2824,
+      "step": 66055
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9056511551993264e-05,
+      "loss": 0.2938,
+      "step": 66060
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.905370055571122e-05,
+      "loss": 0.2722,
+      "step": 66065
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.9050889506794547e-05,
+      "loss": 0.26,
+      "step": 66070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.904807840527975e-05,
+      "loss": 0.2683,
+      "step": 66075
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.904526725120332e-05,
+      "loss": 0.2762,
+      "step": 66080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.904245604460175e-05,
+      "loss": 0.2937,
+      "step": 66085
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.903964478551156e-05,
+      "loss": 0.2587,
+      "step": 66090
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.903683347396925e-05,
+      "loss": 0.2781,
+      "step": 66095
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9034022110011305e-05,
+      "loss": 0.2881,
+      "step": 66100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9031210693674256e-05,
+      "loss": 0.2518,
+      "step": 66105
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9028399224994586e-05,
+      "loss": 0.2767,
+      "step": 66110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.90255877040088e-05,
+      "loss": 0.2892,
+      "step": 66115
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9022776130753414e-05,
+      "loss": 0.2792,
+      "step": 66120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.901996450526494e-05,
+      "loss": 0.2875,
+      "step": 66125
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.901715282757987e-05,
+      "loss": 0.2502,
+      "step": 66130
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9014341097734717e-05,
+      "loss": 0.2566,
+      "step": 66135
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.901152931576599e-05,
+      "loss": 0.2717,
+      "step": 66140
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.90087174817102e-05,
+      "loss": 0.2481,
+      "step": 66145
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.9005905595603854e-05,
+      "loss": 0.2758,
+      "step": 66150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.900309365748347e-05,
+      "loss": 0.2538,
+      "step": 66155
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.900028166738555e-05,
+      "loss": 0.2659,
+      "step": 66160
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.89974696253466e-05,
+      "loss": 0.2753,
+      "step": 66165
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8994657531403153e-05,
+      "loss": 0.2585,
+      "step": 66170
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.899184538559171e-05,
+      "loss": 0.2701,
+      "step": 66175
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8989033187948778e-05,
+      "loss": 0.2448,
+      "step": 66180
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8986220938510884e-05,
+      "loss": 0.2692,
+      "step": 66185
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8983408637314535e-05,
+      "loss": 0.2859,
+      "step": 66190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8980596284396246e-05,
+      "loss": 0.2792,
+      "step": 66195
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.897778387979255e-05,
+      "loss": 0.2653,
+      "step": 66200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8974971423539937e-05,
+      "loss": 0.2734,
+      "step": 66205
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.897215891567495e-05,
+      "loss": 0.2732,
+      "step": 66210
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8969346356234095e-05,
+      "loss": 0.2985,
+      "step": 66215
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8966533745253893e-05,
+      "loss": 0.2597,
+      "step": 66220
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8963721082770863e-05,
+      "loss": 0.2916,
+      "step": 66225
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.896090836882153e-05,
+      "loss": 0.2687,
+      "step": 66230
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.895809560344241e-05,
+      "loss": 0.2851,
+      "step": 66235
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.895528278667003e-05,
+      "loss": 0.249,
+      "step": 66240
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8952469918540908e-05,
+      "loss": 0.2783,
+      "step": 66245
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8949656999091568e-05,
+      "loss": 0.2905,
+      "step": 66250
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8946844028358533e-05,
+      "loss": 0.255,
+      "step": 66255
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8944031006378332e-05,
+      "loss": 0.253,
+      "step": 66260
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8941217933187487e-05,
+      "loss": 0.2725,
+      "step": 66265
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8938404808822527e-05,
+      "loss": 0.2718,
+      "step": 66270
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.893559163331997e-05,
+      "loss": 0.2692,
+      "step": 66275
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8932778406716353e-05,
+      "loss": 0.2846,
+      "step": 66280
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8929965129048197e-05,
+      "loss": 0.2979,
+      "step": 66285
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8927151800352042e-05,
+      "loss": 0.2936,
+      "step": 66290
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8924338420664403e-05,
+      "loss": 0.2795,
+      "step": 66295
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8921524990021816e-05,
+      "loss": 0.2871,
+      "step": 66300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8918711508460805e-05,
+      "loss": 0.2471,
+      "step": 66305
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8915897976017915e-05,
+      "loss": 0.2777,
+      "step": 66310
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.891308439272967e-05,
+      "loss": 0.2781,
+      "step": 66315
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8910270758632606e-05,
+      "loss": 0.2709,
+      "step": 66320
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8907457073763254e-05,
+      "loss": 0.2599,
+      "step": 66325
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.890464333815814e-05,
+      "loss": 0.291,
+      "step": 66330
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.890182955185381e-05,
+      "loss": 0.2766,
+      "step": 66335
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.889901571488679e-05,
+      "loss": 0.2551,
+      "step": 66340
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.889620182729363e-05,
+      "loss": 0.2641,
+      "step": 66345
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8893387889110845e-05,
+      "loss": 0.2837,
+      "step": 66350
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.889057390037499e-05,
+      "loss": 0.2869,
+      "step": 66355
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8887759861122594e-05,
+      "loss": 0.2369,
+      "step": 66360
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.8884945771390203e-05,
+      "loss": 0.2778,
+      "step": 66365
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.888213163121435e-05,
+      "loss": 0.277,
+      "step": 66370
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.887931744063158e-05,
+      "loss": 0.2471,
+      "step": 66375
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8876503199678424e-05,
+      "loss": 0.2734,
+      "step": 66380
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8873688908391426e-05,
+      "loss": 0.2756,
+      "step": 66385
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8870874566807134e-05,
+      "loss": 0.2497,
+      "step": 66390
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8868060174962092e-05,
+      "loss": 0.2631,
+      "step": 66395
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.886524573289283e-05,
+      "loss": 0.2787,
+      "step": 66400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8862431240635906e-05,
+      "loss": 0.2505,
+      "step": 66405
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8859616698227854e-05,
+      "loss": 0.2918,
+      "step": 66410
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8856802105705216e-05,
+      "loss": 0.2516,
+      "step": 66415
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8853987463104554e-05,
+      "loss": 0.2788,
+      "step": 66420
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8851172770462398e-05,
+      "loss": 0.2767,
+      "step": 66425
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.884835802781531e-05,
+      "loss": 0.2687,
+      "step": 66430
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8845543235199824e-05,
+      "loss": 0.2584,
+      "step": 66435
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.884272839265249e-05,
+      "loss": 0.2691,
+      "step": 66440
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8839913500209863e-05,
+      "loss": 0.269,
+      "step": 66445
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8837098557908493e-05,
+      "loss": 0.2869,
+      "step": 66450
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8834283565784927e-05,
+      "loss": 0.266,
+      "step": 66455
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8831468523875715e-05,
+      "loss": 0.2755,
+      "step": 66460
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8828653432217405e-05,
+      "loss": 0.2646,
+      "step": 66465
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.882583829084656e-05,
+      "loss": 0.2931,
+      "step": 66470
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8823023099799717e-05,
+      "loss": 0.2787,
+      "step": 66475
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.882020785911345e-05,
+      "loss": 0.26,
+      "step": 66480
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8817392568824292e-05,
+      "loss": 0.2788,
+      "step": 66485
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8814577228968815e-05,
+      "loss": 0.2496,
+      "step": 66490
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8811761839583558e-05,
+      "loss": 0.2569,
+      "step": 66495
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8808946400705094e-05,
+      "loss": 0.3004,
+      "step": 66500
+    },
+    {
+      "epoch": 2.34,
+      "eval_loss": 0.2670895457267761,
+      "eval_runtime": 10.5428,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 66500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8806130912369966e-05,
+      "loss": 0.2656,
+      "step": 66505
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8803315374614738e-05,
+      "loss": 0.2754,
+      "step": 66510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8800499787475964e-05,
+      "loss": 0.273,
+      "step": 66515
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8797684150990206e-05,
+      "loss": 0.2715,
+      "step": 66520
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8794868465194025e-05,
+      "loss": 0.2471,
+      "step": 66525
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8792052730123976e-05,
+      "loss": 0.2625,
+      "step": 66530
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8789236945816624e-05,
+      "loss": 0.283,
+      "step": 66535
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.878642111230852e-05,
+      "loss": 0.273,
+      "step": 66540
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.878360522963624e-05,
+      "loss": 0.2841,
+      "step": 66545
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8780789297836334e-05,
+      "loss": 0.2855,
+      "step": 66550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8777973316945374e-05,
+      "loss": 0.2768,
+      "step": 66555
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.877515728699992e-05,
+      "loss": 0.2538,
+      "step": 66560
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.877234120803654e-05,
+      "loss": 0.2692,
+      "step": 66565
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.876952508009179e-05,
+      "loss": 0.2605,
+      "step": 66570
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8766708903202245e-05,
+      "loss": 0.2634,
+      "step": 66575
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8763892677404465e-05,
+      "loss": 0.2392,
+      "step": 66580
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.876107640273502e-05,
+      "loss": 0.2734,
+      "step": 66585
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8758260079230475e-05,
+      "loss": 0.2385,
+      "step": 66590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8755443706927403e-05,
+      "loss": 0.2735,
+      "step": 66595
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8752627285862365e-05,
+      "loss": 0.2632,
+      "step": 66600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8749810816071936e-05,
+      "loss": 0.2805,
+      "step": 66605
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8746994297592688e-05,
+      "loss": 0.289,
+      "step": 66610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8744177730461185e-05,
+      "loss": 0.2455,
+      "step": 66615
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8741361114714006e-05,
+      "loss": 0.263,
+      "step": 66620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8738544450387716e-05,
+      "loss": 0.2629,
+      "step": 66625
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8735727737518893e-05,
+      "loss": 0.2732,
+      "step": 66630
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.87329109761441e-05,
+      "loss": 0.2799,
+      "step": 66635
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.873009416629992e-05,
+      "loss": 0.2525,
+      "step": 66640
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.872727730802293e-05,
+      "loss": 0.2636,
+      "step": 66645
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.8724460401349696e-05,
+      "loss": 0.268,
+      "step": 66650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.87216434463168e-05,
+      "loss": 0.2448,
+      "step": 66655
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.871882644296081e-05,
+      "loss": 0.2645,
+      "step": 66660
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8716009391318317e-05,
+      "loss": 0.2918,
+      "step": 66665
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8713192291425893e-05,
+      "loss": 0.2543,
+      "step": 66670
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8710375143320113e-05,
+      "loss": 0.2864,
+      "step": 66675
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8707557947037555e-05,
+      "loss": 0.2812,
+      "step": 66680
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.87047407026148e-05,
+      "loss": 0.2605,
+      "step": 66685
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8701923410088426e-05,
+      "loss": 0.2609,
+      "step": 66690
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.869910606949502e-05,
+      "loss": 0.2737,
+      "step": 66695
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8696288680871164e-05,
+      "loss": 0.2603,
+      "step": 66700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.869347124425343e-05,
+      "loss": 0.2734,
+      "step": 66705
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.869065375967841e-05,
+      "loss": 0.2654,
+      "step": 66710
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8687836227182668e-05,
+      "loss": 0.2671,
+      "step": 66715
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.868501864680282e-05,
+      "loss": 0.2596,
+      "step": 66720
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8682201018575426e-05,
+      "loss": 0.2768,
+      "step": 66725
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8679383342537087e-05,
+      "loss": 0.2812,
+      "step": 66730
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.867656561872437e-05,
+      "loss": 0.2778,
+      "step": 66735
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.867374784717387e-05,
+      "loss": 0.2537,
+      "step": 66740
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8670930027922183e-05,
+      "loss": 0.3001,
+      "step": 66745
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.866811216100589e-05,
+      "loss": 0.2672,
+      "step": 66750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.866529424646157e-05,
+      "loss": 0.2582,
+      "step": 66755
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8662476284325823e-05,
+      "loss": 0.2792,
+      "step": 66760
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.865965827463524e-05,
+      "loss": 0.2657,
+      "step": 66765
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8656840217426394e-05,
+      "loss": 0.2705,
+      "step": 66770
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8654022112735895e-05,
+      "loss": 0.2326,
+      "step": 66775
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8651203960600325e-05,
+      "loss": 0.2847,
+      "step": 66780
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8648385761056285e-05,
+      "loss": 0.2878,
+      "step": 66785
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8645567514140355e-05,
+      "loss": 0.2512,
+      "step": 66790
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.864274921988913e-05,
+      "loss": 0.2916,
+      "step": 66795
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.863993087833921e-05,
+      "loss": 0.2613,
+      "step": 66800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8637112489527195e-05,
+      "loss": 0.2886,
+      "step": 66805
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8634294053489658e-05,
+      "loss": 0.2658,
+      "step": 66810
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8631475570263218e-05,
+      "loss": 0.2933,
+      "step": 66815
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8628657039884455e-05,
+      "loss": 0.2801,
+      "step": 66820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.862583846238997e-05,
+      "loss": 0.2542,
+      "step": 66825
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8623019837816367e-05,
+      "loss": 0.2774,
+      "step": 66830
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8620201166200238e-05,
+      "loss": 0.2683,
+      "step": 66835
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8617382447578178e-05,
+      "loss": 0.2716,
+      "step": 66840
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.86145636819868e-05,
+      "loss": 0.2712,
+      "step": 66845
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8611744869462688e-05,
+      "loss": 0.2498,
+      "step": 66850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8608926010042447e-05,
+      "loss": 0.2782,
+      "step": 66855
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8606107103762688e-05,
+      "loss": 0.2876,
+      "step": 66860
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.860328815066e-05,
+      "loss": 0.2708,
+      "step": 66865
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.860046915077099e-05,
+      "loss": 0.2864,
+      "step": 66870
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8597650104132257e-05,
+      "loss": 0.2729,
+      "step": 66875
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.859483101078041e-05,
+      "loss": 0.2683,
+      "step": 66880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8592011870752062e-05,
+      "loss": 0.2692,
+      "step": 66885
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8589192684083794e-05,
+      "loss": 0.2591,
+      "step": 66890
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8586373450812233e-05,
+      "loss": 0.2617,
+      "step": 66895
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.858355417097397e-05,
+      "loss": 0.2512,
+      "step": 66900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8580734844605624e-05,
+      "loss": 0.2595,
+      "step": 66905
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.857791547174379e-05,
+      "loss": 0.2547,
+      "step": 66910
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.857509605242508e-05,
+      "loss": 0.2712,
+      "step": 66915
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8572276586686107e-05,
+      "loss": 0.2826,
+      "step": 66920
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8569457074563477e-05,
+      "loss": 0.278,
+      "step": 66925
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8566637516093807e-05,
+      "loss": 0.266,
+      "step": 66930
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.8563817911313684e-05,
+      "loss": 0.2973,
+      "step": 66935
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8560998260259748e-05,
+      "loss": 0.2799,
+      "step": 66940
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8558178562968583e-05,
+      "loss": 0.281,
+      "step": 66945
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.855535881947683e-05,
+      "loss": 0.2711,
+      "step": 66950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8552539029821074e-05,
+      "loss": 0.274,
+      "step": 66955
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8549719194037942e-05,
+      "loss": 0.2637,
+      "step": 66960
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8546899312164045e-05,
+      "loss": 0.2716,
+      "step": 66965
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8544079384236004e-05,
+      "loss": 0.2458,
+      "step": 66970
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8541259410290422e-05,
+      "loss": 0.2872,
+      "step": 66975
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8538439390363924e-05,
+      "loss": 0.2763,
+      "step": 66980
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8535619324493125e-05,
+      "loss": 0.2574,
+      "step": 66985
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8532799212714628e-05,
+      "loss": 0.2851,
+      "step": 66990
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8529979055065072e-05,
+      "loss": 0.2583,
+      "step": 66995
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.852715885158106e-05,
+      "loss": 0.2546,
+      "step": 67000
+    },
+    {
+      "epoch": 2.36,
+      "eval_loss": 0.26735541224479675,
+      "eval_runtime": 10.5619,
+      "eval_samples_per_second": 9.468,
+      "eval_steps_per_second": 9.468,
+      "step": 67000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.852433860229921e-05,
+      "loss": 0.2645,
+      "step": 67005
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.852151830725615e-05,
+      "loss": 0.2531,
+      "step": 67010
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8518697966488493e-05,
+      "loss": 0.2846,
+      "step": 67015
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8515877580032868e-05,
+      "loss": 0.2697,
+      "step": 67020
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8513057147925894e-05,
+      "loss": 0.2515,
+      "step": 67025
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8510236670204176e-05,
+      "loss": 0.2567,
+      "step": 67030
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8507416146904358e-05,
+      "loss": 0.2844,
+      "step": 67035
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8504595578063047e-05,
+      "loss": 0.2846,
+      "step": 67040
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.850177496371687e-05,
+      "loss": 0.2816,
+      "step": 67045
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8498954303902452e-05,
+      "loss": 0.2844,
+      "step": 67050
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8496133598656426e-05,
+      "loss": 0.2526,
+      "step": 67055
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8493312848015407e-05,
+      "loss": 0.2688,
+      "step": 67060
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8490492052016028e-05,
+      "loss": 0.2706,
+      "step": 67065
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.84876712106949e-05,
+      "loss": 0.2574,
+      "step": 67070
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8484850324088662e-05,
+      "loss": 0.2623,
+      "step": 67075
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.848202939223394e-05,
+      "loss": 0.2618,
+      "step": 67080
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8479208415167363e-05,
+      "loss": 0.2743,
+      "step": 67085
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8476387392925552e-05,
+      "loss": 0.2593,
+      "step": 67090
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8473566325545148e-05,
+      "loss": 0.2704,
+      "step": 67095
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8470745213062767e-05,
+      "loss": 0.2534,
+      "step": 67100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.846792405551506e-05,
+      "loss": 0.2744,
+      "step": 67105
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.846510285293863e-05,
+      "loss": 0.2651,
+      "step": 67110
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8462281605370133e-05,
+      "loss": 0.2744,
+      "step": 67115
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8459460312846186e-05,
+      "loss": 0.2822,
+      "step": 67120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8456638975403428e-05,
+      "loss": 0.2749,
+      "step": 67125
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8453817593078492e-05,
+      "loss": 0.2597,
+      "step": 67130
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8450996165908006e-05,
+      "loss": 0.2588,
+      "step": 67135
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.844817469392862e-05,
+      "loss": 0.276,
+      "step": 67140
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8445353177176946e-05,
+      "loss": 0.2702,
+      "step": 67145
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8442531615689644e-05,
+      "loss": 0.2784,
+      "step": 67150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8439710009503322e-05,
+      "loss": 0.2722,
+      "step": 67155
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8436888358654644e-05,
+      "loss": 0.2814,
+      "step": 67160
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.843406666318023e-05,
+      "loss": 0.2579,
+      "step": 67165
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.843124492311673e-05,
+      "loss": 0.2651,
+      "step": 67170
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.842842313850077e-05,
+      "loss": 0.2666,
+      "step": 67175
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8425601309369e-05,
+      "loss": 0.2696,
+      "step": 67180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8422779435758045e-05,
+      "loss": 0.2789,
+      "step": 67185
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8419957517704567e-05,
+      "loss": 0.2796,
+      "step": 67190
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8417135555245183e-05,
+      "loss": 0.2679,
+      "step": 67195
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.841431354841656e-05,
+      "loss": 0.2773,
+      "step": 67200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8411491497255316e-05,
+      "loss": 0.2451,
+      "step": 67205
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.8408669401798098e-05,
+      "loss": 0.2846,
+      "step": 67210
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.840584726208156e-05,
+      "loss": 0.2712,
+      "step": 67215
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.840302507814234e-05,
+      "loss": 0.2558,
+      "step": 67220
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8400202850017087e-05,
+      "loss": 0.2631,
+      "step": 67225
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.839738057774244e-05,
+      "loss": 0.2752,
+      "step": 67230
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.839455826135503e-05,
+      "loss": 0.2842,
+      "step": 67235
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.839173590089153e-05,
+      "loss": 0.3026,
+      "step": 67240
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.838891349638858e-05,
+      "loss": 0.2557,
+      "step": 67245
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.838609104788281e-05,
+      "loss": 0.2689,
+      "step": 67250
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.838326855541089e-05,
+      "loss": 0.2744,
+      "step": 67255
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8380446019009448e-05,
+      "loss": 0.2816,
+      "step": 67260
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8377623438715152e-05,
+      "loss": 0.2801,
+      "step": 67265
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8374800814564627e-05,
+      "loss": 0.289,
+      "step": 67270
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.837197814659455e-05,
+      "loss": 0.276,
+      "step": 67275
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8369155434841554e-05,
+      "loss": 0.2614,
+      "step": 67280
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8366332679342305e-05,
+      "loss": 0.2795,
+      "step": 67285
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.836350988013343e-05,
+      "loss": 0.2556,
+      "step": 67290
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8360687037251603e-05,
+      "loss": 0.2745,
+      "step": 67295
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.835786415073347e-05,
+      "loss": 0.2803,
+      "step": 67300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8355041220615685e-05,
+      "loss": 0.2531,
+      "step": 67305
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8352218246934897e-05,
+      "loss": 0.2607,
+      "step": 67310
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8349395229727766e-05,
+      "loss": 0.2564,
+      "step": 67315
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8346572169030938e-05,
+      "loss": 0.2854,
+      "step": 67320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8344313689185543e-05,
+      "loss": 0.2996,
+      "step": 67325
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.834149055029965e-05,
+      "loss": 0.2884,
+      "step": 67330
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.833866736802671e-05,
+      "loss": 0.2499,
+      "step": 67335
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8335844142403366e-05,
+      "loss": 0.245,
+      "step": 67340
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8333020873466297e-05,
+      "loss": 0.2702,
+      "step": 67345
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8330197561252142e-05,
+      "loss": 0.2931,
+      "step": 67350
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8327374205797562e-05,
+      "loss": 0.2842,
+      "step": 67355
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8324550807139223e-05,
+      "loss": 0.2538,
+      "step": 67360
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8321727365313788e-05,
+      "loss": 0.2573,
+      "step": 67365
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8318903880357906e-05,
+      "loss": 0.266,
+      "step": 67370
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.831608035230825e-05,
+      "loss": 0.2564,
+      "step": 67375
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.831325678120147e-05,
+      "loss": 0.2514,
+      "step": 67380
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.831043316707424e-05,
+      "loss": 0.2565,
+      "step": 67385
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.830760950996321e-05,
+      "loss": 0.2854,
+      "step": 67390
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.830478580990506e-05,
+      "loss": 0.2671,
+      "step": 67395
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8301962066936437e-05,
+      "loss": 0.2981,
+      "step": 67400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.829913828109401e-05,
+      "loss": 0.2826,
+      "step": 67405
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8296314452414453e-05,
+      "loss": 0.2819,
+      "step": 67410
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8293490580934422e-05,
+      "loss": 0.2803,
+      "step": 67415
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.829066666669059e-05,
+      "loss": 0.2906,
+      "step": 67420
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8287842709719618e-05,
+      "loss": 0.27,
+      "step": 67425
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8285018710058176e-05,
+      "loss": 0.2333,
+      "step": 67430
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.828219466774293e-05,
+      "loss": 0.2658,
+      "step": 67435
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8279370582810544e-05,
+      "loss": 0.2773,
+      "step": 67440
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8276546455297694e-05,
+      "loss": 0.2568,
+      "step": 67445
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8273722285241056e-05,
+      "loss": 0.2769,
+      "step": 67450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8270898072677282e-05,
+      "loss": 0.2584,
+      "step": 67455
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8268073817643053e-05,
+      "loss": 0.2837,
+      "step": 67460
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.826524952017504e-05,
+      "loss": 0.2787,
+      "step": 67465
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8262425180309927e-05,
+      "loss": 0.2792,
+      "step": 67470
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8259600798084356e-05,
+      "loss": 0.2781,
+      "step": 67475
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8256776373535026e-05,
+      "loss": 0.2648,
+      "step": 67480
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.82539519066986e-05,
+      "loss": 0.2737,
+      "step": 67485
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8251127397611748e-05,
+      "loss": 0.256,
+      "step": 67490
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.8248302846311154e-05,
+      "loss": 0.2713,
+      "step": 67495
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.824547825283349e-05,
+      "loss": 0.2877,
+      "step": 67500
+    },
+    {
+      "epoch": 2.37,
+      "eval_loss": 0.2662014961242676,
+      "eval_runtime": 10.538,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 67500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8242653617215424e-05,
+      "loss": 0.2656,
+      "step": 67505
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8239828939493644e-05,
+      "loss": 0.2672,
+      "step": 67510
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8237004219704817e-05,
+      "loss": 0.2692,
+      "step": 67515
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.823417945788562e-05,
+      "loss": 0.277,
+      "step": 67520
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8231354654072733e-05,
+      "loss": 0.2665,
+      "step": 67525
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8228529808302845e-05,
+      "loss": 0.2728,
+      "step": 67530
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8225704920612616e-05,
+      "loss": 0.2625,
+      "step": 67535
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8222879991038736e-05,
+      "loss": 0.2947,
+      "step": 67540
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8220055019617887e-05,
+      "loss": 0.2678,
+      "step": 67545
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.821723000638674e-05,
+      "loss": 0.2809,
+      "step": 67550
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.821440495138199e-05,
+      "loss": 0.2604,
+      "step": 67555
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.82115798546403e-05,
+      "loss": 0.2859,
+      "step": 67560
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8208754716198376e-05,
+      "loss": 0.2527,
+      "step": 67565
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8205929536092878e-05,
+      "loss": 0.2474,
+      "step": 67570
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8203104314360495e-05,
+      "loss": 0.2517,
+      "step": 67575
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8200279051037914e-05,
+      "loss": 0.305,
+      "step": 67580
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8197453746161827e-05,
+      "loss": 0.2641,
+      "step": 67585
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.81946283997689e-05,
+      "loss": 0.2854,
+      "step": 67590
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.819180301189584e-05,
+      "loss": 0.2652,
+      "step": 67595
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8188977582579308e-05,
+      "loss": 0.2688,
+      "step": 67600
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8186152111856013e-05,
+      "loss": 0.273,
+      "step": 67605
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8183326599762627e-05,
+      "loss": 0.2844,
+      "step": 67610
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.818050104633585e-05,
+      "loss": 0.2694,
+      "step": 67615
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8177675451612358e-05,
+      "loss": 0.2896,
+      "step": 67620
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8174849815628846e-05,
+      "loss": 0.2841,
+      "step": 67625
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8172024138422e-05,
+      "loss": 0.2586,
+      "step": 67630
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.816919842002851e-05,
+      "loss": 0.2634,
+      "step": 67635
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.816637266048507e-05,
+      "loss": 0.2691,
+      "step": 67640
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.816354685982837e-05,
+      "loss": 0.2768,
+      "step": 67645
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8160721018095094e-05,
+      "loss": 0.2891,
+      "step": 67650
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.815789513532194e-05,
+      "loss": 0.2806,
+      "step": 67655
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8155069211545594e-05,
+      "loss": 0.272,
+      "step": 67660
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8152243246802757e-05,
+      "loss": 0.2733,
+      "step": 67665
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8149417241130122e-05,
+      "loss": 0.266,
+      "step": 67670
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8146591194564377e-05,
+      "loss": 0.2856,
+      "step": 67675
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8143765107142218e-05,
+      "loss": 0.2567,
+      "step": 67680
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8140938978900345e-05,
+      "loss": 0.284,
+      "step": 67685
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.813811280987545e-05,
+      "loss": 0.2534,
+      "step": 67690
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8135286600104222e-05,
+      "loss": 0.2557,
+      "step": 67695
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.813246034962337e-05,
+      "loss": 0.252,
+      "step": 67700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.812963405846958e-05,
+      "loss": 0.2668,
+      "step": 67705
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8126807726679556e-05,
+      "loss": 0.2756,
+      "step": 67710
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.812398135428999e-05,
+      "loss": 0.2436,
+      "step": 67715
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.812115494133759e-05,
+      "loss": 0.2669,
+      "step": 67720
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8118328487859053e-05,
+      "loss": 0.2457,
+      "step": 67725
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8115501993891074e-05,
+      "loss": 0.3102,
+      "step": 67730
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.811267545947035e-05,
+      "loss": 0.2651,
+      "step": 67735
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8109848884633588e-05,
+      "loss": 0.2928,
+      "step": 67740
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8107022269417488e-05,
+      "loss": 0.2644,
+      "step": 67745
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8104195613858754e-05,
+      "loss": 0.2744,
+      "step": 67750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8101368917994082e-05,
+      "loss": 0.2925,
+      "step": 67755
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8098542181860183e-05,
+      "loss": 0.2604,
+      "step": 67760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.809571540549375e-05,
+      "loss": 0.2893,
+      "step": 67765
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.80928885889315e-05,
+      "loss": 0.2932,
+      "step": 67770
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.809006173221012e-05,
+      "loss": 0.2549,
+      "step": 67775
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8087234835366333e-05,
+      "loss": 0.2662,
+      "step": 67780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.8084407898436837e-05,
+      "loss": 0.2651,
+      "step": 67785
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8081580921458335e-05,
+      "loss": 0.2785,
+      "step": 67790
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8078753904467532e-05,
+      "loss": 0.2767,
+      "step": 67795
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8075926847501137e-05,
+      "loss": 0.2838,
+      "step": 67800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8073099750595866e-05,
+      "loss": 0.2798,
+      "step": 67805
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8070272613788417e-05,
+      "loss": 0.276,
+      "step": 67810
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8067445437115504e-05,
+      "loss": 0.2661,
+      "step": 67815
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8064618220613826e-05,
+      "loss": 0.2742,
+      "step": 67820
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8061790964320108e-05,
+      "loss": 0.2534,
+      "step": 67825
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8058963668271047e-05,
+      "loss": 0.2752,
+      "step": 67830
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8056136332503362e-05,
+      "loss": 0.2537,
+      "step": 67835
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.805330895705376e-05,
+      "loss": 0.2581,
+      "step": 67840
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.805048154195895e-05,
+      "loss": 0.2517,
+      "step": 67845
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.804765408725565e-05,
+      "loss": 0.2553,
+      "step": 67850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8044826592980577e-05,
+      "loss": 0.2766,
+      "step": 67855
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8041999059170426e-05,
+      "loss": 0.2772,
+      "step": 67860
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8039171485861932e-05,
+      "loss": 0.2644,
+      "step": 67865
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8036343873091793e-05,
+      "loss": 0.2892,
+      "step": 67870
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.803351622089673e-05,
+      "loss": 0.2932,
+      "step": 67875
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.803068852931346e-05,
+      "loss": 0.2749,
+      "step": 67880
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8027860798378697e-05,
+      "loss": 0.2781,
+      "step": 67885
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8025033028129155e-05,
+      "loss": 0.2682,
+      "step": 67890
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.802220521860155e-05,
+      "loss": 0.2645,
+      "step": 67895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8019377369832605e-05,
+      "loss": 0.2776,
+      "step": 67900
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8016549481859035e-05,
+      "loss": 0.2684,
+      "step": 67905
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.801372155471756e-05,
+      "loss": 0.2765,
+      "step": 67910
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8010893588444892e-05,
+      "loss": 0.2764,
+      "step": 67915
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8008065583077763e-05,
+      "loss": 0.2738,
+      "step": 67920
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8005237538652878e-05,
+      "loss": 0.2562,
+      "step": 67925
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.8002409455206964e-05,
+      "loss": 0.2766,
+      "step": 67930
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.799958133277674e-05,
+      "loss": 0.272,
+      "step": 67935
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7996753171398937e-05,
+      "loss": 0.2884,
+      "step": 67940
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7993924971110262e-05,
+      "loss": 0.2617,
+      "step": 67945
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7991096731947454e-05,
+      "loss": 0.2923,
+      "step": 67950
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.798826845394722e-05,
+      "loss": 0.2731,
+      "step": 67955
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7985440137146283e-05,
+      "loss": 0.2692,
+      "step": 67960
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7982611781581382e-05,
+      "loss": 0.2848,
+      "step": 67965
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7979783387289233e-05,
+      "loss": 0.2871,
+      "step": 67970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7976954954306557e-05,
+      "loss": 0.2772,
+      "step": 67975
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7974126482670078e-05,
+      "loss": 0.2792,
+      "step": 67980
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7971297972416527e-05,
+      "loss": 0.2637,
+      "step": 67985
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7968469423582637e-05,
+      "loss": 0.264,
+      "step": 67990
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7965640836205127e-05,
+      "loss": 0.2702,
+      "step": 67995
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7962812210320726e-05,
+      "loss": 0.2813,
+      "step": 68000
+    },
+    {
+      "epoch": 2.39,
+      "eval_loss": 0.2646848261356354,
+      "eval_runtime": 10.557,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 9.472,
+      "step": 68000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7959983545966155e-05,
+      "loss": 0.2767,
+      "step": 68005
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7957154843178153e-05,
+      "loss": 0.2649,
+      "step": 68010
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7954326101993445e-05,
+      "loss": 0.2803,
+      "step": 68015
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7951497322448757e-05,
+      "loss": 0.2935,
+      "step": 68020
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.794866850458082e-05,
+      "loss": 0.2573,
+      "step": 68025
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.794583964842637e-05,
+      "loss": 0.2796,
+      "step": 68030
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7943010754022137e-05,
+      "loss": 0.2686,
+      "step": 68035
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7940181821404832e-05,
+      "loss": 0.263,
+      "step": 68040
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.793735285061122e-05,
+      "loss": 0.2584,
+      "step": 68045
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7934523841678013e-05,
+      "loss": 0.2866,
+      "step": 68050
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.793169479464195e-05,
+      "loss": 0.2616,
+      "step": 68055
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7928865709539754e-05,
+      "loss": 0.253,
+      "step": 68060
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7926036586408172e-05,
+      "loss": 0.2722,
+      "step": 68065
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.7923207425283933e-05,
+      "loss": 0.2763,
+      "step": 68070
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7920378226203775e-05,
+      "loss": 0.2651,
+      "step": 68075
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7917548989204422e-05,
+      "loss": 0.2692,
+      "step": 68080
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7914719714322628e-05,
+      "loss": 0.2624,
+      "step": 68085
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7911890401595114e-05,
+      "loss": 0.2823,
+      "step": 68090
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.790906105105862e-05,
+      "loss": 0.2674,
+      "step": 68095
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7906231662749886e-05,
+      "loss": 0.2812,
+      "step": 68100
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.790340223670566e-05,
+      "loss": 0.262,
+      "step": 68105
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.790057277296266e-05,
+      "loss": 0.2699,
+      "step": 68110
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7897743271557634e-05,
+      "loss": 0.2597,
+      "step": 68115
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.789491373252732e-05,
+      "loss": 0.2699,
+      "step": 68120
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7892084155908465e-05,
+      "loss": 0.2745,
+      "step": 68125
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7889254541737795e-05,
+      "loss": 0.2705,
+      "step": 68130
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7886424890052067e-05,
+      "loss": 0.2734,
+      "step": 68135
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7883595200888014e-05,
+      "loss": 0.2759,
+      "step": 68140
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7880765474282374e-05,
+      "loss": 0.2572,
+      "step": 68145
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7877935710271884e-05,
+      "loss": 0.266,
+      "step": 68150
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7875105908893305e-05,
+      "loss": 0.2584,
+      "step": 68155
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7872276070183373e-05,
+      "loss": 0.2862,
+      "step": 68160
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7869446194178827e-05,
+      "loss": 0.2482,
+      "step": 68165
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7866616280916412e-05,
+      "loss": 0.2581,
+      "step": 68170
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7863786330432873e-05,
+      "loss": 0.2524,
+      "step": 68175
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.786095634276496e-05,
+      "loss": 0.25,
+      "step": 68180
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7858126317949407e-05,
+      "loss": 0.2789,
+      "step": 68185
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.785529625602298e-05,
+      "loss": 0.2712,
+      "step": 68190
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.78524661570224e-05,
+      "loss": 0.2741,
+      "step": 68195
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7849636020984425e-05,
+      "loss": 0.2724,
+      "step": 68200
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.784680584794581e-05,
+      "loss": 0.2737,
+      "step": 68205
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7843975637943303e-05,
+      "loss": 0.2678,
+      "step": 68210
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.784114539101364e-05,
+      "loss": 0.269,
+      "step": 68215
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.783831510719358e-05,
+      "loss": 0.2992,
+      "step": 68220
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7835484786519868e-05,
+      "loss": 0.302,
+      "step": 68225
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7832654429029248e-05,
+      "loss": 0.2639,
+      "step": 68230
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7829824034758483e-05,
+      "loss": 0.2903,
+      "step": 68235
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7826993603744323e-05,
+      "loss": 0.2771,
+      "step": 68240
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7824163136023506e-05,
+      "loss": 0.2762,
+      "step": 68245
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.782133263163279e-05,
+      "loss": 0.2731,
+      "step": 68250
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7818502090608937e-05,
+      "loss": 0.294,
+      "step": 68255
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7815671512988685e-05,
+      "loss": 0.2546,
+      "step": 68260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7812840898808797e-05,
+      "loss": 0.2527,
+      "step": 68265
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7810010248106024e-05,
+      "loss": 0.2724,
+      "step": 68270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7807179560917123e-05,
+      "loss": 0.2793,
+      "step": 68275
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7804348837278836e-05,
+      "loss": 0.263,
+      "step": 68280
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7801518077227935e-05,
+      "loss": 0.2702,
+      "step": 68285
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7798687280801166e-05,
+      "loss": 0.2803,
+      "step": 68290
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.779585644803529e-05,
+      "loss": 0.287,
+      "step": 68295
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.779302557896706e-05,
+      "loss": 0.262,
+      "step": 68300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7790194673633228e-05,
+      "loss": 0.2756,
+      "step": 68305
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.778736373207056e-05,
+      "loss": 0.2656,
+      "step": 68310
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7784532754315812e-05,
+      "loss": 0.2581,
+      "step": 68315
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.778170174040574e-05,
+      "loss": 0.2646,
+      "step": 68320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7778870690377108e-05,
+      "loss": 0.2627,
+      "step": 68325
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7776039604266667e-05,
+      "loss": 0.265,
+      "step": 68330
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7773208482111186e-05,
+      "loss": 0.2562,
+      "step": 68335
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.777037732394741e-05,
+      "loss": 0.2778,
+      "step": 68340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7767546129812123e-05,
+      "loss": 0.2595,
+      "step": 68345
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.7764714899742063e-05,
+      "loss": 0.2597,
+      "step": 68350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.776188363377401e-05,
+      "loss": 0.2698,
+      "step": 68355
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7759052331944713e-05,
+      "loss": 0.3009,
+      "step": 68360
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7756220994290938e-05,
+      "loss": 0.2374,
+      "step": 68365
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7753389620849452e-05,
+      "loss": 0.2609,
+      "step": 68370
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.775055821165702e-05,
+      "loss": 0.2568,
+      "step": 68375
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7747726766750405e-05,
+      "loss": 0.2988,
+      "step": 68380
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7744895286166357e-05,
+      "loss": 0.2684,
+      "step": 68385
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7742063769941662e-05,
+      "loss": 0.2592,
+      "step": 68390
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7739232218113076e-05,
+      "loss": 0.2616,
+      "step": 68395
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.773640063071736e-05,
+      "loss": 0.2722,
+      "step": 68400
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7733569007791284e-05,
+      "loss": 0.262,
+      "step": 68405
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.773073734937162e-05,
+      "loss": 0.276,
+      "step": 68410
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7727905655495135e-05,
+      "loss": 0.2585,
+      "step": 68415
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7725073926198587e-05,
+      "loss": 0.2751,
+      "step": 68420
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7722242161518746e-05,
+      "loss": 0.2607,
+      "step": 68425
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7719410361492393e-05,
+      "loss": 0.2654,
+      "step": 68430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7716578526156285e-05,
+      "loss": 0.2693,
+      "step": 68435
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7713746655547202e-05,
+      "loss": 0.2475,
+      "step": 68440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7710914749701893e-05,
+      "loss": 0.2863,
+      "step": 68445
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7708082808657148e-05,
+      "loss": 0.2811,
+      "step": 68450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7705250832449737e-05,
+      "loss": 0.2761,
+      "step": 68455
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7702418821116426e-05,
+      "loss": 0.27,
+      "step": 68460
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7699586774693986e-05,
+      "loss": 0.2699,
+      "step": 68465
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7696754693219197e-05,
+      "loss": 0.2681,
+      "step": 68470
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7693922576728816e-05,
+      "loss": 0.2782,
+      "step": 68475
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.769109042525962e-05,
+      "loss": 0.2844,
+      "step": 68480
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7688258238848403e-05,
+      "loss": 0.2863,
+      "step": 68485
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7685426017531917e-05,
+      "loss": 0.2722,
+      "step": 68490
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7682593761346948e-05,
+      "loss": 0.2669,
+      "step": 68495
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7679761470330262e-05,
+      "loss": 0.2725,
+      "step": 68500
+    },
+    {
+      "epoch": 2.41,
+      "eval_loss": 0.26537927985191345,
+      "eval_runtime": 10.5494,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 68500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7676929144518642e-05,
+      "loss": 0.2678,
+      "step": 68505
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.767409678394886e-05,
+      "loss": 0.2775,
+      "step": 68510
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.76712643886577e-05,
+      "loss": 0.2876,
+      "step": 68515
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7668431958681924e-05,
+      "loss": 0.2665,
+      "step": 68520
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7665599494058325e-05,
+      "loss": 0.2805,
+      "step": 68525
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7662766994823665e-05,
+      "loss": 0.2676,
+      "step": 68530
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.765993446101473e-05,
+      "loss": 0.2759,
+      "step": 68535
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7657101892668303e-05,
+      "loss": 0.2572,
+      "step": 68540
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7654269289821165e-05,
+      "loss": 0.2697,
+      "step": 68545
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7651436652510082e-05,
+      "loss": 0.2701,
+      "step": 68550
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7648603980771846e-05,
+      "loss": 0.2857,
+      "step": 68555
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7645771274643223e-05,
+      "loss": 0.2857,
+      "step": 68560
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7642938534161016e-05,
+      "loss": 0.2729,
+      "step": 68565
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7640105759361985e-05,
+      "loss": 0.279,
+      "step": 68570
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7637272950282927e-05,
+      "loss": 0.2692,
+      "step": 68575
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7634440106960614e-05,
+      "loss": 0.2535,
+      "step": 68580
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7631607229431834e-05,
+      "loss": 0.2822,
+      "step": 68585
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7628774317733362e-05,
+      "loss": 0.2818,
+      "step": 68590
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7625941371901997e-05,
+      "loss": 0.2645,
+      "step": 68595
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.762310839197451e-05,
+      "loss": 0.2604,
+      "step": 68600
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7620275377987686e-05,
+      "loss": 0.2736,
+      "step": 68605
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7617442329978323e-05,
+      "loss": 0.2814,
+      "step": 68610
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7614609247983193e-05,
+      "loss": 0.2776,
+      "step": 68615
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7611776132039074e-05,
+      "loss": 0.2643,
+      "step": 68620
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7608942982182774e-05,
+      "loss": 0.2721,
+      "step": 68625
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.760610979845107e-05,
+      "loss": 0.2682,
+      "step": 68630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.760327658088074e-05,
+      "loss": 0.2631,
+      "step": 68635
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.7600443329508578e-05,
+      "loss": 0.2766,
+      "step": 68640
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.759761004437138e-05,
+      "loss": 0.2757,
+      "step": 68645
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.759477672550593e-05,
+      "loss": 0.2499,
+      "step": 68650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.759194337294901e-05,
+      "loss": 0.2766,
+      "step": 68655
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7589109986737416e-05,
+      "loss": 0.2666,
+      "step": 68660
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7586276566907925e-05,
+      "loss": 0.2593,
+      "step": 68665
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7583443113497348e-05,
+      "loss": 0.264,
+      "step": 68670
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7580609626542457e-05,
+      "loss": 0.284,
+      "step": 68675
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.757777610608006e-05,
+      "loss": 0.2797,
+      "step": 68680
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.757494255214693e-05,
+      "loss": 0.2877,
+      "step": 68685
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7572108964779873e-05,
+      "loss": 0.2632,
+      "step": 68690
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7569275344015678e-05,
+      "loss": 0.2612,
+      "step": 68695
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.756644168989113e-05,
+      "loss": 0.2647,
+      "step": 68700
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.756360800244303e-05,
+      "loss": 0.2814,
+      "step": 68705
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7560774281708174e-05,
+      "loss": 0.2855,
+      "step": 68710
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7557940527723348e-05,
+      "loss": 0.2755,
+      "step": 68715
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.755510674052535e-05,
+      "loss": 0.2472,
+      "step": 68720
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7552272920150966e-05,
+      "loss": 0.2665,
+      "step": 68725
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.754943906663701e-05,
+      "loss": 0.2592,
+      "step": 68730
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7546605180020273e-05,
+      "loss": 0.2856,
+      "step": 68735
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7543771260337538e-05,
+      "loss": 0.263,
+      "step": 68740
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7540937307625613e-05,
+      "loss": 0.2834,
+      "step": 68745
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7538103321921287e-05,
+      "loss": 0.2778,
+      "step": 68750
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.753526930326137e-05,
+      "loss": 0.2541,
+      "step": 68755
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7532435251682643e-05,
+      "loss": 0.2764,
+      "step": 68760
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.752960116722192e-05,
+      "loss": 0.2536,
+      "step": 68765
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7526767049915992e-05,
+      "loss": 0.2796,
+      "step": 68770
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7523932899801657e-05,
+      "loss": 0.2645,
+      "step": 68775
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.752109871691571e-05,
+      "loss": 0.2534,
+      "step": 68780
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7518264501294964e-05,
+      "loss": 0.2548,
+      "step": 68785
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7515430252976215e-05,
+      "loss": 0.2627,
+      "step": 68790
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7512595971996265e-05,
+      "loss": 0.2931,
+      "step": 68795
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.750976165839191e-05,
+      "loss": 0.2687,
+      "step": 68800
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.750692731219995e-05,
+      "loss": 0.2772,
+      "step": 68805
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7504092933457187e-05,
+      "loss": 0.2637,
+      "step": 68810
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7501258522200436e-05,
+      "loss": 0.2547,
+      "step": 68815
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.749842407846649e-05,
+      "loss": 0.2584,
+      "step": 68820
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7495589602292154e-05,
+      "loss": 0.2928,
+      "step": 68825
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.749275509371423e-05,
+      "loss": 0.2746,
+      "step": 68830
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.748992055276952e-05,
+      "loss": 0.2673,
+      "step": 68835
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7487085979494838e-05,
+      "loss": 0.2751,
+      "step": 68840
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7484251373926977e-05,
+      "loss": 0.2683,
+      "step": 68845
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.748141673610276e-05,
+      "loss": 0.2614,
+      "step": 68850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7478582066058974e-05,
+      "loss": 0.2732,
+      "step": 68855
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7475747363832432e-05,
+      "loss": 0.2437,
+      "step": 68860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7472912629459946e-05,
+      "loss": 0.265,
+      "step": 68865
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7470077862978322e-05,
+      "loss": 0.2429,
+      "step": 68870
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7467243064424358e-05,
+      "loss": 0.2671,
+      "step": 68875
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7464408233834876e-05,
+      "loss": 0.2554,
+      "step": 68880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7461573371246674e-05,
+      "loss": 0.271,
+      "step": 68885
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.745873847669656e-05,
+      "loss": 0.2693,
+      "step": 68890
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.745590355022135e-05,
+      "loss": 0.2692,
+      "step": 68895
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7453068591857857e-05,
+      "loss": 0.2953,
+      "step": 68900
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7450233601642878e-05,
+      "loss": 0.2494,
+      "step": 68905
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7447398579613233e-05,
+      "loss": 0.2435,
+      "step": 68910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7444563525805728e-05,
+      "loss": 0.284,
+      "step": 68915
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.7441728440257174e-05,
+      "loss": 0.2486,
+      "step": 68920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 2.743889332300439e-05,
+      "loss": 0.255,
+      "step": 68925
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7436058174084188e-05,
+      "loss": 0.2625,
+      "step": 68930
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.743322299353337e-05,
+      "loss": 0.276,
+      "step": 68935
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7430387781388756e-05,
+      "loss": 0.2862,
+      "step": 68940
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.742755253768715e-05,
+      "loss": 0.2511,
+      "step": 68945
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.742471726246538e-05,
+      "loss": 0.2575,
+      "step": 68950
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7421881955760265e-05,
+      "loss": 0.2964,
+      "step": 68955
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7419046617608597e-05,
+      "loss": 0.2514,
+      "step": 68960
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7416211248047212e-05,
+      "loss": 0.2529,
+      "step": 68965
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.741337584711291e-05,
+      "loss": 0.2616,
+      "step": 68970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.741054041484251e-05,
+      "loss": 0.2552,
+      "step": 68975
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7407704951272833e-05,
+      "loss": 0.2736,
+      "step": 68980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7404869456440703e-05,
+      "loss": 0.2452,
+      "step": 68985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.740203393038292e-05,
+      "loss": 0.2571,
+      "step": 68990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.739919837313631e-05,
+      "loss": 0.2656,
+      "step": 68995
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7396362784737682e-05,
+      "loss": 0.2927,
+      "step": 69000
+    },
+    {
+      "epoch": 2.43,
+      "eval_loss": 0.2651735842227936,
+      "eval_runtime": 10.5525,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 69000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7393527165223875e-05,
+      "loss": 0.2494,
+      "step": 69005
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7390691514631693e-05,
+      "loss": 0.2765,
+      "step": 69010
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.738785583299796e-05,
+      "loss": 0.2723,
+      "step": 69015
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7385020120359483e-05,
+      "loss": 0.2716,
+      "step": 69020
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7382184376753096e-05,
+      "loss": 0.2638,
+      "step": 69025
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7379348602215616e-05,
+      "loss": 0.2854,
+      "step": 69030
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7376512796783867e-05,
+      "loss": 0.263,
+      "step": 69035
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.737367696049466e-05,
+      "loss": 0.2827,
+      "step": 69040
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7370841093384824e-05,
+      "loss": 0.295,
+      "step": 69045
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7368005195491176e-05,
+      "loss": 0.2831,
+      "step": 69050
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.736516926685054e-05,
+      "loss": 0.2635,
+      "step": 69055
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.736290050182495e-05,
+      "loss": 0.2612,
+      "step": 69060
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7360064517932533e-05,
+      "loss": 0.2668,
+      "step": 69065
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.735722850339623e-05,
+      "loss": 0.2729,
+      "step": 69070
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7354392458252876e-05,
+      "loss": 0.2574,
+      "step": 69075
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7351556382539283e-05,
+      "loss": 0.2612,
+      "step": 69080
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7348720276292284e-05,
+      "loss": 0.2877,
+      "step": 69085
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.734588413954869e-05,
+      "loss": 0.2746,
+      "step": 69090
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7343047972345347e-05,
+      "loss": 0.2714,
+      "step": 69095
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7340211774719064e-05,
+      "loss": 0.2651,
+      "step": 69100
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7337375546706677e-05,
+      "loss": 0.2713,
+      "step": 69105
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7334539288345012e-05,
+      "loss": 0.2508,
+      "step": 69110
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7331702999670883e-05,
+      "loss": 0.2733,
+      "step": 69115
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7328866680721134e-05,
+      "loss": 0.2548,
+      "step": 69120
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7326030331532586e-05,
+      "loss": 0.2789,
+      "step": 69125
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.732319395214207e-05,
+      "loss": 0.2904,
+      "step": 69130
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7320357542586405e-05,
+      "loss": 0.2713,
+      "step": 69135
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7317521102902433e-05,
+      "loss": 0.2806,
+      "step": 69140
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7314684633126964e-05,
+      "loss": 0.2691,
+      "step": 69145
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7311848133296853e-05,
+      "loss": 0.2929,
+      "step": 69150
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7309011603448914e-05,
+      "loss": 0.2708,
+      "step": 69155
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.730617504361998e-05,
+      "loss": 0.2812,
+      "step": 69160
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7303338453846884e-05,
+      "loss": 0.2603,
+      "step": 69165
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7300501834166452e-05,
+      "loss": 0.2747,
+      "step": 69170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7297665184615522e-05,
+      "loss": 0.2567,
+      "step": 69175
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7294828505230936e-05,
+      "loss": 0.2734,
+      "step": 69180
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.72919917960495e-05,
+      "loss": 0.2757,
+      "step": 69185
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.728915505710807e-05,
+      "loss": 0.2778,
+      "step": 69190
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.728631828844346e-05,
+      "loss": 0.2741,
+      "step": 69195
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7283481490092516e-05,
+      "loss": 0.2572,
+      "step": 69200
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.728064466209207e-05,
+      "loss": 0.2774,
+      "step": 69205
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 2.7277807804478965e-05,
+      "loss": 0.2795,
+      "step": 69210
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7274970917290017e-05,
+      "loss": 0.2495,
+      "step": 69215
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7272134000562076e-05,
+      "loss": 0.2688,
+      "step": 69220
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7269297054331965e-05,
+      "loss": 0.2808,
+      "step": 69225
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7266460078636534e-05,
+      "loss": 0.2556,
+      "step": 69230
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.726362307351261e-05,
+      "loss": 0.2753,
+      "step": 69235
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7260786038997045e-05,
+      "loss": 0.286,
+      "step": 69240
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7257948975126647e-05,
+      "loss": 0.2793,
+      "step": 69245
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7255111881938273e-05,
+      "loss": 0.2538,
+      "step": 69250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7252274759468754e-05,
+      "loss": 0.2577,
+      "step": 69255
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7249437607754937e-05,
+      "loss": 0.2796,
+      "step": 69260
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7246600426833656e-05,
+      "loss": 0.272,
+      "step": 69265
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7243763216741745e-05,
+      "loss": 0.2648,
+      "step": 69270
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.724092597751606e-05,
+      "loss": 0.2778,
+      "step": 69275
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.723808870919341e-05,
+      "loss": 0.2485,
+      "step": 69280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7235251411810657e-05,
+      "loss": 0.2654,
+      "step": 69285
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7232414085404643e-05,
+      "loss": 0.2599,
+      "step": 69290
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7229576730012207e-05,
+      "loss": 0.232,
+      "step": 69295
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7226739345670176e-05,
+      "loss": 0.2623,
+      "step": 69300
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7223901932415415e-05,
+      "loss": 0.2579,
+      "step": 69305
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7221064490284738e-05,
+      "loss": 0.2594,
+      "step": 69310
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.721822701931501e-05,
+      "loss": 0.2513,
+      "step": 69315
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7215389519543068e-05,
+      "loss": 0.3011,
+      "step": 69320
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7212551991005753e-05,
+      "loss": 0.256,
+      "step": 69325
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7209714433739903e-05,
+      "loss": 0.2623,
+      "step": 69330
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.720687684778237e-05,
+      "loss": 0.2762,
+      "step": 69335
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7204039233169992e-05,
+      "loss": 0.2626,
+      "step": 69340
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.720120158993963e-05,
+      "loss": 0.281,
+      "step": 69345
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.71983639181281e-05,
+      "loss": 0.2703,
+      "step": 69350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7195526217772278e-05,
+      "loss": 0.2765,
+      "step": 69355
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7192688488908984e-05,
+      "loss": 0.282,
+      "step": 69360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7189850731575077e-05,
+      "loss": 0.2712,
+      "step": 69365
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.71870129458074e-05,
+      "loss": 0.2931,
+      "step": 69370
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7184175131642803e-05,
+      "loss": 0.2832,
+      "step": 69375
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7181337289118137e-05,
+      "loss": 0.2733,
+      "step": 69380
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7178499418270238e-05,
+      "loss": 0.2858,
+      "step": 69385
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7175661519135958e-05,
+      "loss": 0.2905,
+      "step": 69390
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.717282359175215e-05,
+      "loss": 0.2579,
+      "step": 69395
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7169985636155663e-05,
+      "loss": 0.2739,
+      "step": 69400
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7167147652383334e-05,
+      "loss": 0.2675,
+      "step": 69405
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.716430964047203e-05,
+      "loss": 0.2827,
+      "step": 69410
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7161471600458588e-05,
+      "loss": 0.2544,
+      "step": 69415
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7158633532379863e-05,
+      "loss": 0.301,
+      "step": 69420
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7155795436272696e-05,
+      "loss": 0.2817,
+      "step": 69425
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.715295731217396e-05,
+      "loss": 0.2906,
+      "step": 69430
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.715011916012048e-05,
+      "loss": 0.2861,
+      "step": 69435
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.714728098014913e-05,
+      "loss": 0.2622,
+      "step": 69440
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7144442772296742e-05,
+      "loss": 0.2833,
+      "step": 69445
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7141604536600185e-05,
+      "loss": 0.246,
+      "step": 69450
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7138766273096296e-05,
+      "loss": 0.2596,
+      "step": 69455
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7135927981821947e-05,
+      "loss": 0.2619,
+      "step": 69460
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7133089662813977e-05,
+      "loss": 0.2508,
+      "step": 69465
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7130251316109233e-05,
+      "loss": 0.2653,
+      "step": 69470
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.712741294174459e-05,
+      "loss": 0.2602,
+      "step": 69475
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.71245745397569e-05,
+      "loss": 0.2717,
+      "step": 69480
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.7121736110182994e-05,
+      "loss": 0.2808,
+      "step": 69485
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 2.711889765305975e-05,
+      "loss": 0.2695,
+      "step": 69490
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.711605916842402e-05,
+      "loss": 0.281,
+      "step": 69495
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.711322065631265e-05,
+      "loss": 0.2699,
+      "step": 69500
+    },
+    {
+      "epoch": 2.45,
+      "eval_loss": 0.2653324604034424,
+      "eval_runtime": 10.5362,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 69500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.71103821167625e-05,
+      "loss": 0.2805,
+      "step": 69505
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7107543549810433e-05,
+      "loss": 0.2772,
+      "step": 69510
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7104704955493303e-05,
+      "loss": 0.2752,
+      "step": 69515
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7101866333847962e-05,
+      "loss": 0.2571,
+      "step": 69520
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.709902768491128e-05,
+      "loss": 0.2747,
+      "step": 69525
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7096189008720096e-05,
+      "loss": 0.262,
+      "step": 69530
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7093350305311288e-05,
+      "loss": 0.2717,
+      "step": 69535
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7090511574721705e-05,
+      "loss": 0.2975,
+      "step": 69540
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.708767281698821e-05,
+      "loss": 0.2871,
+      "step": 69545
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.708483403214765e-05,
+      "loss": 0.2548,
+      "step": 69550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7081995220236893e-05,
+      "loss": 0.269,
+      "step": 69555
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.707915638129281e-05,
+      "loss": 0.2799,
+      "step": 69560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7076317515352247e-05,
+      "loss": 0.2619,
+      "step": 69565
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.707347862245207e-05,
+      "loss": 0.2628,
+      "step": 69570
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7070639702629142e-05,
+      "loss": 0.2729,
+      "step": 69575
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.706780075592032e-05,
+      "loss": 0.2924,
+      "step": 69580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7064961782362462e-05,
+      "loss": 0.2756,
+      "step": 69585
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7062122781992438e-05,
+      "loss": 0.2735,
+      "step": 69590
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.705928375484712e-05,
+      "loss": 0.2717,
+      "step": 69595
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7056444700963346e-05,
+      "loss": 0.258,
+      "step": 69600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7053605620377998e-05,
+      "loss": 0.2592,
+      "step": 69605
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7050766513127934e-05,
+      "loss": 0.2745,
+      "step": 69610
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7047927379250016e-05,
+      "loss": 0.274,
+      "step": 69615
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7045088218781116e-05,
+      "loss": 0.2726,
+      "step": 69620
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7042249031758088e-05,
+      "loss": 0.2749,
+      "step": 69625
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.703940981821781e-05,
+      "loss": 0.2606,
+      "step": 69630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7036570578197134e-05,
+      "loss": 0.2564,
+      "step": 69635
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.703373131173293e-05,
+      "loss": 0.2579,
+      "step": 69640
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7030892018862066e-05,
+      "loss": 0.2936,
+      "step": 69645
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.702805269962141e-05,
+      "loss": 0.2619,
+      "step": 69650
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7025213354047824e-05,
+      "loss": 0.2891,
+      "step": 69655
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7022373982178182e-05,
+      "loss": 0.2682,
+      "step": 69660
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7019534584049333e-05,
+      "loss": 0.2676,
+      "step": 69665
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7016695159698174e-05,
+      "loss": 0.2531,
+      "step": 69670
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.701385570916155e-05,
+      "loss": 0.271,
+      "step": 69675
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7011016232476344e-05,
+      "loss": 0.2571,
+      "step": 69680
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7008176729679407e-05,
+      "loss": 0.2842,
+      "step": 69685
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7005337200807624e-05,
+      "loss": 0.2799,
+      "step": 69690
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.7002497645897855e-05,
+      "loss": 0.266,
+      "step": 69695
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6999658064986983e-05,
+      "loss": 0.2721,
+      "step": 69700
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6996818458111862e-05,
+      "loss": 0.2712,
+      "step": 69705
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6993978825309374e-05,
+      "loss": 0.2511,
+      "step": 69710
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6991139166616375e-05,
+      "loss": 0.2838,
+      "step": 69715
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.698829948206975e-05,
+      "loss": 0.2664,
+      "step": 69720
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6985459771706367e-05,
+      "loss": 0.2854,
+      "step": 69725
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.69826200355631e-05,
+      "loss": 0.2769,
+      "step": 69730
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.697978027367682e-05,
+      "loss": 0.2712,
+      "step": 69735
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6976940486084394e-05,
+      "loss": 0.2626,
+      "step": 69740
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6974100672822705e-05,
+      "loss": 0.2676,
+      "step": 69745
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6971260833928607e-05,
+      "loss": 0.2453,
+      "step": 69750
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6968420969439002e-05,
+      "loss": 0.2894,
+      "step": 69755
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6965581079390734e-05,
+      "loss": 0.2795,
+      "step": 69760
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6962741163820703e-05,
+      "loss": 0.2667,
+      "step": 69765
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.6959901222765764e-05,
+      "loss": 0.2664,
+      "step": 69770
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 2.69570612562628e-05,
+      "loss": 0.2816,
+      "step": 69775
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6954221264348684e-05,
+      "loss": 0.2869,
+      "step": 69780
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.69513812470603e-05,
+      "loss": 0.2762,
+      "step": 69785
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.694854120443451e-05,
+      "loss": 0.278,
+      "step": 69790
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.69457011365082e-05,
+      "loss": 0.2699,
+      "step": 69795
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.694286104331824e-05,
+      "loss": 0.2747,
+      "step": 69800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.694002092490151e-05,
+      "loss": 0.2802,
+      "step": 69805
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6937180781294885e-05,
+      "loss": 0.288,
+      "step": 69810
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6934340612535247e-05,
+      "loss": 0.2781,
+      "step": 69815
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6931500418659473e-05,
+      "loss": 0.284,
+      "step": 69820
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6928660199704435e-05,
+      "loss": 0.2531,
+      "step": 69825
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6925819955707014e-05,
+      "loss": 0.2725,
+      "step": 69830
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6922979686704097e-05,
+      "loss": 0.2799,
+      "step": 69835
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6920139392732552e-05,
+      "loss": 0.2975,
+      "step": 69840
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.691729907382926e-05,
+      "loss": 0.2886,
+      "step": 69845
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.691445873003111e-05,
+      "loss": 0.2714,
+      "step": 69850
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.691161836137497e-05,
+      "loss": 0.2786,
+      "step": 69855
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6908777967897718e-05,
+      "loss": 0.2714,
+      "step": 69860
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.690593754963625e-05,
+      "loss": 0.2534,
+      "step": 69865
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6903097106627447e-05,
+      "loss": 0.2817,
+      "step": 69870
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6900256638908174e-05,
+      "loss": 0.2838,
+      "step": 69875
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.689741614651532e-05,
+      "loss": 0.2638,
+      "step": 69880
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6894575629485762e-05,
+      "loss": 0.2688,
+      "step": 69885
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.68917350878564e-05,
+      "loss": 0.2442,
+      "step": 69890
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.68888945216641e-05,
+      "loss": 0.2397,
+      "step": 69895
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6886053930945755e-05,
+      "loss": 0.2594,
+      "step": 69900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6883213315738232e-05,
+      "loss": 0.257,
+      "step": 69905
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6880372676078423e-05,
+      "loss": 0.2749,
+      "step": 69910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6877532012003225e-05,
+      "loss": 0.2849,
+      "step": 69915
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.687469132354951e-05,
+      "loss": 0.2624,
+      "step": 69920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6871850610754156e-05,
+      "loss": 0.2645,
+      "step": 69925
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6869009873654066e-05,
+      "loss": 0.2821,
+      "step": 69930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6866169112286105e-05,
+      "loss": 0.2655,
+      "step": 69935
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.686332832668717e-05,
+      "loss": 0.2726,
+      "step": 69940
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6860487516894144e-05,
+      "loss": 0.2955,
+      "step": 69945
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.685764668294392e-05,
+      "loss": 0.2693,
+      "step": 69950
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6854805824873376e-05,
+      "loss": 0.2529,
+      "step": 69955
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6851964942719394e-05,
+      "loss": 0.2596,
+      "step": 69960
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.684912403651888e-05,
+      "loss": 0.2836,
+      "step": 69965
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.684628310630869e-05,
+      "loss": 0.2747,
+      "step": 69970
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6843442152125748e-05,
+      "loss": 0.2686,
+      "step": 69975
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6840601174006918e-05,
+      "loss": 0.2522,
+      "step": 69980
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6837760171989094e-05,
+      "loss": 0.2775,
+      "step": 69985
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6834919146109167e-05,
+      "loss": 0.2693,
+      "step": 69990
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.683207809640402e-05,
+      "loss": 0.2827,
+      "step": 69995
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6829237022910552e-05,
+      "loss": 0.238,
+      "step": 70000
+    },
+    {
+      "epoch": 2.46,
+      "eval_loss": 0.26445090770721436,
+      "eval_runtime": 10.5622,
+      "eval_samples_per_second": 9.468,
+      "eval_steps_per_second": 9.468,
+      "step": 70000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.682639592566565e-05,
+      "loss": 0.2826,
+      "step": 70005
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6823554804706192e-05,
+      "loss": 0.2633,
+      "step": 70010
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6820713660069087e-05,
+      "loss": 0.2642,
+      "step": 70015
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.68178724917912e-05,
+      "loss": 0.2651,
+      "step": 70020
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.681503129990945e-05,
+      "loss": 0.2732,
+      "step": 70025
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6812190084460704e-05,
+      "loss": 0.2791,
+      "step": 70030
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6809348845481873e-05,
+      "loss": 0.2664,
+      "step": 70035
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6806507583009837e-05,
+      "loss": 0.2579,
+      "step": 70040
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.680366629708149e-05,
+      "loss": 0.2769,
+      "step": 70045
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.680082498773372e-05,
+      "loss": 0.2742,
+      "step": 70050
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6797983655003433e-05,
+      "loss": 0.2812,
+      "step": 70055
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 2.6795142298927507e-05,
+      "loss": 0.2794,
+      "step": 70060
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6792300919542845e-05,
+      "loss": 0.2661,
+      "step": 70065
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6789459516886334e-05,
+      "loss": 0.2733,
+      "step": 70070
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.678661809099487e-05,
+      "loss": 0.2739,
+      "step": 70075
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6783776641905346e-05,
+      "loss": 0.2775,
+      "step": 70080
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6780935169654657e-05,
+      "loss": 0.2773,
+      "step": 70085
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6778093674279708e-05,
+      "loss": 0.2615,
+      "step": 70090
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6775252155817375e-05,
+      "loss": 0.3045,
+      "step": 70095
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6772410614304572e-05,
+      "loss": 0.2739,
+      "step": 70100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6769569049778175e-05,
+      "loss": 0.2818,
+      "step": 70105
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6766727462275098e-05,
+      "loss": 0.2558,
+      "step": 70110
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6763885851832228e-05,
+      "loss": 0.2704,
+      "step": 70115
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6761044218486457e-05,
+      "loss": 0.2763,
+      "step": 70120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6758202562274693e-05,
+      "loss": 0.2631,
+      "step": 70125
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6755360883233826e-05,
+      "loss": 0.2597,
+      "step": 70130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.675251918140075e-05,
+      "loss": 0.2679,
+      "step": 70135
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6749677456812376e-05,
+      "loss": 0.2464,
+      "step": 70140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6746835709505586e-05,
+      "loss": 0.2559,
+      "step": 70145
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.674399393951729e-05,
+      "loss": 0.2623,
+      "step": 70150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6741152146884386e-05,
+      "loss": 0.2735,
+      "step": 70155
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6738310331643755e-05,
+      "loss": 0.2737,
+      "step": 70160
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6735468493832322e-05,
+      "loss": 0.2854,
+      "step": 70165
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6732626633486973e-05,
+      "loss": 0.2754,
+      "step": 70170
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6729784750644605e-05,
+      "loss": 0.2867,
+      "step": 70175
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.672694284534213e-05,
+      "loss": 0.2767,
+      "step": 70180
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.672410091761643e-05,
+      "loss": 0.2667,
+      "step": 70185
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.672125896750441e-05,
+      "loss": 0.2848,
+      "step": 70190
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.671841699504299e-05,
+      "loss": 0.2816,
+      "step": 70195
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6715575000269055e-05,
+      "loss": 0.2844,
+      "step": 70200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6712732983219514e-05,
+      "loss": 0.2748,
+      "step": 70205
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6709890943931252e-05,
+      "loss": 0.2603,
+      "step": 70210
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.670704888244119e-05,
+      "loss": 0.2777,
+      "step": 70215
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.670420679878622e-05,
+      "loss": 0.2673,
+      "step": 70220
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6701364693003255e-05,
+      "loss": 0.2878,
+      "step": 70225
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.669852256512918e-05,
+      "loss": 0.2912,
+      "step": 70230
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.669568041520092e-05,
+      "loss": 0.2746,
+      "step": 70235
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6692838243255364e-05,
+      "loss": 0.2432,
+      "step": 70240
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6689996049329413e-05,
+      "loss": 0.2665,
+      "step": 70245
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6687153833459982e-05,
+      "loss": 0.2876,
+      "step": 70250
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.668431159568397e-05,
+      "loss": 0.3013,
+      "step": 70255
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.668146933603828e-05,
+      "loss": 0.2851,
+      "step": 70260
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6678627054559824e-05,
+      "loss": 0.2648,
+      "step": 70265
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6675784751285494e-05,
+      "loss": 0.2773,
+      "step": 70270
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6672942426252212e-05,
+      "loss": 0.2921,
+      "step": 70275
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.667010007949687e-05,
+      "loss": 0.2451,
+      "step": 70280
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.666725771105638e-05,
+      "loss": 0.2775,
+      "step": 70285
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6664415320967644e-05,
+      "loss": 0.2334,
+      "step": 70290
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6661572909267574e-05,
+      "loss": 0.2681,
+      "step": 70295
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6658730475993077e-05,
+      "loss": 0.2791,
+      "step": 70300
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6655888021181052e-05,
+      "loss": 0.2672,
+      "step": 70305
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.665304554486842e-05,
+      "loss": 0.2755,
+      "step": 70310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.665020304709208e-05,
+      "loss": 0.2966,
+      "step": 70315
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6647360527888943e-05,
+      "loss": 0.2693,
+      "step": 70320
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6644517987295903e-05,
+      "loss": 0.2733,
+      "step": 70325
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6641675425349894e-05,
+      "loss": 0.263,
+      "step": 70330
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6638832842087807e-05,
+      "loss": 0.2887,
+      "step": 70335
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.663599023754656e-05,
+      "loss": 0.2671,
+      "step": 70340
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 2.6633147611763047e-05,
+      "loss": 0.2613,
+      "step": 70345
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6630304964774195e-05,
+      "loss": 0.2824,
+      "step": 70350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6627462296616906e-05,
+      "loss": 0.2821,
+      "step": 70355
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6624619607328093e-05,
+      "loss": 0.2637,
+      "step": 70360
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6621776896944666e-05,
+      "loss": 0.283,
+      "step": 70365
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.661893416550354e-05,
+      "loss": 0.2689,
+      "step": 70370
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.661609141304161e-05,
+      "loss": 0.2667,
+      "step": 70375
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.66132486395958e-05,
+      "loss": 0.2465,
+      "step": 70380
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.661040584520302e-05,
+      "loss": 0.2643,
+      "step": 70385
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.660756302990019e-05,
+      "loss": 0.2812,
+      "step": 70390
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.66047201937242e-05,
+      "loss": 0.2553,
+      "step": 70395
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.660187733671199e-05,
+      "loss": 0.2678,
+      "step": 70400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6599034458900452e-05,
+      "loss": 0.2585,
+      "step": 70405
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.65961915603265e-05,
+      "loss": 0.2741,
+      "step": 70410
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6593348641027056e-05,
+      "loss": 0.2839,
+      "step": 70415
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6590505701039032e-05,
+      "loss": 0.2588,
+      "step": 70420
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6587662740399338e-05,
+      "loss": 0.2524,
+      "step": 70425
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6584819759144892e-05,
+      "loss": 0.2595,
+      "step": 70430
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.65819767573126e-05,
+      "loss": 0.2641,
+      "step": 70435
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6579133734939383e-05,
+      "loss": 0.2606,
+      "step": 70440
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6576290692062162e-05,
+      "loss": 0.2499,
+      "step": 70445
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.657344762871784e-05,
+      "loss": 0.2609,
+      "step": 70450
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6570604544943346e-05,
+      "loss": 0.2638,
+      "step": 70455
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.656776144077558e-05,
+      "loss": 0.2473,
+      "step": 70460
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6564918316251463e-05,
+      "loss": 0.2855,
+      "step": 70465
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6562075171407914e-05,
+      "loss": 0.2692,
+      "step": 70470
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6559232006281853e-05,
+      "loss": 0.2746,
+      "step": 70475
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6556388820910183e-05,
+      "loss": 0.2747,
+      "step": 70480
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6553545615329838e-05,
+      "loss": 0.2568,
+      "step": 70485
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.655070238957772e-05,
+      "loss": 0.2693,
+      "step": 70490
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6547859143690762e-05,
+      "loss": 0.265,
+      "step": 70495
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6545015877705865e-05,
+      "loss": 0.2696,
+      "step": 70500
+    },
+    {
+      "epoch": 2.48,
+      "eval_loss": 0.2644537389278412,
+      "eval_runtime": 10.5368,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 70500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6542172591659962e-05,
+      "loss": 0.2577,
+      "step": 70505
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6539329285589963e-05,
+      "loss": 0.2593,
+      "step": 70510
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.653648595953278e-05,
+      "loss": 0.2707,
+      "step": 70515
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6533642613525345e-05,
+      "loss": 0.2658,
+      "step": 70520
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6530799247604577e-05,
+      "loss": 0.2745,
+      "step": 70525
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6527955861807385e-05,
+      "loss": 0.2485,
+      "step": 70530
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6525112456170692e-05,
+      "loss": 0.2742,
+      "step": 70535
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6522269030731416e-05,
+      "loss": 0.2785,
+      "step": 70540
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.651942558552648e-05,
+      "loss": 0.2837,
+      "step": 70545
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.651658212059281e-05,
+      "loss": 0.2881,
+      "step": 70550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6513738635967316e-05,
+      "loss": 0.2577,
+      "step": 70555
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6510895131686932e-05,
+      "loss": 0.2778,
+      "step": 70560
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.650805160778856e-05,
+      "loss": 0.2704,
+      "step": 70565
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6505208064309134e-05,
+      "loss": 0.2713,
+      "step": 70570
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.650236450128557e-05,
+      "loss": 0.2803,
+      "step": 70575
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6499520918754805e-05,
+      "loss": 0.2946,
+      "step": 70580
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6496677316753738e-05,
+      "loss": 0.2721,
+      "step": 70585
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6493833695319308e-05,
+      "loss": 0.2775,
+      "step": 70590
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.649099005448843e-05,
+      "loss": 0.2534,
+      "step": 70595
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.648814639429803e-05,
+      "loss": 0.2831,
+      "step": 70600
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6485302714785027e-05,
+      "loss": 0.2819,
+      "step": 70605
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6482459015986348e-05,
+      "loss": 0.256,
+      "step": 70610
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6479615297938914e-05,
+      "loss": 0.2702,
+      "step": 70615
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6476771560679657e-05,
+      "loss": 0.2729,
+      "step": 70620
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.647392780424549e-05,
+      "loss": 0.2365,
+      "step": 70625
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 2.6471084028673344e-05,
+      "loss": 0.2691,
+      "step": 70630
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6468240234000134e-05,
+      "loss": 0.2795,
+      "step": 70635
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.64653964202628e-05,
+      "loss": 0.2887,
+      "step": 70640
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6462552587498256e-05,
+      "loss": 0.2906,
+      "step": 70645
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.645970873574343e-05,
+      "loss": 0.2476,
+      "step": 70650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6456864865035246e-05,
+      "loss": 0.2553,
+      "step": 70655
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6454020975410636e-05,
+      "loss": 0.2771,
+      "step": 70660
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6451177066906525e-05,
+      "loss": 0.2717,
+      "step": 70665
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6448333139559834e-05,
+      "loss": 0.2877,
+      "step": 70670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.644548919340749e-05,
+      "loss": 0.2463,
+      "step": 70675
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.644264522848642e-05,
+      "loss": 0.2666,
+      "step": 70680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6439801244833552e-05,
+      "loss": 0.2872,
+      "step": 70685
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6436957242485817e-05,
+      "loss": 0.255,
+      "step": 70690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.643411322148014e-05,
+      "loss": 0.2743,
+      "step": 70695
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.643126918185344e-05,
+      "loss": 0.2633,
+      "step": 70700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6428425123642664e-05,
+      "loss": 0.2505,
+      "step": 70705
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6425581046884718e-05,
+      "loss": 0.2791,
+      "step": 70710
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.642273695161655e-05,
+      "loss": 0.2804,
+      "step": 70715
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6419892837875077e-05,
+      "loss": 0.2931,
+      "step": 70720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6417048705697232e-05,
+      "loss": 0.276,
+      "step": 70725
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.641420455511994e-05,
+      "loss": 0.2728,
+      "step": 70730
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.641136038618014e-05,
+      "loss": 0.2851,
+      "step": 70735
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.640851619891475e-05,
+      "loss": 0.2897,
+      "step": 70740
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6405671993360716e-05,
+      "loss": 0.2527,
+      "step": 70745
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6402827769554943e-05,
+      "loss": 0.2513,
+      "step": 70750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6399983527534388e-05,
+      "loss": 0.2765,
+      "step": 70755
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6397139267335963e-05,
+      "loss": 0.2954,
+      "step": 70760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6394294988996604e-05,
+      "loss": 0.2749,
+      "step": 70765
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6391450692553243e-05,
+      "loss": 0.2695,
+      "step": 70770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6388606378042812e-05,
+      "loss": 0.2888,
+      "step": 70775
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.638576204550225e-05,
+      "loss": 0.2716,
+      "step": 70780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6382917694968474e-05,
+      "loss": 0.2771,
+      "step": 70785
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.638007332647843e-05,
+      "loss": 0.2758,
+      "step": 70790
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6377228940069032e-05,
+      "loss": 0.2839,
+      "step": 70795
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6374384535777235e-05,
+      "loss": 0.2812,
+      "step": 70800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6371540113639952e-05,
+      "loss": 0.2791,
+      "step": 70805
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6368695673694133e-05,
+      "loss": 0.2697,
+      "step": 70810
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6365851215976695e-05,
+      "loss": 0.3063,
+      "step": 70815
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6363006740524583e-05,
+      "loss": 0.2718,
+      "step": 70820
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.636016224737472e-05,
+      "loss": 0.2651,
+      "step": 70825
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6357317736564058e-05,
+      "loss": 0.2578,
+      "step": 70830
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6354473208129516e-05,
+      "loss": 0.2959,
+      "step": 70835
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.635162866210803e-05,
+      "loss": 0.2696,
+      "step": 70840
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.634878409853654e-05,
+      "loss": 0.2828,
+      "step": 70845
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6345939517451968e-05,
+      "loss": 0.2892,
+      "step": 70850
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6343094918891264e-05,
+      "loss": 0.268,
+      "step": 70855
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.634025030289136e-05,
+      "loss": 0.28,
+      "step": 70860
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6337405669489183e-05,
+      "loss": 0.301,
+      "step": 70865
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.633456101872168e-05,
+      "loss": 0.2413,
+      "step": 70870
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.633171635062578e-05,
+      "loss": 0.2937,
+      "step": 70875
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6328871665238425e-05,
+      "loss": 0.2646,
+      "step": 70880
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.632602696259654e-05,
+      "loss": 0.2688,
+      "step": 70885
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.632318224273707e-05,
+      "loss": 0.2705,
+      "step": 70890
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6320337505696956e-05,
+      "loss": 0.2516,
+      "step": 70895
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6317492751513128e-05,
+      "loss": 0.2815,
+      "step": 70900
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.6314647980222524e-05,
+      "loss": 0.2841,
+      "step": 70905
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.631180319186208e-05,
+      "loss": 0.2748,
+      "step": 70910
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 2.630895838646874e-05,
+      "loss": 0.2425,
+      "step": 70915
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.630611356407943e-05,
+      "loss": 0.2694,
+      "step": 70920
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6303268724731112e-05,
+      "loss": 0.2638,
+      "step": 70925
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6300423868460688e-05,
+      "loss": 0.2543,
+      "step": 70930
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6297578995305133e-05,
+      "loss": 0.2726,
+      "step": 70935
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.629473410530136e-05,
+      "loss": 0.2665,
+      "step": 70940
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.629188919848633e-05,
+      "loss": 0.2788,
+      "step": 70945
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6289044274896963e-05,
+      "loss": 0.2554,
+      "step": 70950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.62861993345702e-05,
+      "loss": 0.2959,
+      "step": 70955
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.628335437754299e-05,
+      "loss": 0.2694,
+      "step": 70960
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6280509403852277e-05,
+      "loss": 0.2644,
+      "step": 70965
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.627766441353498e-05,
+      "loss": 0.2655,
+      "step": 70970
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6274819406628066e-05,
+      "loss": 0.2675,
+      "step": 70975
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6271974383168453e-05,
+      "loss": 0.2728,
+      "step": 70980
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6269129343193093e-05,
+      "loss": 0.246,
+      "step": 70985
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6266284286738924e-05,
+      "loss": 0.2662,
+      "step": 70990
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6263439213842893e-05,
+      "loss": 0.2847,
+      "step": 70995
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6260594124541932e-05,
+      "loss": 0.2974,
+      "step": 71000
+    },
+    {
+      "epoch": 2.5,
+      "eval_loss": 0.26469552516937256,
+      "eval_runtime": 10.5298,
+      "eval_samples_per_second": 9.497,
+      "eval_steps_per_second": 9.497,
+      "step": 71000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6257749018872983e-05,
+      "loss": 0.2703,
+      "step": 71005
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6254903896872996e-05,
+      "loss": 0.2498,
+      "step": 71010
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6252058758578907e-05,
+      "loss": 0.2616,
+      "step": 71015
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6249213604027667e-05,
+      "loss": 0.2757,
+      "step": 71020
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6246368433256206e-05,
+      "loss": 0.2803,
+      "step": 71025
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6243523246301477e-05,
+      "loss": 0.2634,
+      "step": 71030
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6240678043200416e-05,
+      "loss": 0.2533,
+      "step": 71035
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6237832823989967e-05,
+      "loss": 0.2823,
+      "step": 71040
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.623498758870707e-05,
+      "loss": 0.2766,
+      "step": 71045
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6232142337388684e-05,
+      "loss": 0.3033,
+      "step": 71050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6229297070071734e-05,
+      "loss": 0.2837,
+      "step": 71055
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6226451786793177e-05,
+      "loss": 0.2547,
+      "step": 71060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6223606487589946e-05,
+      "loss": 0.2853,
+      "step": 71065
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6220761172498993e-05,
+      "loss": 0.2503,
+      "step": 71070
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6217915841557255e-05,
+      "loss": 0.2546,
+      "step": 71075
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6215070494801693e-05,
+      "loss": 0.2663,
+      "step": 71080
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6212225132269234e-05,
+      "loss": 0.2875,
+      "step": 71085
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6209379753996833e-05,
+      "loss": 0.2916,
+      "step": 71090
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6206534360021427e-05,
+      "loss": 0.286,
+      "step": 71095
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6203688950379978e-05,
+      "loss": 0.2548,
+      "step": 71100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6200843525109418e-05,
+      "loss": 0.2594,
+      "step": 71105
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.61979980842467e-05,
+      "loss": 0.2408,
+      "step": 71110
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.619515262782876e-05,
+      "loss": 0.2577,
+      "step": 71115
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6192307155892547e-05,
+      "loss": 0.2733,
+      "step": 71120
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6189461668475018e-05,
+      "loss": 0.2338,
+      "step": 71125
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.618661616561311e-05,
+      "loss": 0.2518,
+      "step": 71130
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.618377064734378e-05,
+      "loss": 0.2524,
+      "step": 71135
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.618092511370396e-05,
+      "loss": 0.244,
+      "step": 71140
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6178079564730613e-05,
+      "loss": 0.2627,
+      "step": 71145
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.617523400046067e-05,
+      "loss": 0.2476,
+      "step": 71150
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6172388420931093e-05,
+      "loss": 0.2441,
+      "step": 71155
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6169542826178822e-05,
+      "loss": 0.2498,
+      "step": 71160
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6166697216240816e-05,
+      "loss": 0.2706,
+      "step": 71165
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.616385159115401e-05,
+      "loss": 0.275,
+      "step": 71170
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6161005950955358e-05,
+      "loss": 0.2578,
+      "step": 71175
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6158160295681804e-05,
+      "loss": 0.2837,
+      "step": 71180
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6155314625370315e-05,
+      "loss": 0.2616,
+      "step": 71185
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.615246894005782e-05,
+      "loss": 0.2713,
+      "step": 71190
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 2.6149623239781272e-05,
+      "loss": 0.2844,
+      "step": 71195
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6146777524577627e-05,
+      "loss": 0.283,
+      "step": 71200
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6143931794483827e-05,
+      "loss": 0.2695,
+      "step": 71205
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6141086049536833e-05,
+      "loss": 0.2553,
+      "step": 71210
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6138240289773592e-05,
+      "loss": 0.2613,
+      "step": 71215
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6135394515231037e-05,
+      "loss": 0.2718,
+      "step": 71220
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6132548725946147e-05,
+      "loss": 0.2702,
+      "step": 71225
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6129702921955844e-05,
+      "loss": 0.2659,
+      "step": 71230
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.612685710329711e-05,
+      "loss": 0.2463,
+      "step": 71235
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6124011270006866e-05,
+      "loss": 0.2611,
+      "step": 71240
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.612116542212208e-05,
+      "loss": 0.2812,
+      "step": 71245
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.61183195596797e-05,
+      "loss": 0.2586,
+      "step": 71250
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.611547368271668e-05,
+      "loss": 0.2627,
+      "step": 71255
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.611262779126996e-05,
+      "loss": 0.2599,
+      "step": 71260
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.610978188537651e-05,
+      "loss": 0.2799,
+      "step": 71265
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6106935965073275e-05,
+      "loss": 0.261,
+      "step": 71270
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6104090030397206e-05,
+      "loss": 0.2442,
+      "step": 71275
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6101244081385257e-05,
+      "loss": 0.2656,
+      "step": 71280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.609839811807437e-05,
+      "loss": 0.263,
+      "step": 71285
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6095552140501512e-05,
+      "loss": 0.251,
+      "step": 71290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.609270614870363e-05,
+      "loss": 0.2839,
+      "step": 71295
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6089860142717682e-05,
+      "loss": 0.2422,
+      "step": 71300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6087014122580617e-05,
+      "loss": 0.2632,
+      "step": 71305
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6084168088329384e-05,
+      "loss": 0.293,
+      "step": 71310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.608132204000095e-05,
+      "loss": 0.2771,
+      "step": 71315
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6078475977632268e-05,
+      "loss": 0.2842,
+      "step": 71320
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6075629901260272e-05,
+      "loss": 0.2775,
+      "step": 71325
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6072783810921946e-05,
+      "loss": 0.2772,
+      "step": 71330
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6069937706654218e-05,
+      "loss": 0.2903,
+      "step": 71335
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.606709158849405e-05,
+      "loss": 0.274,
+      "step": 71340
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.606424545647841e-05,
+      "loss": 0.2756,
+      "step": 71345
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6061399310644246e-05,
+      "loss": 0.2881,
+      "step": 71350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.605855315102851e-05,
+      "loss": 0.2835,
+      "step": 71355
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6055706977668147e-05,
+      "loss": 0.2657,
+      "step": 71360
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6052860790600143e-05,
+      "loss": 0.2821,
+      "step": 71365
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6050014589861416e-05,
+      "loss": 0.2426,
+      "step": 71370
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6047168375488956e-05,
+      "loss": 0.2741,
+      "step": 71375
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.60443221475197e-05,
+      "loss": 0.2739,
+      "step": 71380
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6041475905990614e-05,
+      "loss": 0.2688,
+      "step": 71385
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6038629650938644e-05,
+      "loss": 0.2555,
+      "step": 71390
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.603578338240075e-05,
+      "loss": 0.2414,
+      "step": 71395
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6032937100413896e-05,
+      "loss": 0.2605,
+      "step": 71400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.603009080501504e-05,
+      "loss": 0.2634,
+      "step": 71405
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.602724449624112e-05,
+      "loss": 0.2889,
+      "step": 71410
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6024398174129122e-05,
+      "loss": 0.2832,
+      "step": 71415
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.602155183871598e-05,
+      "loss": 0.2571,
+      "step": 71420
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.601870549003866e-05,
+      "loss": 0.2695,
+      "step": 71425
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6015859128134124e-05,
+      "loss": 0.28,
+      "step": 71430
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.601301275303933e-05,
+      "loss": 0.2626,
+      "step": 71435
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6010166364791223e-05,
+      "loss": 0.2621,
+      "step": 71440
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.600731996342678e-05,
+      "loss": 0.2734,
+      "step": 71445
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6004473548982945e-05,
+      "loss": 0.2753,
+      "step": 71450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.6001627121496687e-05,
+      "loss": 0.2534,
+      "step": 71455
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.599878068100496e-05,
+      "loss": 0.289,
+      "step": 71460
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5995934227544728e-05,
+      "loss": 0.2631,
+      "step": 71465
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.599308776115294e-05,
+      "loss": 0.2656,
+      "step": 71470
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.599024128186656e-05,
+      "loss": 0.263,
+      "step": 71475
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 2.5987394789722554e-05,
+      "loss": 0.2728,
+      "step": 71480
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5984548284757877e-05,
+      "loss": 0.2724,
+      "step": 71485
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.598170176700949e-05,
+      "loss": 0.2847,
+      "step": 71490
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.597885523651435e-05,
+      "loss": 0.2776,
+      "step": 71495
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.597600869330943e-05,
+      "loss": 0.2942,
+      "step": 71500
+    },
+    {
+      "epoch": 2.52,
+      "eval_loss": 0.2644185721874237,
+      "eval_runtime": 10.558,
+      "eval_samples_per_second": 9.471,
+      "eval_steps_per_second": 9.471,
+      "step": 71500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5973162137431668e-05,
+      "loss": 0.2832,
+      "step": 71505
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.597031556891804e-05,
+      "loss": 0.2688,
+      "step": 71510
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5967468987805504e-05,
+      "loss": 0.2701,
+      "step": 71515
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5964622394131023e-05,
+      "loss": 0.2681,
+      "step": 71520
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5961775787931553e-05,
+      "loss": 0.2971,
+      "step": 71525
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5958929169244063e-05,
+      "loss": 0.312,
+      "step": 71530
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5956082538105504e-05,
+      "loss": 0.2934,
+      "step": 71535
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5953235894552847e-05,
+      "loss": 0.271,
+      "step": 71540
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5950389238623047e-05,
+      "loss": 0.2588,
+      "step": 71545
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5947542570353077e-05,
+      "loss": 0.2516,
+      "step": 71550
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.594469588977988e-05,
+      "loss": 0.2648,
+      "step": 71555
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.594184919694044e-05,
+      "loss": 0.2894,
+      "step": 71560
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.59390024918717e-05,
+      "loss": 0.2683,
+      "step": 71565
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.593615577461064e-05,
+      "loss": 0.2838,
+      "step": 71570
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5933309045194204e-05,
+      "loss": 0.2564,
+      "step": 71575
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5930462303659375e-05,
+      "loss": 0.2731,
+      "step": 71580
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.59276155500431e-05,
+      "loss": 0.2804,
+      "step": 71585
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.592476878438234e-05,
+      "loss": 0.267,
+      "step": 71590
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5921922006714085e-05,
+      "loss": 0.2696,
+      "step": 71595
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5919075217075268e-05,
+      "loss": 0.2778,
+      "step": 71600
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5916228415502875e-05,
+      "loss": 0.2754,
+      "step": 71605
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5913381602033847e-05,
+      "loss": 0.2525,
+      "step": 71610
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.591053477670517e-05,
+      "loss": 0.2565,
+      "step": 71615
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5907687939553794e-05,
+      "loss": 0.2581,
+      "step": 71620
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5904841090616695e-05,
+      "loss": 0.2581,
+      "step": 71625
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5901994229930824e-05,
+      "loss": 0.2522,
+      "step": 71630
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.589914735753316e-05,
+      "loss": 0.2773,
+      "step": 71635
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5896300473460655e-05,
+      "loss": 0.2666,
+      "step": 71640
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5893453577750277e-05,
+      "loss": 0.2703,
+      "step": 71645
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5890606670438994e-05,
+      "loss": 0.2686,
+      "step": 71650
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5887759751563777e-05,
+      "loss": 0.2582,
+      "step": 71655
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.588491282116158e-05,
+      "loss": 0.2581,
+      "step": 71660
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5882065879269373e-05,
+      "loss": 0.259,
+      "step": 71665
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5879218925924116e-05,
+      "loss": 0.2782,
+      "step": 71670
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.587637196116279e-05,
+      "loss": 0.2828,
+      "step": 71675
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5873524985022347e-05,
+      "loss": 0.2558,
+      "step": 71680
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.587067799753976e-05,
+      "loss": 0.2871,
+      "step": 71685
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.586783099875199e-05,
+      "loss": 0.2673,
+      "step": 71690
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5864983988696007e-05,
+      "loss": 0.2785,
+      "step": 71695
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5862136967408773e-05,
+      "loss": 0.2656,
+      "step": 71700
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.585928993492727e-05,
+      "loss": 0.2634,
+      "step": 71705
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5856442891288447e-05,
+      "loss": 0.2504,
+      "step": 71710
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5853595836529277e-05,
+      "loss": 0.2628,
+      "step": 71715
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5850748770686728e-05,
+      "loss": 0.255,
+      "step": 71720
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5847901693797765e-05,
+      "loss": 0.2527,
+      "step": 71725
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.584505460589936e-05,
+      "loss": 0.2515,
+      "step": 71730
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5842207507028478e-05,
+      "loss": 0.297,
+      "step": 71735
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5839360397222083e-05,
+      "loss": 0.2261,
+      "step": 71740
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.583651327651715e-05,
+      "loss": 0.2677,
+      "step": 71745
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.583366614495064e-05,
+      "loss": 0.2555,
+      "step": 71750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5830819002559526e-05,
+      "loss": 0.2722,
+      "step": 71755
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.582797184938078e-05,
+      "loss": 0.256,
+      "step": 71760
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 2.5825124685451363e-05,
+      "loss": 0.2765,
+      "step": 71765
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5822277510808246e-05,
+      "loss": 0.2664,
+      "step": 71770
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5819430325488393e-05,
+      "loss": 0.2628,
+      "step": 71775
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5816583129528776e-05,
+      "loss": 0.2742,
+      "step": 71780
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.581373592296637e-05,
+      "loss": 0.2894,
+      "step": 71785
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5810888705838136e-05,
+      "loss": 0.2772,
+      "step": 71790
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.580804147818105e-05,
+      "loss": 0.2661,
+      "step": 71795
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5805194240032083e-05,
+      "loss": 0.2905,
+      "step": 71800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.580234699142819e-05,
+      "loss": 0.2824,
+      "step": 71805
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.579949973240635e-05,
+      "loss": 0.2648,
+      "step": 71810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5796652463003535e-05,
+      "loss": 0.2654,
+      "step": 71815
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.579380518325672e-05,
+      "loss": 0.2823,
+      "step": 71820
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5790957893202856e-05,
+      "loss": 0.2606,
+      "step": 71825
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5788110592878928e-05,
+      "loss": 0.2698,
+      "step": 71830
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.578526328232191e-05,
+      "loss": 0.273,
+      "step": 71835
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5782415961568756e-05,
+      "loss": 0.2733,
+      "step": 71840
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5779568630656452e-05,
+      "loss": 0.2512,
+      "step": 71845
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5776721289621965e-05,
+      "loss": 0.264,
+      "step": 71850
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5773873938502263e-05,
+      "loss": 0.2789,
+      "step": 71855
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.577102657733432e-05,
+      "loss": 0.2529,
+      "step": 71860
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5768179206155098e-05,
+      "loss": 0.2685,
+      "step": 71865
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5765331825001577e-05,
+      "loss": 0.2743,
+      "step": 71870
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5762484433910732e-05,
+      "loss": 0.2568,
+      "step": 71875
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5759637032919527e-05,
+      "loss": 0.2392,
+      "step": 71880
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5756789622064937e-05,
+      "loss": 0.2603,
+      "step": 71885
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5753942201383925e-05,
+      "loss": 0.2906,
+      "step": 71890
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5751094770913476e-05,
+      "loss": 0.2739,
+      "step": 71895
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5748247330690557e-05,
+      "loss": 0.2532,
+      "step": 71900
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5745399880752143e-05,
+      "loss": 0.2563,
+      "step": 71905
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5742552421135195e-05,
+      "loss": 0.2803,
+      "step": 71910
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.573970495187669e-05,
+      "loss": 0.2673,
+      "step": 71915
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5736857473013616e-05,
+      "loss": 0.2969,
+      "step": 71920
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.573400998458293e-05,
+      "loss": 0.2601,
+      "step": 71925
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5731162486621602e-05,
+      "loss": 0.2824,
+      "step": 71930
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5728314979166618e-05,
+      "loss": 0.2398,
+      "step": 71935
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.572546746225494e-05,
+      "loss": 0.253,
+      "step": 71940
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.572261993592354e-05,
+      "loss": 0.266,
+      "step": 71945
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.57197724002094e-05,
+      "loss": 0.2691,
+      "step": 71950
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5716924855149495e-05,
+      "loss": 0.2848,
+      "step": 71955
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5714077300780792e-05,
+      "loss": 0.2687,
+      "step": 71960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5711229737140263e-05,
+      "loss": 0.2578,
+      "step": 71965
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5708382164264883e-05,
+      "loss": 0.247,
+      "step": 71970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5705534582191626e-05,
+      "loss": 0.2634,
+      "step": 71975
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5702686990957474e-05,
+      "loss": 0.2676,
+      "step": 71980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.569983939059939e-05,
+      "loss": 0.2804,
+      "step": 71985
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.569699178115436e-05,
+      "loss": 0.2901,
+      "step": 71990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5694144162659344e-05,
+      "loss": 0.2714,
+      "step": 71995
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.569129653515132e-05,
+      "loss": 0.2763,
+      "step": 72000
+    },
+    {
+      "epoch": 2.53,
+      "eval_loss": 0.26415836811065674,
+      "eval_runtime": 10.5519,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 72000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5688448898667274e-05,
+      "loss": 0.2591,
+      "step": 72005
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5685601253244172e-05,
+      "loss": 0.2776,
+      "step": 72010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5682753598918985e-05,
+      "loss": 0.2975,
+      "step": 72015
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.56799059357287e-05,
+      "loss": 0.2368,
+      "step": 72020
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5677058263710274e-05,
+      "loss": 0.2852,
+      "step": 72025
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5674210582900698e-05,
+      "loss": 0.3003,
+      "step": 72030
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.567136289333694e-05,
+      "loss": 0.2511,
+      "step": 72035
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5668515195055988e-05,
+      "loss": 0.2747,
+      "step": 72040
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.566566748809479e-05,
+      "loss": 0.2807,
+      "step": 72045
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 2.5662819772490348e-05,
+      "loss": 0.272,
+      "step": 72050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5659972048279623e-05,
+      "loss": 0.2606,
+      "step": 72055
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5657124315499598e-05,
+      "loss": 0.2775,
+      "step": 72060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5654276574187254e-05,
+      "loss": 0.2746,
+      "step": 72065
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.565142882437955e-05,
+      "loss": 0.2642,
+      "step": 72070
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.564858106611348e-05,
+      "loss": 0.2627,
+      "step": 72075
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5645733299426006e-05,
+      "loss": 0.2554,
+      "step": 72080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.564288552435411e-05,
+      "loss": 0.2609,
+      "step": 72085
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.564003774093477e-05,
+      "loss": 0.2533,
+      "step": 72090
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.563718994920497e-05,
+      "loss": 0.2824,
+      "step": 72095
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5634342149201667e-05,
+      "loss": 0.2591,
+      "step": 72100
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5631494340961858e-05,
+      "loss": 0.2786,
+      "step": 72105
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5628646524522503e-05,
+      "loss": 0.2509,
+      "step": 72110
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5625798699920596e-05,
+      "loss": 0.2841,
+      "step": 72115
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5622950867193095e-05,
+      "loss": 0.2978,
+      "step": 72120
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5620103026377003e-05,
+      "loss": 0.2747,
+      "step": 72125
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.561725517750927e-05,
+      "loss": 0.2687,
+      "step": 72130
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5614407320626886e-05,
+      "loss": 0.2913,
+      "step": 72135
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5611559455766825e-05,
+      "loss": 0.2905,
+      "step": 72140
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5608711582966077e-05,
+      "loss": 0.2821,
+      "step": 72145
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5605863702261607e-05,
+      "loss": 0.2703,
+      "step": 72150
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5603015813690396e-05,
+      "loss": 0.2875,
+      "step": 72155
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.560016791728942e-05,
+      "loss": 0.2575,
+      "step": 72160
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5597320013095656e-05,
+      "loss": 0.2838,
+      "step": 72165
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.559447210114609e-05,
+      "loss": 0.2722,
+      "step": 72170
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5591624181477693e-05,
+      "loss": 0.2918,
+      "step": 72175
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.558877625412745e-05,
+      "loss": 0.2727,
+      "step": 72180
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.558592831913233e-05,
+      "loss": 0.2597,
+      "step": 72185
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5583080376529318e-05,
+      "loss": 0.2616,
+      "step": 72190
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.558023242635539e-05,
+      "loss": 0.2793,
+      "step": 72195
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5577384468647537e-05,
+      "loss": 0.2957,
+      "step": 72200
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5574536503442715e-05,
+      "loss": 0.2607,
+      "step": 72205
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5571688530777928e-05,
+      "loss": 0.2442,
+      "step": 72210
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5568840550690127e-05,
+      "loss": 0.2868,
+      "step": 72215
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5565992563216313e-05,
+      "loss": 0.2637,
+      "step": 72220
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5563144568393456e-05,
+      "loss": 0.2579,
+      "step": 72225
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5560296566258546e-05,
+      "loss": 0.2757,
+      "step": 72230
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5557448556848545e-05,
+      "loss": 0.2823,
+      "step": 72235
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.555460054020045e-05,
+      "loss": 0.3012,
+      "step": 72240
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.555175251635123e-05,
+      "loss": 0.2845,
+      "step": 72245
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5548904485337864e-05,
+      "loss": 0.2942,
+      "step": 72250
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5546056447197335e-05,
+      "loss": 0.2608,
+      "step": 72255
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.554320840196663e-05,
+      "loss": 0.2834,
+      "step": 72260
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5540360349682713e-05,
+      "loss": 0.2809,
+      "step": 72265
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5537512290382576e-05,
+      "loss": 0.2622,
+      "step": 72270
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5534664224103193e-05,
+      "loss": 0.2804,
+      "step": 72275
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5531816150881554e-05,
+      "loss": 0.2637,
+      "step": 72280
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.552896807075463e-05,
+      "loss": 0.2903,
+      "step": 72285
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5526119983759404e-05,
+      "loss": 0.2552,
+      "step": 72290
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.552327188993286e-05,
+      "loss": 0.2634,
+      "step": 72295
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5520423789311972e-05,
+      "loss": 0.2673,
+      "step": 72300
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.551757568193372e-05,
+      "loss": 0.2947,
+      "step": 72305
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.551472756783509e-05,
+      "loss": 0.258,
+      "step": 72310
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5511879447053068e-05,
+      "loss": 0.2688,
+      "step": 72315
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5509031319624627e-05,
+      "loss": 0.2813,
+      "step": 72320
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5506183185586752e-05,
+      "loss": 0.2619,
+      "step": 72325
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5503335044976405e-05,
+      "loss": 0.2896,
+      "step": 72330
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 2.5500486897830607e-05,
+      "loss": 0.2547,
+      "step": 72335
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5497638744186304e-05,
+      "loss": 0.2922,
+      "step": 72340
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5494790584080498e-05,
+      "loss": 0.2762,
+      "step": 72345
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5491942417550154e-05,
+      "loss": 0.2727,
+      "step": 72350
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.548909424463226e-05,
+      "loss": 0.2848,
+      "step": 72355
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5486246065363807e-05,
+      "loss": 0.2811,
+      "step": 72360
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5483397879781768e-05,
+      "loss": 0.2931,
+      "step": 72365
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5480549687923123e-05,
+      "loss": 0.2764,
+      "step": 72370
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5477701489824863e-05,
+      "loss": 0.2898,
+      "step": 72375
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5474853285523958e-05,
+      "loss": 0.2849,
+      "step": 72380
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.54720050750574e-05,
+      "loss": 0.267,
+      "step": 72385
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5469156858462156e-05,
+      "loss": 0.2518,
+      "step": 72390
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5466308635775238e-05,
+      "loss": 0.2578,
+      "step": 72395
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5463460407033597e-05,
+      "loss": 0.2703,
+      "step": 72400
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5460612172274234e-05,
+      "loss": 0.2511,
+      "step": 72405
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5457763931534117e-05,
+      "loss": 0.2541,
+      "step": 72410
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5454915684850238e-05,
+      "loss": 0.2788,
+      "step": 72415
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5452067432259585e-05,
+      "loss": 0.2648,
+      "step": 72420
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5449219173799132e-05,
+      "loss": 0.2611,
+      "step": 72425
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5446370909505868e-05,
+      "loss": 0.2631,
+      "step": 72430
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5443522639416767e-05,
+      "loss": 0.2589,
+      "step": 72435
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.544067436356882e-05,
+      "loss": 0.2697,
+      "step": 72440
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5437826081999e-05,
+      "loss": 0.2398,
+      "step": 72445
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.543497779474431e-05,
+      "loss": 0.2626,
+      "step": 72450
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5432129501841704e-05,
+      "loss": 0.2503,
+      "step": 72455
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5429281203328198e-05,
+      "loss": 0.2749,
+      "step": 72460
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.542643289924075e-05,
+      "loss": 0.2486,
+      "step": 72465
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5423584589616346e-05,
+      "loss": 0.2838,
+      "step": 72470
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.542073627449198e-05,
+      "loss": 0.2734,
+      "step": 72475
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5417887953904633e-05,
+      "loss": 0.295,
+      "step": 72480
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5415039627891287e-05,
+      "loss": 0.2504,
+      "step": 72485
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.541219129648892e-05,
+      "loss": 0.2823,
+      "step": 72490
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5409342959734522e-05,
+      "loss": 0.2523,
+      "step": 72495
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5406494617665084e-05,
+      "loss": 0.2634,
+      "step": 72500
+    },
+    {
+      "epoch": 2.55,
+      "eval_loss": 0.26341068744659424,
+      "eval_runtime": 10.5346,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 72500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5403646270317573e-05,
+      "loss": 0.2763,
+      "step": 72505
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.540079791772898e-05,
+      "loss": 0.2665,
+      "step": 72510
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5397949559936297e-05,
+      "loss": 0.2826,
+      "step": 72515
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5395101196976488e-05,
+      "loss": 0.2844,
+      "step": 72520
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.539225282888656e-05,
+      "loss": 0.2834,
+      "step": 72525
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5389404455703487e-05,
+      "loss": 0.2522,
+      "step": 72530
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5386556077464258e-05,
+      "loss": 0.2763,
+      "step": 72535
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5383707694205843e-05,
+      "loss": 0.2537,
+      "step": 72540
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.538085930596525e-05,
+      "loss": 0.2708,
+      "step": 72545
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5378010912779432e-05,
+      "loss": 0.2655,
+      "step": 72550
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5375162514685408e-05,
+      "loss": 0.309,
+      "step": 72555
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5372314111720135e-05,
+      "loss": 0.2705,
+      "step": 72560
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.536946570392062e-05,
+      "loss": 0.2456,
+      "step": 72565
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5366617291323825e-05,
+      "loss": 0.2688,
+      "step": 72570
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5363768873966747e-05,
+      "loss": 0.2542,
+      "step": 72575
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5360920451886373e-05,
+      "loss": 0.2535,
+      "step": 72580
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5358072025119687e-05,
+      "loss": 0.2642,
+      "step": 72585
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.535522359370367e-05,
+      "loss": 0.272,
+      "step": 72590
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.535237515767531e-05,
+      "loss": 0.2651,
+      "step": 72595
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.534952671707159e-05,
+      "loss": 0.2561,
+      "step": 72600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.534667827192949e-05,
+      "loss": 0.2688,
+      "step": 72605
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5343829822286008e-05,
+      "loss": 0.2615,
+      "step": 72610
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.5340981368178125e-05,
+      "loss": 0.2712,
+      "step": 72615
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 2.533813290964282e-05,
+      "loss": 0.3022,
+      "step": 72620
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5335284446717083e-05,
+      "loss": 0.2799,
+      "step": 72625
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.533243597943789e-05,
+      "loss": 0.2795,
+      "step": 72630
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5329587507842246e-05,
+      "loss": 0.2702,
+      "step": 72635
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5326739031967118e-05,
+      "loss": 0.2808,
+      "step": 72640
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.53238905518495e-05,
+      "loss": 0.2706,
+      "step": 72645
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5321042067526383e-05,
+      "loss": 0.2738,
+      "step": 72650
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5318193579034737e-05,
+      "loss": 0.27,
+      "step": 72655
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.531534508641156e-05,
+      "loss": 0.2633,
+      "step": 72660
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.531249658969384e-05,
+      "loss": 0.2696,
+      "step": 72665
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5309648088918558e-05,
+      "loss": 0.2552,
+      "step": 72670
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5306799584122693e-05,
+      "loss": 0.2602,
+      "step": 72675
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5303951075343245e-05,
+      "loss": 0.2569,
+      "step": 72680
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5301102562617186e-05,
+      "loss": 0.2762,
+      "step": 72685
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.529825404598151e-05,
+      "loss": 0.2722,
+      "step": 72690
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5295405525473205e-05,
+      "loss": 0.2748,
+      "step": 72695
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5292557001129252e-05,
+      "loss": 0.2611,
+      "step": 72700
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5289708472986638e-05,
+      "loss": 0.2969,
+      "step": 72705
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5286859941082347e-05,
+      "loss": 0.2757,
+      "step": 72710
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.528401140545337e-05,
+      "loss": 0.259,
+      "step": 72715
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5281162866136705e-05,
+      "loss": 0.2553,
+      "step": 72720
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.527831432316931e-05,
+      "loss": 0.252,
+      "step": 72725
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.52754657765882e-05,
+      "loss": 0.2745,
+      "step": 72730
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.527261722643034e-05,
+      "loss": 0.2559,
+      "step": 72735
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5269768672732726e-05,
+      "loss": 0.2649,
+      "step": 72740
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5266920115532338e-05,
+      "loss": 0.262,
+      "step": 72745
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5264071554866176e-05,
+      "loss": 0.2882,
+      "step": 72750
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5261222990771218e-05,
+      "loss": 0.2757,
+      "step": 72755
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.525837442328445e-05,
+      "loss": 0.2497,
+      "step": 72760
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5255525852442863e-05,
+      "loss": 0.2631,
+      "step": 72765
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5252677278283432e-05,
+      "loss": 0.2461,
+      "step": 72770
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5249828700843165e-05,
+      "loss": 0.2635,
+      "step": 72775
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5246980120159026e-05,
+      "loss": 0.2524,
+      "step": 72780
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5244131536268024e-05,
+      "loss": 0.2744,
+      "step": 72785
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.524128294920713e-05,
+      "loss": 0.2592,
+      "step": 72790
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.523843435901333e-05,
+      "loss": 0.2661,
+      "step": 72795
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5235585765723624e-05,
+      "loss": 0.2602,
+      "step": 72800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5232737169374993e-05,
+      "loss": 0.2876,
+      "step": 72805
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5229888570004416e-05,
+      "loss": 0.2709,
+      "step": 72810
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.52270399676489e-05,
+      "loss": 0.2644,
+      "step": 72815
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.522419136234541e-05,
+      "loss": 0.2693,
+      "step": 72820
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5221342754130943e-05,
+      "loss": 0.2959,
+      "step": 72825
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.521849414304249e-05,
+      "loss": 0.2704,
+      "step": 72830
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5215645529117033e-05,
+      "loss": 0.2734,
+      "step": 72835
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.521279691239156e-05,
+      "loss": 0.2744,
+      "step": 72840
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5209948292903063e-05,
+      "loss": 0.254,
+      "step": 72845
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5207099670688516e-05,
+      "loss": 0.2688,
+      "step": 72850
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5204251045784927e-05,
+      "loss": 0.2712,
+      "step": 72855
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5201402418229265e-05,
+      "loss": 0.2834,
+      "step": 72860
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5198553788058533e-05,
+      "loss": 0.2676,
+      "step": 72865
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5195705155309705e-05,
+      "loss": 0.2897,
+      "step": 72870
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5192856520019774e-05,
+      "loss": 0.2422,
+      "step": 72875
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.519000788222573e-05,
+      "loss": 0.2494,
+      "step": 72880
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.518715924196456e-05,
+      "loss": 0.2354,
+      "step": 72885
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.518431059927325e-05,
+      "loss": 0.2622,
+      "step": 72890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.518146195418879e-05,
+      "loss": 0.2665,
+      "step": 72895
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 2.5178613306748166e-05,
+      "loss": 0.2761,
+      "step": 72900
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5175764656988365e-05,
+      "loss": 0.2379,
+      "step": 72905
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5172916004946372e-05,
+      "loss": 0.2582,
+      "step": 72910
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5170067350659183e-05,
+      "loss": 0.2882,
+      "step": 72915
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5167218694163786e-05,
+      "loss": 0.28,
+      "step": 72920
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5164370035497152e-05,
+      "loss": 0.2516,
+      "step": 72925
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.516152137469629e-05,
+      "loss": 0.2771,
+      "step": 72930
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5158672711798174e-05,
+      "loss": 0.2699,
+      "step": 72935
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5155824046839803e-05,
+      "loss": 0.2599,
+      "step": 72940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5152975379858155e-05,
+      "loss": 0.2754,
+      "step": 72945
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5150126710890232e-05,
+      "loss": 0.2775,
+      "step": 72950
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5147278039973005e-05,
+      "loss": 0.2599,
+      "step": 72955
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.514442936714347e-05,
+      "loss": 0.2474,
+      "step": 72960
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.514158069243861e-05,
+      "loss": 0.2508,
+      "step": 72965
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.513873201589543e-05,
+      "loss": 0.2771,
+      "step": 72970
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5135883337550898e-05,
+      "loss": 0.2758,
+      "step": 72975
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.513303465744201e-05,
+      "loss": 0.2695,
+      "step": 72980
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5130185975605756e-05,
+      "loss": 0.2611,
+      "step": 72985
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5127337292079122e-05,
+      "loss": 0.2607,
+      "step": 72990
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.51244886068991e-05,
+      "loss": 0.2707,
+      "step": 72995
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5121639920102675e-05,
+      "loss": 0.2887,
+      "step": 73000
+    },
+    {
+      "epoch": 2.57,
+      "eval_loss": 0.2631213068962097,
+      "eval_runtime": 10.5493,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 73000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5118791231726835e-05,
+      "loss": 0.2693,
+      "step": 73005
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5115942541808568e-05,
+      "loss": 0.2585,
+      "step": 73010
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5113093850384863e-05,
+      "loss": 0.2537,
+      "step": 73015
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5110245157492707e-05,
+      "loss": 0.2874,
+      "step": 73020
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5107396463169097e-05,
+      "loss": 0.2721,
+      "step": 73025
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.510454776745101e-05,
+      "loss": 0.2768,
+      "step": 73030
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5101699070375446e-05,
+      "loss": 0.2624,
+      "step": 73035
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5098850371979375e-05,
+      "loss": 0.2713,
+      "step": 73040
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5096001672299806e-05,
+      "loss": 0.2588,
+      "step": 73045
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5093152971373712e-05,
+      "loss": 0.2834,
+      "step": 73050
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5090304269238096e-05,
+      "loss": 0.25,
+      "step": 73055
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5087455565929936e-05,
+      "loss": 0.2558,
+      "step": 73060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.508460686148622e-05,
+      "loss": 0.2671,
+      "step": 73065
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5081758155943935e-05,
+      "loss": 0.2846,
+      "step": 73070
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.507890944934009e-05,
+      "loss": 0.2752,
+      "step": 73075
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5076060741711643e-05,
+      "loss": 0.2523,
+      "step": 73080
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5073212033095605e-05,
+      "loss": 0.2534,
+      "step": 73085
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5070363323528956e-05,
+      "loss": 0.2498,
+      "step": 73090
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5067514613048683e-05,
+      "loss": 0.271,
+      "step": 73095
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5064665901691775e-05,
+      "loss": 0.2751,
+      "step": 73100
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.506181718949523e-05,
+      "loss": 0.2606,
+      "step": 73105
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5058968476496024e-05,
+      "loss": 0.2337,
+      "step": 73110
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.505611976273115e-05,
+      "loss": 0.2835,
+      "step": 73115
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.50532710482376e-05,
+      "loss": 0.2665,
+      "step": 73120
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5050422333052366e-05,
+      "loss": 0.2758,
+      "step": 73125
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.504757361721242e-05,
+      "loss": 0.2706,
+      "step": 73130
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5044724900754775e-05,
+      "loss": 0.2624,
+      "step": 73135
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.50418761837164e-05,
+      "loss": 0.2817,
+      "step": 73140
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5039027466134292e-05,
+      "loss": 0.2912,
+      "step": 73145
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.503617874804543e-05,
+      "loss": 0.2722,
+      "step": 73150
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5033330029486818e-05,
+      "loss": 0.3058,
+      "step": 73155
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.503048131049544e-05,
+      "loss": 0.2577,
+      "step": 73160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.502763259110828e-05,
+      "loss": 0.2728,
+      "step": 73165
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5024783871362327e-05,
+      "loss": 0.2937,
+      "step": 73170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5021935151294573e-05,
+      "loss": 0.2686,
+      "step": 73175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.5019086430942006e-05,
+      "loss": 0.2744,
+      "step": 73180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 2.501623771034161e-05,
+      "loss": 0.2831,
+      "step": 73185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5013388989530388e-05,
+      "loss": 0.2692,
+      "step": 73190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5010540268545314e-05,
+      "loss": 0.2712,
+      "step": 73195
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5007691547423377e-05,
+      "loss": 0.2667,
+      "step": 73200
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5004842826201574e-05,
+      "loss": 0.2671,
+      "step": 73205
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.5001994104916888e-05,
+      "loss": 0.2742,
+      "step": 73210
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4999145383606313e-05,
+      "loss": 0.2682,
+      "step": 73215
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4996296662306834e-05,
+      "loss": 0.2402,
+      "step": 73220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4993447941055446e-05,
+      "loss": 0.2721,
+      "step": 73225
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4990599219889114e-05,
+      "loss": 0.2458,
+      "step": 73230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.498775049884486e-05,
+      "loss": 0.2727,
+      "step": 73235
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4984901777959655e-05,
+      "loss": 0.2571,
+      "step": 73240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4982053057270483e-05,
+      "loss": 0.2896,
+      "step": 73245
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.497920433681435e-05,
+      "loss": 0.276,
+      "step": 73250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4976355616628236e-05,
+      "loss": 0.266,
+      "step": 73255
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4973506896749123e-05,
+      "loss": 0.291,
+      "step": 73260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4970658177214007e-05,
+      "loss": 0.2903,
+      "step": 73265
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4967809458059873e-05,
+      "loss": 0.2851,
+      "step": 73270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.496496073932372e-05,
+      "loss": 0.2526,
+      "step": 73275
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4962112021042523e-05,
+      "loss": 0.2813,
+      "step": 73280
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.495926330325327e-05,
+      "loss": 0.2818,
+      "step": 73285
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4956414585992963e-05,
+      "loss": 0.2748,
+      "step": 73290
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4953565869298586e-05,
+      "loss": 0.262,
+      "step": 73295
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.495071715320713e-05,
+      "loss": 0.2837,
+      "step": 73300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4947868437755563e-05,
+      "loss": 0.2752,
+      "step": 73305
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4945019722980904e-05,
+      "loss": 0.2715,
+      "step": 73310
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.494217100892012e-05,
+      "loss": 0.273,
+      "step": 73315
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4939322295610216e-05,
+      "loss": 0.2473,
+      "step": 73320
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.493647358308816e-05,
+      "loss": 0.2795,
+      "step": 73325
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4933624871390962e-05,
+      "loss": 0.2874,
+      "step": 73330
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4930776160555603e-05,
+      "loss": 0.2605,
+      "step": 73335
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.492792745061906e-05,
+      "loss": 0.2756,
+      "step": 73340
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4925078741618338e-05,
+      "loss": 0.2642,
+      "step": 73345
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4922230033590415e-05,
+      "loss": 0.2525,
+      "step": 73350
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.491938132657229e-05,
+      "loss": 0.2883,
+      "step": 73355
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4916532620600937e-05,
+      "loss": 0.2925,
+      "step": 73360
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4913683915713362e-05,
+      "loss": 0.2917,
+      "step": 73365
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.491083521194654e-05,
+      "loss": 0.2734,
+      "step": 73370
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4907986509337467e-05,
+      "loss": 0.2758,
+      "step": 73375
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4905137807923116e-05,
+      "loss": 0.2821,
+      "step": 73380
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4902289107740508e-05,
+      "loss": 0.2439,
+      "step": 73385
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4899440408826602e-05,
+      "loss": 0.2789,
+      "step": 73390
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4896591711218394e-05,
+      "loss": 0.2544,
+      "step": 73395
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4893743014952875e-05,
+      "loss": 0.2764,
+      "step": 73400
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.489089432006703e-05,
+      "loss": 0.2583,
+      "step": 73405
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.488804562659786e-05,
+      "loss": 0.2913,
+      "step": 73410
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4885196934582343e-05,
+      "loss": 0.2417,
+      "step": 73415
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.488234824405746e-05,
+      "loss": 0.28,
+      "step": 73420
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.487949955506021e-05,
+      "loss": 0.2668,
+      "step": 73425
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.487665086762758e-05,
+      "loss": 0.2542,
+      "step": 73430
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.487380218179656e-05,
+      "loss": 0.2687,
+      "step": 73435
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4870953497604126e-05,
+      "loss": 0.2709,
+      "step": 73440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4868104815087284e-05,
+      "loss": 0.2685,
+      "step": 73445
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.486525613428301e-05,
+      "loss": 0.2622,
+      "step": 73450
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4862407455228303e-05,
+      "loss": 0.291,
+      "step": 73455
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4859558777960135e-05,
+      "loss": 0.265,
+      "step": 73460
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4856710102515513e-05,
+      "loss": 0.2648,
+      "step": 73465
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 2.4853861428931412e-05,
+      "loss": 0.3027,
+      "step": 73470
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4851012757244818e-05,
+      "loss": 0.3025,
+      "step": 73475
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4848164087492735e-05,
+      "loss": 0.2561,
+      "step": 73480
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4845315419712135e-05,
+      "loss": 0.2634,
+      "step": 73485
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.484246675394002e-05,
+      "loss": 0.2709,
+      "step": 73490
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4839618090213357e-05,
+      "loss": 0.2787,
+      "step": 73495
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4836769428569158e-05,
+      "loss": 0.2796,
+      "step": 73500
+    },
+    {
+      "epoch": 2.59,
+      "eval_loss": 0.2626335024833679,
+      "eval_runtime": 10.5465,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 73500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4833920769044402e-05,
+      "loss": 0.2616,
+      "step": 73505
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.483107211167607e-05,
+      "loss": 0.2657,
+      "step": 73510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.482822345650115e-05,
+      "loss": 0.245,
+      "step": 73515
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4825374803556643e-05,
+      "loss": 0.2641,
+      "step": 73520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4822526152879533e-05,
+      "loss": 0.2623,
+      "step": 73525
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4819677504506795e-05,
+      "loss": 0.2704,
+      "step": 73530
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.481682885847543e-05,
+      "loss": 0.2854,
+      "step": 73535
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4813980214822418e-05,
+      "loss": 0.2653,
+      "step": 73540
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4811131573584757e-05,
+      "loss": 0.2757,
+      "step": 73545
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.480828293479943e-05,
+      "loss": 0.2732,
+      "step": 73550
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4805434298503407e-05,
+      "loss": 0.2669,
+      "step": 73555
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4802585664733708e-05,
+      "loss": 0.2607,
+      "step": 73560
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4799737033527296e-05,
+      "loss": 0.2629,
+      "step": 73565
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4797458130432597e-05,
+      "loss": 0.2736,
+      "step": 73570
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4794609503933324e-05,
+      "loss": 0.2633,
+      "step": 73575
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4791760880100916e-05,
+      "loss": 0.2567,
+      "step": 73580
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4788912258972347e-05,
+      "loss": 0.2552,
+      "step": 73585
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4786063640584623e-05,
+      "loss": 0.2663,
+      "step": 73590
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4783215024974717e-05,
+      "loss": 0.2864,
+      "step": 73595
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4780366412179627e-05,
+      "loss": 0.2785,
+      "step": 73600
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4777517802236323e-05,
+      "loss": 0.2649,
+      "step": 73605
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4774669195181813e-05,
+      "loss": 0.2524,
+      "step": 73610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4771820591053075e-05,
+      "loss": 0.2715,
+      "step": 73615
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.476897198988709e-05,
+      "loss": 0.2782,
+      "step": 73620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4766123391720855e-05,
+      "loss": 0.2799,
+      "step": 73625
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4763274796591352e-05,
+      "loss": 0.2584,
+      "step": 73630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4760426204535575e-05,
+      "loss": 0.2706,
+      "step": 73635
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4757577615590507e-05,
+      "loss": 0.2862,
+      "step": 73640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.475472902979312e-05,
+      "loss": 0.2709,
+      "step": 73645
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4751880447180427e-05,
+      "loss": 0.2287,
+      "step": 73650
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.47490318677894e-05,
+      "loss": 0.2704,
+      "step": 73655
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4746183291657022e-05,
+      "loss": 0.2665,
+      "step": 73660
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4743334718820298e-05,
+      "loss": 0.2821,
+      "step": 73665
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4740486149316204e-05,
+      "loss": 0.2545,
+      "step": 73670
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4737637583181716e-05,
+      "loss": 0.2813,
+      "step": 73675
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4734789020453835e-05,
+      "loss": 0.2617,
+      "step": 73680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4731940461169543e-05,
+      "loss": 0.2424,
+      "step": 73685
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4729091905365835e-05,
+      "loss": 0.277,
+      "step": 73690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4726243353079685e-05,
+      "loss": 0.2643,
+      "step": 73695
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.472339480434808e-05,
+      "loss": 0.273,
+      "step": 73700
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.472054625920802e-05,
+      "loss": 0.2446,
+      "step": 73705
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4717697717696475e-05,
+      "loss": 0.2484,
+      "step": 73710
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4714849179850448e-05,
+      "loss": 0.2785,
+      "step": 73715
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.47120006457069e-05,
+      "loss": 0.2813,
+      "step": 73720
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.470915211530285e-05,
+      "loss": 0.2686,
+      "step": 73725
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4706303588675267e-05,
+      "loss": 0.2724,
+      "step": 73730
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4703455065861133e-05,
+      "loss": 0.268,
+      "step": 73735
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.4700606546897432e-05,
+      "loss": 0.2808,
+      "step": 73740
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.469775803182117e-05,
+      "loss": 0.2872,
+      "step": 73745
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.469490952066932e-05,
+      "loss": 0.2504,
+      "step": 73750
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 2.469206101347887e-05,
+      "loss": 0.2543,
+      "step": 73755
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4689212510286793e-05,
+      "loss": 0.2908,
+      "step": 73760
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4686364011130094e-05,
+      "loss": 0.2758,
+      "step": 73765
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4683515516045756e-05,
+      "loss": 0.2654,
+      "step": 73770
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.468066702507075e-05,
+      "loss": 0.2751,
+      "step": 73775
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.467781853824208e-05,
+      "loss": 0.2788,
+      "step": 73780
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4674970055596727e-05,
+      "loss": 0.2844,
+      "step": 73785
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4672121577171667e-05,
+      "loss": 0.2669,
+      "step": 73790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4669273103003886e-05,
+      "loss": 0.2827,
+      "step": 73795
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4666424633130393e-05,
+      "loss": 0.2853,
+      "step": 73800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.466357616758815e-05,
+      "loss": 0.2893,
+      "step": 73805
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4660727706414148e-05,
+      "loss": 0.2785,
+      "step": 73810
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4657879249645366e-05,
+      "loss": 0.2621,
+      "step": 73815
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4655030797318797e-05,
+      "loss": 0.2654,
+      "step": 73820
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4652182349471437e-05,
+      "loss": 0.2726,
+      "step": 73825
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4649333906140256e-05,
+      "loss": 0.2621,
+      "step": 73830
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4646485467362234e-05,
+      "loss": 0.287,
+      "step": 73835
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4643637033174378e-05,
+      "loss": 0.2776,
+      "step": 73840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4640788603613652e-05,
+      "loss": 0.2683,
+      "step": 73845
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4637940178717053e-05,
+      "loss": 0.2783,
+      "step": 73850
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4635091758521554e-05,
+      "loss": 0.2617,
+      "step": 73855
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4632243343064158e-05,
+      "loss": 0.2614,
+      "step": 73860
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4629394932381842e-05,
+      "loss": 0.2846,
+      "step": 73865
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.462654652651158e-05,
+      "loss": 0.2668,
+      "step": 73870
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4623698125490364e-05,
+      "loss": 0.2688,
+      "step": 73875
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4620849729355186e-05,
+      "loss": 0.267,
+      "step": 73880
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4618001338143027e-05,
+      "loss": 0.2587,
+      "step": 73885
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.461515295189086e-05,
+      "loss": 0.2811,
+      "step": 73890
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4612304570635684e-05,
+      "loss": 0.2919,
+      "step": 73895
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4609456194414473e-05,
+      "loss": 0.2534,
+      "step": 73900
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4606607823264225e-05,
+      "loss": 0.2737,
+      "step": 73905
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4603759457221904e-05,
+      "loss": 0.2857,
+      "step": 73910
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4600911096324513e-05,
+      "loss": 0.2629,
+      "step": 73915
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4598062740609028e-05,
+      "loss": 0.2631,
+      "step": 73920
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.459521439011243e-05,
+      "loss": 0.2634,
+      "step": 73925
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4592366044871708e-05,
+      "loss": 0.2748,
+      "step": 73930
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4589517704923843e-05,
+      "loss": 0.2676,
+      "step": 73935
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4586669370305825e-05,
+      "loss": 0.2539,
+      "step": 73940
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4583821041054634e-05,
+      "loss": 0.2718,
+      "step": 73945
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.458097271720724e-05,
+      "loss": 0.2683,
+      "step": 73950
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4578124398800645e-05,
+      "loss": 0.2919,
+      "step": 73955
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.457527608587183e-05,
+      "loss": 0.282,
+      "step": 73960
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.457242777845778e-05,
+      "loss": 0.2822,
+      "step": 73965
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.456957947659546e-05,
+      "loss": 0.2758,
+      "step": 73970
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4566731180321877e-05,
+      "loss": 0.2454,
+      "step": 73975
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4563882889674e-05,
+      "loss": 0.2699,
+      "step": 73980
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4561034604688822e-05,
+      "loss": 0.2696,
+      "step": 73985
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4558186325403305e-05,
+      "loss": 0.276,
+      "step": 73990
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4555338051854465e-05,
+      "loss": 0.2672,
+      "step": 73995
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4552489784079262e-05,
+      "loss": 0.286,
+      "step": 74000
+    },
+    {
+      "epoch": 2.6,
+      "eval_loss": 0.26184189319610596,
+      "eval_runtime": 10.5446,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 74000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4549641522114678e-05,
+      "loss": 0.2697,
+      "step": 74005
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4546793265997708e-05,
+      "loss": 0.2528,
+      "step": 74010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4543945015765327e-05,
+      "loss": 0.2671,
+      "step": 74015
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4541096771454525e-05,
+      "loss": 0.2972,
+      "step": 74020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4538248533102265e-05,
+      "loss": 0.2878,
+      "step": 74025
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4535400300745555e-05,
+      "loss": 0.282,
+      "step": 74030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.4532552074421363e-05,
+      "loss": 0.2478,
+      "step": 74035
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 2.452970385416668e-05,
+      "loss": 0.2523,
+      "step": 74040
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.452685564001847e-05,
+      "loss": 0.2701,
+      "step": 74045
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.452400743201374e-05,
+      "loss": 0.2813,
+      "step": 74050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4521159230189456e-05,
+      "loss": 0.2408,
+      "step": 74055
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.45183110345826e-05,
+      "loss": 0.2699,
+      "step": 74060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4515462845230158e-05,
+      "loss": 0.2767,
+      "step": 74065
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4512614662169116e-05,
+      "loss": 0.2737,
+      "step": 74070
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.450976648543645e-05,
+      "loss": 0.2806,
+      "step": 74075
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4506918315069148e-05,
+      "loss": 0.2699,
+      "step": 74080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4504070151104172e-05,
+      "loss": 0.262,
+      "step": 74085
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4501221993578533e-05,
+      "loss": 0.2613,
+      "step": 74090
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4498373842529187e-05,
+      "loss": 0.2739,
+      "step": 74095
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4495525697993134e-05,
+      "loss": 0.2618,
+      "step": 74100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.449267756000734e-05,
+      "loss": 0.261,
+      "step": 74105
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4489829428608802e-05,
+      "loss": 0.2796,
+      "step": 74110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4486981303834485e-05,
+      "loss": 0.2656,
+      "step": 74115
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.448413318572138e-05,
+      "loss": 0.2802,
+      "step": 74120
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4481285074306468e-05,
+      "loss": 0.2774,
+      "step": 74125
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4478436969626727e-05,
+      "loss": 0.2624,
+      "step": 74130
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4475588871719144e-05,
+      "loss": 0.2983,
+      "step": 74135
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4472740780620678e-05,
+      "loss": 0.2542,
+      "step": 74140
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.446989269636834e-05,
+      "loss": 0.2763,
+      "step": 74145
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.446704461899909e-05,
+      "loss": 0.2834,
+      "step": 74150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.446419654854992e-05,
+      "loss": 0.2549,
+      "step": 74155
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4461348485057792e-05,
+      "loss": 0.2625,
+      "step": 74160
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.445850042855971e-05,
+      "loss": 0.2741,
+      "step": 74165
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4455652379092644e-05,
+      "loss": 0.2667,
+      "step": 74170
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.445280433669357e-05,
+      "loss": 0.2613,
+      "step": 74175
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4449956301399464e-05,
+      "loss": 0.2699,
+      "step": 74180
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.444710827324732e-05,
+      "loss": 0.2531,
+      "step": 74185
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4444260252274116e-05,
+      "loss": 0.2748,
+      "step": 74190
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4441412238516823e-05,
+      "loss": 0.2512,
+      "step": 74195
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4438564232012412e-05,
+      "loss": 0.2867,
+      "step": 74200
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4435716232797882e-05,
+      "loss": 0.2848,
+      "step": 74205
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.443286824091021e-05,
+      "loss": 0.2665,
+      "step": 74210
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.443002025638637e-05,
+      "loss": 0.2785,
+      "step": 74215
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4427172279263333e-05,
+      "loss": 0.2872,
+      "step": 74220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4424324309578096e-05,
+      "loss": 0.2626,
+      "step": 74225
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4421476347367623e-05,
+      "loss": 0.2686,
+      "step": 74230
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.441862839266889e-05,
+      "loss": 0.2656,
+      "step": 74235
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.44157804455189e-05,
+      "loss": 0.2722,
+      "step": 74240
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.441293250595461e-05,
+      "loss": 0.266,
+      "step": 74245
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.441008457401301e-05,
+      "loss": 0.2711,
+      "step": 74250
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4407236649731056e-05,
+      "loss": 0.2852,
+      "step": 74255
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4404388733145756e-05,
+      "loss": 0.2881,
+      "step": 74260
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.440154082429408e-05,
+      "loss": 0.2974,
+      "step": 74265
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4398692923213e-05,
+      "loss": 0.294,
+      "step": 74270
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.439584502993948e-05,
+      "loss": 0.2635,
+      "step": 74275
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4392997144510533e-05,
+      "loss": 0.2973,
+      "step": 74280
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.439014926696311e-05,
+      "loss": 0.2617,
+      "step": 74285
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4387301397334203e-05,
+      "loss": 0.2876,
+      "step": 74290
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.438445353566077e-05,
+      "loss": 0.247,
+      "step": 74295
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4381605681979817e-05,
+      "loss": 0.2682,
+      "step": 74300
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4378757836328304e-05,
+      "loss": 0.2722,
+      "step": 74305
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4375909998743205e-05,
+      "loss": 0.259,
+      "step": 74310
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4373062169261498e-05,
+      "loss": 0.2706,
+      "step": 74315
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.437021434792018e-05,
+      "loss": 0.2927,
+      "step": 74320
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 2.4367366534756213e-05,
+      "loss": 0.282,
+      "step": 74325
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4364518729806572e-05,
+      "loss": 0.2638,
+      "step": 74330
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4361670933108224e-05,
+      "loss": 0.2674,
+      "step": 74335
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.435882314469817e-05,
+      "loss": 0.2617,
+      "step": 74340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4355975364613372e-05,
+      "loss": 0.2523,
+      "step": 74345
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4353127592890804e-05,
+      "loss": 0.2723,
+      "step": 74350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.435027982956746e-05,
+      "loss": 0.2797,
+      "step": 74355
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.43474320746803e-05,
+      "loss": 0.2813,
+      "step": 74360
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.43445843282663e-05,
+      "loss": 0.2605,
+      "step": 74365
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4341736590362445e-05,
+      "loss": 0.2604,
+      "step": 74370
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.43388888610057e-05,
+      "loss": 0.2687,
+      "step": 74375
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4336041140233058e-05,
+      "loss": 0.2694,
+      "step": 74380
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4333193428081483e-05,
+      "loss": 0.2787,
+      "step": 74385
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4330345724587937e-05,
+      "loss": 0.2734,
+      "step": 74390
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4327498029789422e-05,
+      "loss": 0.2889,
+      "step": 74395
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4324650343722905e-05,
+      "loss": 0.2638,
+      "step": 74400
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4321802666425357e-05,
+      "loss": 0.2766,
+      "step": 74405
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4318954997933743e-05,
+      "loss": 0.2545,
+      "step": 74410
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4316107338285062e-05,
+      "loss": 0.2671,
+      "step": 74415
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.431325968751627e-05,
+      "loss": 0.2798,
+      "step": 74420
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.431041204566436e-05,
+      "loss": 0.2563,
+      "step": 74425
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4307564412766276e-05,
+      "loss": 0.2845,
+      "step": 74430
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4304716788859024e-05,
+      "loss": 0.2695,
+      "step": 74435
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4301869173979568e-05,
+      "loss": 0.2853,
+      "step": 74440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4299021568164876e-05,
+      "loss": 0.2705,
+      "step": 74445
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.429617397145193e-05,
+      "loss": 0.2638,
+      "step": 74450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4293326383877694e-05,
+      "loss": 0.2754,
+      "step": 74455
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4290478805479158e-05,
+      "loss": 0.2483,
+      "step": 74460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4287631236293287e-05,
+      "loss": 0.25,
+      "step": 74465
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4284783676357044e-05,
+      "loss": 0.2723,
+      "step": 74470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4281936125707418e-05,
+      "loss": 0.2838,
+      "step": 74475
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4279088584381386e-05,
+      "loss": 0.2648,
+      "step": 74480
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4276241052415898e-05,
+      "loss": 0.2948,
+      "step": 74485
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.427339352984796e-05,
+      "loss": 0.2717,
+      "step": 74490
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4270546016714525e-05,
+      "loss": 0.2738,
+      "step": 74495
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4267698513052565e-05,
+      "loss": 0.2835,
+      "step": 74500
+    },
+    {
+      "epoch": 2.62,
+      "eval_loss": 0.2619902491569519,
+      "eval_runtime": 10.556,
+      "eval_samples_per_second": 9.473,
+      "eval_steps_per_second": 9.473,
+      "step": 74500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4264851018899058e-05,
+      "loss": 0.2681,
+      "step": 74505
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4262003534290972e-05,
+      "loss": 0.254,
+      "step": 74510
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4259156059265292e-05,
+      "loss": 0.2792,
+      "step": 74515
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4256308593858982e-05,
+      "loss": 0.2778,
+      "step": 74520
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4253461138109004e-05,
+      "loss": 0.2737,
+      "step": 74525
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4250613692052352e-05,
+      "loss": 0.2618,
+      "step": 74530
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.424776625572598e-05,
+      "loss": 0.2901,
+      "step": 74535
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4244918829166876e-05,
+      "loss": 0.275,
+      "step": 74540
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4242071412411994e-05,
+      "loss": 0.2533,
+      "step": 74545
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.423922400549832e-05,
+      "loss": 0.2717,
+      "step": 74550
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.423637660846282e-05,
+      "loss": 0.2673,
+      "step": 74555
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4233529221342473e-05,
+      "loss": 0.2636,
+      "step": 74560
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4230681844174228e-05,
+      "loss": 0.2511,
+      "step": 74565
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4227834476995086e-05,
+      "loss": 0.2544,
+      "step": 74570
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4224987119842006e-05,
+      "loss": 0.276,
+      "step": 74575
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4222139772751952e-05,
+      "loss": 0.2681,
+      "step": 74580
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4219292435761895e-05,
+      "loss": 0.2883,
+      "step": 74585
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4216445108908817e-05,
+      "loss": 0.2474,
+      "step": 74590
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4213597792229685e-05,
+      "loss": 0.2678,
+      "step": 74595
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4210750485761457e-05,
+      "loss": 0.271,
+      "step": 74600
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.4207903189541126e-05,
+      "loss": 0.2604,
+      "step": 74605
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 2.420505590360565e-05,
+      "loss": 0.2676,
+      "step": 74610
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4202208627991994e-05,
+      "loss": 0.2623,
+      "step": 74615
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4199361362737127e-05,
+      "loss": 0.2597,
+      "step": 74620
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4196514107878038e-05,
+      "loss": 0.254,
+      "step": 74625
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.419366686345168e-05,
+      "loss": 0.2658,
+      "step": 74630
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.419081962949503e-05,
+      "loss": 0.2685,
+      "step": 74635
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.418797240604504e-05,
+      "loss": 0.2688,
+      "step": 74640
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.41851251931387e-05,
+      "loss": 0.2555,
+      "step": 74645
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4182277990812976e-05,
+      "loss": 0.2566,
+      "step": 74650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4179430799104834e-05,
+      "loss": 0.2696,
+      "step": 74655
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4176583618051236e-05,
+      "loss": 0.2613,
+      "step": 74660
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4173736447689165e-05,
+      "loss": 0.2532,
+      "step": 74665
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4170889288055572e-05,
+      "loss": 0.2818,
+      "step": 74670
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4168042139187447e-05,
+      "loss": 0.2913,
+      "step": 74675
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4165195001121732e-05,
+      "loss": 0.2811,
+      "step": 74680
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.416234787389542e-05,
+      "loss": 0.2603,
+      "step": 74685
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4159500757545473e-05,
+      "loss": 0.2629,
+      "step": 74690
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4156653652108847e-05,
+      "loss": 0.2686,
+      "step": 74695
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.415380655762251e-05,
+      "loss": 0.2839,
+      "step": 74700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4150959474123454e-05,
+      "loss": 0.2787,
+      "step": 74705
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4148112401648624e-05,
+      "loss": 0.2747,
+      "step": 74710
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4145265340234983e-05,
+      "loss": 0.2692,
+      "step": 74715
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.414241828991952e-05,
+      "loss": 0.2783,
+      "step": 74720
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4139571250739183e-05,
+      "loss": 0.2737,
+      "step": 74725
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4136724222730957e-05,
+      "loss": 0.2638,
+      "step": 74730
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.413387720593178e-05,
+      "loss": 0.2488,
+      "step": 74735
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4131030200378653e-05,
+      "loss": 0.2688,
+      "step": 74740
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4128183206108524e-05,
+      "loss": 0.2412,
+      "step": 74745
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4125336223158358e-05,
+      "loss": 0.2759,
+      "step": 74750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4122489251565117e-05,
+      "loss": 0.302,
+      "step": 74755
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4119642291365785e-05,
+      "loss": 0.2656,
+      "step": 74760
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.411679534259732e-05,
+      "loss": 0.2767,
+      "step": 74765
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4113948405296684e-05,
+      "loss": 0.2715,
+      "step": 74770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4111101479500835e-05,
+      "loss": 0.2581,
+      "step": 74775
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.410825456524675e-05,
+      "loss": 0.2648,
+      "step": 74780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.41054076625714e-05,
+      "loss": 0.2657,
+      "step": 74785
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.410256077151174e-05,
+      "loss": 0.2561,
+      "step": 74790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4099713892104727e-05,
+      "loss": 0.2544,
+      "step": 74795
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4096867024387345e-05,
+      "loss": 0.2728,
+      "step": 74800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4094020168396548e-05,
+      "loss": 0.2933,
+      "step": 74805
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4091173324169304e-05,
+      "loss": 0.2865,
+      "step": 74810
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4088326491742567e-05,
+      "loss": 0.2505,
+      "step": 74815
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.408547967115332e-05,
+      "loss": 0.2515,
+      "step": 74820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4082632862438514e-05,
+      "loss": 0.2667,
+      "step": 74825
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4079786065635107e-05,
+      "loss": 0.2822,
+      "step": 74830
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4076939280780075e-05,
+      "loss": 0.261,
+      "step": 74835
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4074092507910386e-05,
+      "loss": 0.2563,
+      "step": 74840
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4071245747062996e-05,
+      "loss": 0.2818,
+      "step": 74845
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4068398998274854e-05,
+      "loss": 0.2844,
+      "step": 74850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4065552261582952e-05,
+      "loss": 0.2628,
+      "step": 74855
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.406270553702423e-05,
+      "loss": 0.2968,
+      "step": 74860
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4059858824635663e-05,
+      "loss": 0.263,
+      "step": 74865
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4057012124454205e-05,
+      "loss": 0.2644,
+      "step": 74870
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4054165436516825e-05,
+      "loss": 0.256,
+      "step": 74875
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4051318760860488e-05,
+      "loss": 0.2898,
+      "step": 74880
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.4048472097522148e-05,
+      "loss": 0.2598,
+      "step": 74885
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 2.404562544653877e-05,
+      "loss": 0.2734,
+      "step": 74890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4042778807947313e-05,
+      "loss": 0.2798,
+      "step": 74895
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4039932181784754e-05,
+      "loss": 0.272,
+      "step": 74900
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.403708556808804e-05,
+      "loss": 0.2886,
+      "step": 74905
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.403423896689412e-05,
+      "loss": 0.2677,
+      "step": 74910
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.403139237823998e-05,
+      "loss": 0.2875,
+      "step": 74915
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4028545802162578e-05,
+      "loss": 0.2836,
+      "step": 74920
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.402569923869887e-05,
+      "loss": 0.2768,
+      "step": 74925
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.40228526878858e-05,
+      "loss": 0.3029,
+      "step": 74930
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4020006149760355e-05,
+      "loss": 0.26,
+      "step": 74935
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.401715962435948e-05,
+      "loss": 0.2837,
+      "step": 74940
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.401431311172014e-05,
+      "loss": 0.2707,
+      "step": 74945
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4011466611879297e-05,
+      "loss": 0.2681,
+      "step": 74950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.400862012487391e-05,
+      "loss": 0.2783,
+      "step": 74955
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.400577365074094e-05,
+      "loss": 0.2826,
+      "step": 74960
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.400292718951733e-05,
+      "loss": 0.2472,
+      "step": 74965
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.4000080741240066e-05,
+      "loss": 0.2662,
+      "step": 74970
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.399723430594609e-05,
+      "loss": 0.2496,
+      "step": 74975
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.399438788367237e-05,
+      "loss": 0.2745,
+      "step": 74980
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.399154147445585e-05,
+      "loss": 0.2454,
+      "step": 74985
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.398869507833351e-05,
+      "loss": 0.2886,
+      "step": 74990
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3985848695342294e-05,
+      "loss": 0.2567,
+      "step": 74995
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.398300232551917e-05,
+      "loss": 0.2774,
+      "step": 75000
+    },
+    {
+      "epoch": 2.64,
+      "eval_loss": 0.2608758807182312,
+      "eval_runtime": 10.5423,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 75000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3980155968901078e-05,
+      "loss": 0.2513,
+      "step": 75005
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3977309625524998e-05,
+      "loss": 0.2618,
+      "step": 75010
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.397446329542788e-05,
+      "loss": 0.2612,
+      "step": 75015
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3971616978646678e-05,
+      "loss": 0.2762,
+      "step": 75020
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.396877067521835e-05,
+      "loss": 0.2579,
+      "step": 75025
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3965924385179853e-05,
+      "loss": 0.2517,
+      "step": 75030
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.396307810856815e-05,
+      "loss": 0.2847,
+      "step": 75035
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.39602318454202e-05,
+      "loss": 0.2795,
+      "step": 75040
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.395738559577294e-05,
+      "loss": 0.2711,
+      "step": 75045
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.395453935966335e-05,
+      "loss": 0.2744,
+      "step": 75050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3951693137128377e-05,
+      "loss": 0.2652,
+      "step": 75055
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.394884692820497e-05,
+      "loss": 0.2746,
+      "step": 75060
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3946000732930105e-05,
+      "loss": 0.2683,
+      "step": 75065
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3943154551340724e-05,
+      "loss": 0.2474,
+      "step": 75070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3940308383473788e-05,
+      "loss": 0.2929,
+      "step": 75075
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3937462229366234e-05,
+      "loss": 0.2724,
+      "step": 75080
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3934616089055038e-05,
+      "loss": 0.2569,
+      "step": 75085
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.393176996257716e-05,
+      "loss": 0.2704,
+      "step": 75090
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3928923849969543e-05,
+      "loss": 0.2602,
+      "step": 75095
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.392607775126913e-05,
+      "loss": 0.2738,
+      "step": 75100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3923231666512906e-05,
+      "loss": 0.2869,
+      "step": 75105
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.39203855957378e-05,
+      "loss": 0.2515,
+      "step": 75110
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.391753953898078e-05,
+      "loss": 0.2623,
+      "step": 75115
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3914693496278787e-05,
+      "loss": 0.2683,
+      "step": 75120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3911847467668793e-05,
+      "loss": 0.2926,
+      "step": 75125
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3909001453187747e-05,
+      "loss": 0.2708,
+      "step": 75130
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3906155452872588e-05,
+      "loss": 0.2827,
+      "step": 75135
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3903309466760274e-05,
+      "loss": 0.3036,
+      "step": 75140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3900463494887776e-05,
+      "loss": 0.2739,
+      "step": 75145
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3897617537292036e-05,
+      "loss": 0.2748,
+      "step": 75150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3894771594010004e-05,
+      "loss": 0.2815,
+      "step": 75155
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3891925665078624e-05,
+      "loss": 0.2412,
+      "step": 75160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3889079750534865e-05,
+      "loss": 0.2747,
+      "step": 75165
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.388623385041568e-05,
+      "loss": 0.2702,
+      "step": 75170
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 2.3883387964758e-05,
+      "loss": 0.2612,
+      "step": 75175
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3880542093598805e-05,
+      "loss": 0.2415,
+      "step": 75180
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3877696236975034e-05,
+      "loss": 0.268,
+      "step": 75185
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.387485039492363e-05,
+      "loss": 0.2636,
+      "step": 75190
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.387200456748155e-05,
+      "loss": 0.2544,
+      "step": 75195
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3869158754685753e-05,
+      "loss": 0.2749,
+      "step": 75200
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3866312956573188e-05,
+      "loss": 0.2571,
+      "step": 75205
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.38634671731808e-05,
+      "loss": 0.2724,
+      "step": 75210
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.386062140454553e-05,
+      "loss": 0.2596,
+      "step": 75215
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.385777565070435e-05,
+      "loss": 0.2694,
+      "step": 75220
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3854929911694203e-05,
+      "loss": 0.2482,
+      "step": 75225
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.385208418755204e-05,
+      "loss": 0.2841,
+      "step": 75230
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.384923847831479e-05,
+      "loss": 0.2664,
+      "step": 75235
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3846392784019436e-05,
+      "loss": 0.2859,
+      "step": 75240
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3843547104702903e-05,
+      "loss": 0.2856,
+      "step": 75245
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3840701440402156e-05,
+      "loss": 0.2696,
+      "step": 75250
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3837855791154127e-05,
+      "loss": 0.2782,
+      "step": 75255
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3835010156995783e-05,
+      "loss": 0.2685,
+      "step": 75260
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3832164537964066e-05,
+      "loss": 0.2724,
+      "step": 75265
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3829318934095918e-05,
+      "loss": 0.2717,
+      "step": 75270
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3826473345428285e-05,
+      "loss": 0.2621,
+      "step": 75275
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3823627771998134e-05,
+      "loss": 0.2779,
+      "step": 75280
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.38207822138424e-05,
+      "loss": 0.2559,
+      "step": 75285
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.381793667099802e-05,
+      "loss": 0.2871,
+      "step": 75290
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3815091143501968e-05,
+      "loss": 0.2464,
+      "step": 75295
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3812245631391168e-05,
+      "loss": 0.266,
+      "step": 75300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3809400134702582e-05,
+      "loss": 0.2678,
+      "step": 75305
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380655465347314e-05,
+      "loss": 0.244,
+      "step": 75310
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380370918773981e-05,
+      "loss": 0.2402,
+      "step": 75315
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.380086373753953e-05,
+      "loss": 0.2624,
+      "step": 75320
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3798018302909235e-05,
+      "loss": 0.265,
+      "step": 75325
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.379517288388588e-05,
+      "loss": 0.2735,
+      "step": 75330
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3792327480506412e-05,
+      "loss": 0.2528,
+      "step": 75335
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3789482092807784e-05,
+      "loss": 0.2571,
+      "step": 75340
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3786636720826934e-05,
+      "loss": 0.2701,
+      "step": 75345
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.378379136460079e-05,
+      "loss": 0.2734,
+      "step": 75350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.378094602416632e-05,
+      "loss": 0.2922,
+      "step": 75355
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3778100699560474e-05,
+      "loss": 0.2715,
+      "step": 75360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.377525539082018e-05,
+      "loss": 0.2577,
+      "step": 75365
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.377241009798238e-05,
+      "loss": 0.2751,
+      "step": 75370
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.376956482108403e-05,
+      "loss": 0.2759,
+      "step": 75375
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.376671956016207e-05,
+      "loss": 0.2581,
+      "step": 75380
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.376387431525345e-05,
+      "loss": 0.2553,
+      "step": 75385
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3761029086395095e-05,
+      "loss": 0.2709,
+      "step": 75390
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.375818387362397e-05,
+      "loss": 0.2801,
+      "step": 75395
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3755338676977012e-05,
+      "loss": 0.2996,
+      "step": 75400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3752493496491148e-05,
+      "loss": 0.2742,
+      "step": 75405
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3749648332203346e-05,
+      "loss": 0.257,
+      "step": 75410
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.374680318415053e-05,
+      "loss": 0.278,
+      "step": 75415
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3743958052369652e-05,
+      "loss": 0.2875,
+      "step": 75420
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3741112936897646e-05,
+      "loss": 0.2619,
+      "step": 75425
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3738267837771464e-05,
+      "loss": 0.2756,
+      "step": 75430
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.373542275502804e-05,
+      "loss": 0.2495,
+      "step": 75435
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3732577688704323e-05,
+      "loss": 0.2631,
+      "step": 75440
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3729732638837238e-05,
+      "loss": 0.2677,
+      "step": 75445
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3726887605463748e-05,
+      "loss": 0.2637,
+      "step": 75450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3724042588620786e-05,
+      "loss": 0.2702,
+      "step": 75455
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 2.3721197588345285e-05,
+      "loss": 0.2665,
+      "step": 75460
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3718352604674187e-05,
+      "loss": 0.2869,
+      "step": 75465
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3715507637644437e-05,
+      "loss": 0.2535,
+      "step": 75470
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.371266268729298e-05,
+      "loss": 0.2708,
+      "step": 75475
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3709817753656748e-05,
+      "loss": 0.27,
+      "step": 75480
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3706972836772673e-05,
+      "loss": 0.2495,
+      "step": 75485
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3704127936677713e-05,
+      "loss": 0.2591,
+      "step": 75490
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3701283053408795e-05,
+      "loss": 0.2735,
+      "step": 75495
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3698438187002863e-05,
+      "loss": 0.2667,
+      "step": 75500
+    },
+    {
+      "epoch": 2.66,
+      "eval_loss": 0.2609610855579376,
+      "eval_runtime": 10.5588,
+      "eval_samples_per_second": 9.471,
+      "eval_steps_per_second": 9.471,
+      "step": 75500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3695593337496847e-05,
+      "loss": 0.2506,
+      "step": 75505
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.36927485049277e-05,
+      "loss": 0.2631,
+      "step": 75510
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3689903689332353e-05,
+      "loss": 0.2895,
+      "step": 75515
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.368705889074773e-05,
+      "loss": 0.266,
+      "step": 75520
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3684214109210788e-05,
+      "loss": 0.2753,
+      "step": 75525
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3681369344758464e-05,
+      "loss": 0.2822,
+      "step": 75530
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.367852459742769e-05,
+      "loss": 0.2874,
+      "step": 75535
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.367567986725539e-05,
+      "loss": 0.2457,
+      "step": 75540
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3672835154278534e-05,
+      "loss": 0.2878,
+      "step": 75545
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3669990458534023e-05,
+      "loss": 0.2828,
+      "step": 75550
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3667145780058823e-05,
+      "loss": 0.263,
+      "step": 75555
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3664301118889842e-05,
+      "loss": 0.2772,
+      "step": 75560
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3661456475064043e-05,
+      "loss": 0.2477,
+      "step": 75565
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3658611848618346e-05,
+      "loss": 0.2818,
+      "step": 75570
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3655767239589687e-05,
+      "loss": 0.259,
+      "step": 75575
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3652922648015e-05,
+      "loss": 0.2561,
+      "step": 75580
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3650078073931237e-05,
+      "loss": 0.2811,
+      "step": 75585
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3647233517375317e-05,
+      "loss": 0.2783,
+      "step": 75590
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.364438897838418e-05,
+      "loss": 0.2699,
+      "step": 75595
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3641544456994747e-05,
+      "loss": 0.2716,
+      "step": 75600
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.363869995324397e-05,
+      "loss": 0.2684,
+      "step": 75605
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3635855467168783e-05,
+      "loss": 0.292,
+      "step": 75610
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3633010998806113e-05,
+      "loss": 0.2756,
+      "step": 75615
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.363016654819288e-05,
+      "loss": 0.2436,
+      "step": 75620
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.362732211536605e-05,
+      "loss": 0.2806,
+      "step": 75625
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3624477700362526e-05,
+      "loss": 0.271,
+      "step": 75630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.362163330321925e-05,
+      "loss": 0.2794,
+      "step": 75635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3618788923973167e-05,
+      "loss": 0.2581,
+      "step": 75640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.36159445626612e-05,
+      "loss": 0.2853,
+      "step": 75645
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3613100219320282e-05,
+      "loss": 0.2975,
+      "step": 75650
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3610255893987328e-05,
+      "loss": 0.2612,
+      "step": 75655
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3607411586699296e-05,
+      "loss": 0.2624,
+      "step": 75660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.360456729749311e-05,
+      "loss": 0.2454,
+      "step": 75665
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.36017230264057e-05,
+      "loss": 0.2618,
+      "step": 75670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3598878773473982e-05,
+      "loss": 0.2853,
+      "step": 75675
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3596034538734908e-05,
+      "loss": 0.2617,
+      "step": 75680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.35931903222254e-05,
+      "loss": 0.2671,
+      "step": 75685
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3590346123982394e-05,
+      "loss": 0.2818,
+      "step": 75690
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.35875019440428e-05,
+      "loss": 0.2639,
+      "step": 75695
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3584657782443575e-05,
+      "loss": 0.2507,
+      "step": 75700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3581813639221637e-05,
+      "loss": 0.2559,
+      "step": 75705
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3578969514413908e-05,
+      "loss": 0.2618,
+      "step": 75710
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3576125408057317e-05,
+      "loss": 0.2597,
+      "step": 75715
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3573281320188813e-05,
+      "loss": 0.2762,
+      "step": 75720
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3570437250845306e-05,
+      "loss": 0.2716,
+      "step": 75725
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3567593200063733e-05,
+      "loss": 0.2634,
+      "step": 75730
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3564749167881007e-05,
+      "loss": 0.2626,
+      "step": 75735
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.356190515433407e-05,
+      "loss": 0.2668,
+      "step": 75740
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 2.3559061159459854e-05,
+      "loss": 0.2505,
+      "step": 75745
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.355621718329528e-05,
+      "loss": 0.2781,
+      "step": 75750
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3553373225877263e-05,
+      "loss": 0.2879,
+      "step": 75755
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.355052928724275e-05,
+      "loss": 0.2771,
+      "step": 75760
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3547685367428656e-05,
+      "loss": 0.2718,
+      "step": 75765
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.354484146647191e-05,
+      "loss": 0.2845,
+      "step": 75770
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3541997584409438e-05,
+      "loss": 0.2798,
+      "step": 75775
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.353915372127817e-05,
+      "loss": 0.2676,
+      "step": 75780
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3536309877115033e-05,
+      "loss": 0.2759,
+      "step": 75785
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3533466051956938e-05,
+      "loss": 0.2623,
+      "step": 75790
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3530622245840818e-05,
+      "loss": 0.2667,
+      "step": 75795
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.352777845880361e-05,
+      "loss": 0.2681,
+      "step": 75800
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.352493469088223e-05,
+      "loss": 0.2697,
+      "step": 75805
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3522659690333325e-05,
+      "loss": 0.2463,
+      "step": 75810
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3519815956913475e-05,
+      "loss": 0.2921,
+      "step": 75815
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3516972242712848e-05,
+      "loss": 0.2793,
+      "step": 75820
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3514128547768342e-05,
+      "loss": 0.2738,
+      "step": 75825
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3511284872116906e-05,
+      "loss": 0.2991,
+      "step": 75830
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.350844121579546e-05,
+      "loss": 0.2796,
+      "step": 75835
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.350559757884091e-05,
+      "loss": 0.2683,
+      "step": 75840
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3502753961290204e-05,
+      "loss": 0.2594,
+      "step": 75845
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3499910363180246e-05,
+      "loss": 0.2886,
+      "step": 75850
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.349706678454796e-05,
+      "loss": 0.2805,
+      "step": 75855
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3494223225430266e-05,
+      "loss": 0.2628,
+      "step": 75860
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3491379685864105e-05,
+      "loss": 0.272,
+      "step": 75865
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3488536165886384e-05,
+      "loss": 0.2463,
+      "step": 75870
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.348569266553402e-05,
+      "loss": 0.2638,
+      "step": 75875
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.348284918484393e-05,
+      "loss": 0.2661,
+      "step": 75880
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.348000572385305e-05,
+      "loss": 0.2953,
+      "step": 75885
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.34771622825983e-05,
+      "loss": 0.2886,
+      "step": 75890
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3474318861116594e-05,
+      "loss": 0.2792,
+      "step": 75895
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.347147545944484e-05,
+      "loss": 0.2792,
+      "step": 75900
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.346863207761998e-05,
+      "loss": 0.2595,
+      "step": 75905
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3465788715678922e-05,
+      "loss": 0.263,
+      "step": 75910
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3462945373658588e-05,
+      "loss": 0.2568,
+      "step": 75915
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3460102051595885e-05,
+      "loss": 0.2462,
+      "step": 75920
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.345725874952775e-05,
+      "loss": 0.2756,
+      "step": 75925
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3454415467491102e-05,
+      "loss": 0.2813,
+      "step": 75930
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.345157220552284e-05,
+      "loss": 0.259,
+      "step": 75935
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3448728963659885e-05,
+      "loss": 0.2697,
+      "step": 75940
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3445885741939175e-05,
+      "loss": 0.2627,
+      "step": 75945
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3443042540397615e-05,
+      "loss": 0.2602,
+      "step": 75950
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.344019935907211e-05,
+      "loss": 0.2678,
+      "step": 75955
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.34373561979996e-05,
+      "loss": 0.2482,
+      "step": 75960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3434513057216984e-05,
+      "loss": 0.2637,
+      "step": 75965
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.343166993676119e-05,
+      "loss": 0.2795,
+      "step": 75970
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3428826836669116e-05,
+      "loss": 0.277,
+      "step": 75975
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.34259837569777e-05,
+      "loss": 0.246,
+      "step": 75980
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.342314069772385e-05,
+      "loss": 0.2809,
+      "step": 75985
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3420297658944472e-05,
+      "loss": 0.267,
+      "step": 75990
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.341745464067649e-05,
+      "loss": 0.274,
+      "step": 75995
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3414611642956816e-05,
+      "loss": 0.2866,
+      "step": 76000
+    },
+    {
+      "epoch": 2.67,
+      "eval_loss": 0.2612769603729248,
+      "eval_runtime": 10.5493,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 76000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.341176866582237e-05,
+      "loss": 0.2699,
+      "step": 76005
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3408925709310063e-05,
+      "loss": 0.2581,
+      "step": 76010
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3406082773456792e-05,
+      "loss": 0.2552,
+      "step": 76015
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.340323985829949e-05,
+      "loss": 0.259,
+      "step": 76020
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.340039696387508e-05,
+      "loss": 0.2766,
+      "step": 76025
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 2.3397554090220456e-05,
+      "loss": 0.2715,
+      "step": 76030
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3394711237372525e-05,
+      "loss": 0.2591,
+      "step": 76035
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3391868405368218e-05,
+      "loss": 0.2483,
+      "step": 76040
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.338902559424444e-05,
+      "loss": 0.2635,
+      "step": 76045
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3386182804038108e-05,
+      "loss": 0.2694,
+      "step": 76050
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3383340034786118e-05,
+      "loss": 0.2895,
+      "step": 76055
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.33804972865254e-05,
+      "loss": 0.2687,
+      "step": 76060
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3377654559292862e-05,
+      "loss": 0.279,
+      "step": 76065
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3374811853125393e-05,
+      "loss": 0.275,
+      "step": 76070
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3371969168059937e-05,
+      "loss": 0.2605,
+      "step": 76075
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.336912650413338e-05,
+      "loss": 0.2477,
+      "step": 76080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3366283861382648e-05,
+      "loss": 0.2602,
+      "step": 76085
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3363441239844632e-05,
+      "loss": 0.2465,
+      "step": 76090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3360598639556268e-05,
+      "loss": 0.2853,
+      "step": 76095
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3357756060554445e-05,
+      "loss": 0.2683,
+      "step": 76100
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3354913502876072e-05,
+      "loss": 0.282,
+      "step": 76105
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.335207096655806e-05,
+      "loss": 0.2867,
+      "step": 76110
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3349228451637327e-05,
+      "loss": 0.2511,
+      "step": 76115
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3346385958150778e-05,
+      "loss": 0.2747,
+      "step": 76120
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3343543486135314e-05,
+      "loss": 0.2776,
+      "step": 76125
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3340701035627843e-05,
+      "loss": 0.2435,
+      "step": 76130
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.333785860666528e-05,
+      "loss": 0.2765,
+      "step": 76135
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3335016199284528e-05,
+      "loss": 0.2888,
+      "step": 76140
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3332173813522497e-05,
+      "loss": 0.2604,
+      "step": 76145
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3329331449416077e-05,
+      "loss": 0.2646,
+      "step": 76150
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3326489107002194e-05,
+      "loss": 0.2574,
+      "step": 76155
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3323646786317748e-05,
+      "loss": 0.2908,
+      "step": 76160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3320804487399647e-05,
+      "loss": 0.2683,
+      "step": 76165
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3317962210284784e-05,
+      "loss": 0.2552,
+      "step": 76170
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.331511995501008e-05,
+      "loss": 0.2676,
+      "step": 76175
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3312277721612434e-05,
+      "loss": 0.2623,
+      "step": 76180
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.330943551012874e-05,
+      "loss": 0.2738,
+      "step": 76185
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3306593320595918e-05,
+      "loss": 0.2851,
+      "step": 76190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.330375115305087e-05,
+      "loss": 0.2665,
+      "step": 76195
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3300909007530493e-05,
+      "loss": 0.2861,
+      "step": 76200
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3298066884071686e-05,
+      "loss": 0.2728,
+      "step": 76205
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.329522478271137e-05,
+      "loss": 0.2538,
+      "step": 76210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3292382703486428e-05,
+      "loss": 0.2773,
+      "step": 76215
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3289540646433777e-05,
+      "loss": 0.2909,
+      "step": 76220
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3286698611590304e-05,
+      "loss": 0.2853,
+      "step": 76225
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.328385659899293e-05,
+      "loss": 0.2606,
+      "step": 76230
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.328101460867855e-05,
+      "loss": 0.2718,
+      "step": 76235
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3278172640684056e-05,
+      "loss": 0.2591,
+      "step": 76240
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.327533069504635e-05,
+      "loss": 0.264,
+      "step": 76245
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.327248877180235e-05,
+      "loss": 0.2514,
+      "step": 76250
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3269646870988944e-05,
+      "loss": 0.2671,
+      "step": 76255
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.326680499264303e-05,
+      "loss": 0.2581,
+      "step": 76260
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3263963136801505e-05,
+      "loss": 0.281,
+      "step": 76265
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3261121303501274e-05,
+      "loss": 0.2668,
+      "step": 76270
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.325827949277925e-05,
+      "loss": 0.2715,
+      "step": 76275
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3255437704672312e-05,
+      "loss": 0.2676,
+      "step": 76280
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3252595939217357e-05,
+      "loss": 0.3085,
+      "step": 76285
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3249754196451304e-05,
+      "loss": 0.2605,
+      "step": 76290
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.324691247641103e-05,
+      "loss": 0.2552,
+      "step": 76295
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3244070779133442e-05,
+      "loss": 0.2588,
+      "step": 76300
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.324122910465545e-05,
+      "loss": 0.2745,
+      "step": 76305
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3238387453013934e-05,
+      "loss": 0.2781,
+      "step": 76310
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 2.3235545824245797e-05,
+      "loss": 0.2418,
+      "step": 76315
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3232704218387923e-05,
+      "loss": 0.2643,
+      "step": 76320
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3229862635477228e-05,
+      "loss": 0.2813,
+      "step": 76325
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3227021075550606e-05,
+      "loss": 0.2506,
+      "step": 76330
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3224179538644944e-05,
+      "loss": 0.2613,
+      "step": 76335
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.322133802479713e-05,
+      "loss": 0.2741,
+      "step": 76340
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3218496534044082e-05,
+      "loss": 0.2694,
+      "step": 76345
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.321565506642268e-05,
+      "loss": 0.2591,
+      "step": 76350
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.321281362196982e-05,
+      "loss": 0.2826,
+      "step": 76355
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.320997220072239e-05,
+      "loss": 0.2798,
+      "step": 76360
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3207130802717304e-05,
+      "loss": 0.278,
+      "step": 76365
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3204289427991437e-05,
+      "loss": 0.2806,
+      "step": 76370
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3201448076581688e-05,
+      "loss": 0.2719,
+      "step": 76375
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.319860674852494e-05,
+      "loss": 0.2532,
+      "step": 76380
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3195765443858116e-05,
+      "loss": 0.2692,
+      "step": 76385
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.319292416261808e-05,
+      "loss": 0.2773,
+      "step": 76390
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3190082904841737e-05,
+      "loss": 0.272,
+      "step": 76395
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.318724167056596e-05,
+      "loss": 0.2646,
+      "step": 76400
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3184400459827666e-05,
+      "loss": 0.2476,
+      "step": 76405
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3181559272663737e-05,
+      "loss": 0.2573,
+      "step": 76410
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3178718109111052e-05,
+      "loss": 0.248,
+      "step": 76415
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3175876969206524e-05,
+      "loss": 0.2775,
+      "step": 76420
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3173035852987036e-05,
+      "loss": 0.2733,
+      "step": 76425
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.317019476048946e-05,
+      "loss": 0.2875,
+      "step": 76430
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3167353691750703e-05,
+      "loss": 0.2397,
+      "step": 76435
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.316451264680765e-05,
+      "loss": 0.2967,
+      "step": 76440
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3161671625697197e-05,
+      "loss": 0.2505,
+      "step": 76445
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3158830628456225e-05,
+      "loss": 0.2745,
+      "step": 76450
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3155989655121614e-05,
+      "loss": 0.2636,
+      "step": 76455
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.315314870573027e-05,
+      "loss": 0.2789,
+      "step": 76460
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.315030778031907e-05,
+      "loss": 0.2802,
+      "step": 76465
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3147466878924915e-05,
+      "loss": 0.2532,
+      "step": 76470
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3144626001584663e-05,
+      "loss": 0.2909,
+      "step": 76475
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3141785148335234e-05,
+      "loss": 0.256,
+      "step": 76480
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3138944319213497e-05,
+      "loss": 0.2513,
+      "step": 76485
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.313610351425634e-05,
+      "loss": 0.2494,
+      "step": 76490
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3133262733500646e-05,
+      "loss": 0.2747,
+      "step": 76495
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3130421976983314e-05,
+      "loss": 0.2887,
+      "step": 76500
+    },
+    {
+      "epoch": 2.69,
+      "eval_loss": 0.2611083686351776,
+      "eval_runtime": 10.5437,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 76500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.312758124474122e-05,
+      "loss": 0.2711,
+      "step": 76505
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3124740536811245e-05,
+      "loss": 0.2671,
+      "step": 76510
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3121899853230278e-05,
+      "loss": 0.2561,
+      "step": 76515
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3119059194035202e-05,
+      "loss": 0.2709,
+      "step": 76520
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3116218559262912e-05,
+      "loss": 0.2708,
+      "step": 76525
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3113377948950267e-05,
+      "loss": 0.2657,
+      "step": 76530
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.311053736313418e-05,
+      "loss": 0.2494,
+      "step": 76535
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3107696801851513e-05,
+      "loss": 0.264,
+      "step": 76540
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3104856265139164e-05,
+      "loss": 0.2555,
+      "step": 76545
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3102015753033996e-05,
+      "loss": 0.2725,
+      "step": 76550
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3099175265572912e-05,
+      "loss": 0.2636,
+      "step": 76555
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3096334802792787e-05,
+      "loss": 0.2668,
+      "step": 76560
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.309349436473049e-05,
+      "loss": 0.2711,
+      "step": 76565
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3090653951422914e-05,
+      "loss": 0.2713,
+      "step": 76570
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.308781356290694e-05,
+      "loss": 0.2534,
+      "step": 76575
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3084973199219452e-05,
+      "loss": 0.2498,
+      "step": 76580
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.308213286039732e-05,
+      "loss": 0.271,
+      "step": 76585
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3079292546477423e-05,
+      "loss": 0.2704,
+      "step": 76590
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 2.3076452257496658e-05,
+      "loss": 0.2688,
+      "step": 76595
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3073611993491885e-05,
+      "loss": 0.2491,
+      "step": 76600
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.30707717545e-05,
+      "loss": 0.2701,
+      "step": 76605
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3067931540557855e-05,
+      "loss": 0.27,
+      "step": 76610
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.306509135170236e-05,
+      "loss": 0.2353,
+      "step": 76615
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.306225118797038e-05,
+      "loss": 0.2765,
+      "step": 76620
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3059411049398784e-05,
+      "loss": 0.261,
+      "step": 76625
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.305657093602445e-05,
+      "loss": 0.2716,
+      "step": 76630
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3053730847884274e-05,
+      "loss": 0.2438,
+      "step": 76635
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3050890785015122e-05,
+      "loss": 0.2526,
+      "step": 76640
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3048050747453857e-05,
+      "loss": 0.2388,
+      "step": 76645
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3045210735237373e-05,
+      "loss": 0.2811,
+      "step": 76650
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.304237074840254e-05,
+      "loss": 0.2666,
+      "step": 76655
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3039530786986237e-05,
+      "loss": 0.2566,
+      "step": 76660
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3036690851025324e-05,
+      "loss": 0.2593,
+      "step": 76665
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.303385094055669e-05,
+      "loss": 0.2807,
+      "step": 76670
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3031011055617217e-05,
+      "loss": 0.2642,
+      "step": 76675
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3028171196243756e-05,
+      "loss": 0.2517,
+      "step": 76680
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3025331362473187e-05,
+      "loss": 0.2673,
+      "step": 76685
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.30224915543424e-05,
+      "loss": 0.2477,
+      "step": 76690
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.301965177188826e-05,
+      "loss": 0.2783,
+      "step": 76695
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3016812015147635e-05,
+      "loss": 0.2897,
+      "step": 76700
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3013972284157386e-05,
+      "loss": 0.2525,
+      "step": 76705
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3011132578954407e-05,
+      "loss": 0.2556,
+      "step": 76710
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3008292899575562e-05,
+      "loss": 0.2586,
+      "step": 76715
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.3005453246057725e-05,
+      "loss": 0.2679,
+      "step": 76720
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.300261361843775e-05,
+      "loss": 0.2646,
+      "step": 76725
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.299977401675253e-05,
+      "loss": 0.259,
+      "step": 76730
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2996934441038925e-05,
+      "loss": 0.2711,
+      "step": 76735
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.299409489133381e-05,
+      "loss": 0.2804,
+      "step": 76740
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2991255367674043e-05,
+      "loss": 0.2622,
+      "step": 76745
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.298841587009651e-05,
+      "loss": 0.2763,
+      "step": 76750
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.298557639863807e-05,
+      "loss": 0.2788,
+      "step": 76755
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.298273695333558e-05,
+      "loss": 0.275,
+      "step": 76760
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2979897534225928e-05,
+      "loss": 0.2933,
+      "step": 76765
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.297705814134598e-05,
+      "loss": 0.2498,
+      "step": 76770
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2974218774732604e-05,
+      "loss": 0.2771,
+      "step": 76775
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2971379434422647e-05,
+      "loss": 0.2682,
+      "step": 76780
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2968540120453003e-05,
+      "loss": 0.2858,
+      "step": 76785
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.296570083286052e-05,
+      "loss": 0.2574,
+      "step": 76790
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2962861571682075e-05,
+      "loss": 0.2707,
+      "step": 76795
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.296002233695452e-05,
+      "loss": 0.2781,
+      "step": 76800
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.295718312871475e-05,
+      "loss": 0.2666,
+      "step": 76805
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2954343946999602e-05,
+      "loss": 0.2449,
+      "step": 76810
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2951504791845945e-05,
+      "loss": 0.2578,
+      "step": 76815
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2948665663290644e-05,
+      "loss": 0.2534,
+      "step": 76820
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.294582656137058e-05,
+      "loss": 0.2635,
+      "step": 76825
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.29429874861226e-05,
+      "loss": 0.2841,
+      "step": 76830
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2940148437583575e-05,
+      "loss": 0.2808,
+      "step": 76835
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2937309415790353e-05,
+      "loss": 0.2633,
+      "step": 76840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2934470420779815e-05,
+      "loss": 0.2678,
+      "step": 76845
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2931631452588827e-05,
+      "loss": 0.2716,
+      "step": 76850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2928792511254235e-05,
+      "loss": 0.2717,
+      "step": 76855
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2925953596812898e-05,
+      "loss": 0.2703,
+      "step": 76860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.29231147093017e-05,
+      "loss": 0.2869,
+      "step": 76865
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.292027584875748e-05,
+      "loss": 0.2913,
+      "step": 76870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.291743701521711e-05,
+      "loss": 0.2903,
+      "step": 76875
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.2914598208717446e-05,
+      "loss": 0.2753,
+      "step": 76880
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2911759429295357e-05,
+      "loss": 0.274,
+      "step": 76885
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2908920676987693e-05,
+      "loss": 0.2704,
+      "step": 76890
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2906081951831308e-05,
+      "loss": 0.2785,
+      "step": 76895
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2903243253863075e-05,
+      "loss": 0.2702,
+      "step": 76900
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.290040458311985e-05,
+      "loss": 0.2588,
+      "step": 76905
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2897565939638487e-05,
+      "loss": 0.2606,
+      "step": 76910
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2894727323455835e-05,
+      "loss": 0.2743,
+      "step": 76915
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2891888734608775e-05,
+      "loss": 0.249,
+      "step": 76920
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.288905017313414e-05,
+      "loss": 0.2775,
+      "step": 76925
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2886211639068806e-05,
+      "loss": 0.2406,
+      "step": 76930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2883373132449608e-05,
+      "loss": 0.2914,
+      "step": 76935
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2880534653313428e-05,
+      "loss": 0.2439,
+      "step": 76940
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2877696201697107e-05,
+      "loss": 0.2857,
+      "step": 76945
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.28748577776375e-05,
+      "loss": 0.2752,
+      "step": 76950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.287201938117146e-05,
+      "loss": 0.2735,
+      "step": 76955
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2869181012335848e-05,
+      "loss": 0.2812,
+      "step": 76960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.286634267116752e-05,
+      "loss": 0.254,
+      "step": 76965
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2863504357703328e-05,
+      "loss": 0.2942,
+      "step": 76970
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.286066607198011e-05,
+      "loss": 0.2571,
+      "step": 76975
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2857827814034743e-05,
+      "loss": 0.2972,
+      "step": 76980
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2854989583904074e-05,
+      "loss": 0.2408,
+      "step": 76985
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2852151381624952e-05,
+      "loss": 0.261,
+      "step": 76990
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2849313207234217e-05,
+      "loss": 0.2507,
+      "step": 76995
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2846475060768743e-05,
+      "loss": 0.2842,
+      "step": 77000
+    },
+    {
+      "epoch": 2.71,
+      "eval_loss": 0.2610674202442169,
+      "eval_runtime": 10.5451,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 77000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2843636942265367e-05,
+      "loss": 0.2581,
+      "step": 77005
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.284079885176094e-05,
+      "loss": 0.2934,
+      "step": 77010
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2837960789292322e-05,
+      "loss": 0.2504,
+      "step": 77015
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2835122754896367e-05,
+      "loss": 0.2727,
+      "step": 77020
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.283228474860991e-05,
+      "loss": 0.2551,
+      "step": 77025
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2829446770469796e-05,
+      "loss": 0.2866,
+      "step": 77030
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.28266088205129e-05,
+      "loss": 0.2693,
+      "step": 77035
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2823770898776047e-05,
+      "loss": 0.2849,
+      "step": 77040
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.28209330052961e-05,
+      "loss": 0.2797,
+      "step": 77045
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2818095140109892e-05,
+      "loss": 0.2796,
+      "step": 77050
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2815257303254296e-05,
+      "loss": 0.2764,
+      "step": 77055
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.281241949476614e-05,
+      "loss": 0.2723,
+      "step": 77060
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2809581714682268e-05,
+      "loss": 0.2533,
+      "step": 77065
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.280674396303953e-05,
+      "loss": 0.2869,
+      "step": 77070
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.280390623987479e-05,
+      "loss": 0.2614,
+      "step": 77075
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2801068545224876e-05,
+      "loss": 0.2631,
+      "step": 77080
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2798230879126636e-05,
+      "loss": 0.2867,
+      "step": 77085
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2795393241616913e-05,
+      "loss": 0.2628,
+      "step": 77090
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.279255563273256e-05,
+      "loss": 0.2489,
+      "step": 77095
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2789718052510424e-05,
+      "loss": 0.2667,
+      "step": 77100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2786880500987336e-05,
+      "loss": 0.2879,
+      "step": 77105
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2784042978200142e-05,
+      "loss": 0.2719,
+      "step": 77110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2781205484185695e-05,
+      "loss": 0.2512,
+      "step": 77115
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2778368018980827e-05,
+      "loss": 0.2795,
+      "step": 77120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2775530582622385e-05,
+      "loss": 0.2752,
+      "step": 77125
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.277269317514722e-05,
+      "loss": 0.2545,
+      "step": 77130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.276985579659217e-05,
+      "loss": 0.2642,
+      "step": 77135
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.276701844699407e-05,
+      "loss": 0.2255,
+      "step": 77140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2764181126389753e-05,
+      "loss": 0.2544,
+      "step": 77145
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2761343834816072e-05,
+      "loss": 0.2894,
+      "step": 77150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2758506572309878e-05,
+      "loss": 0.2732,
+      "step": 77155
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.2755669338907997e-05,
+      "loss": 0.2722,
+      "step": 77160
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.275283213464726e-05,
+      "loss": 0.2762,
+      "step": 77165
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2749994959564528e-05,
+      "loss": 0.2824,
+      "step": 77170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.274715781369662e-05,
+      "loss": 0.2778,
+      "step": 77175
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2744320697080394e-05,
+      "loss": 0.2657,
+      "step": 77180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.274148360975266e-05,
+      "loss": 0.2686,
+      "step": 77185
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2738646551750288e-05,
+      "loss": 0.2729,
+      "step": 77190
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.27358095231101e-05,
+      "loss": 0.2537,
+      "step": 77195
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2732972523868926e-05,
+      "loss": 0.2587,
+      "step": 77200
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2730135554063607e-05,
+      "loss": 0.2661,
+      "step": 77205
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.272729861373099e-05,
+      "loss": 0.2629,
+      "step": 77210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2724461702907905e-05,
+      "loss": 0.2707,
+      "step": 77215
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2721624821631183e-05,
+      "loss": 0.2674,
+      "step": 77220
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.271878796993765e-05,
+      "loss": 0.2789,
+      "step": 77225
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.271595114786416e-05,
+      "loss": 0.2739,
+      "step": 77230
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.271311435544754e-05,
+      "loss": 0.2704,
+      "step": 77235
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2710277592724616e-05,
+      "loss": 0.2693,
+      "step": 77240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.270744085973224e-05,
+      "loss": 0.2677,
+      "step": 77245
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.270460415650723e-05,
+      "loss": 0.2619,
+      "step": 77250
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.270176748308642e-05,
+      "loss": 0.267,
+      "step": 77255
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.269893083950664e-05,
+      "loss": 0.2798,
+      "step": 77260
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2696094225804735e-05,
+      "loss": 0.2499,
+      "step": 77265
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.269325764201753e-05,
+      "loss": 0.2957,
+      "step": 77270
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2690421088181852e-05,
+      "loss": 0.2505,
+      "step": 77275
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2687584564334524e-05,
+      "loss": 0.2756,
+      "step": 77280
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.268474807051239e-05,
+      "loss": 0.2733,
+      "step": 77285
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2681911606752286e-05,
+      "loss": 0.2585,
+      "step": 77290
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.267907517309103e-05,
+      "loss": 0.279,
+      "step": 77295
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.267623876956544e-05,
+      "loss": 0.2715,
+      "step": 77300
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2673402396212374e-05,
+      "loss": 0.2629,
+      "step": 77305
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2670566053068637e-05,
+      "loss": 0.2589,
+      "step": 77310
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.266772974017107e-05,
+      "loss": 0.2688,
+      "step": 77315
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2664893457556486e-05,
+      "loss": 0.2875,
+      "step": 77320
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.266205720526173e-05,
+      "loss": 0.2793,
+      "step": 77325
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.265922098332362e-05,
+      "loss": 0.2463,
+      "step": 77330
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2656384791778977e-05,
+      "loss": 0.2618,
+      "step": 77335
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2653548630664632e-05,
+      "loss": 0.2857,
+      "step": 77340
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.265071250001742e-05,
+      "loss": 0.2459,
+      "step": 77345
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2647876399874156e-05,
+      "loss": 0.2834,
+      "step": 77350
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2645040330271658e-05,
+      "loss": 0.2687,
+      "step": 77355
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2642204291246772e-05,
+      "loss": 0.271,
+      "step": 77360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.26393682828363e-05,
+      "loss": 0.2495,
+      "step": 77365
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2636532305077083e-05,
+      "loss": 0.2624,
+      "step": 77370
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2633696358005928e-05,
+      "loss": 0.2692,
+      "step": 77375
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.263086044165967e-05,
+      "loss": 0.2721,
+      "step": 77380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2628024556075134e-05,
+      "loss": 0.2617,
+      "step": 77385
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.262518870128913e-05,
+      "loss": 0.246,
+      "step": 77390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2622352877338482e-05,
+      "loss": 0.274,
+      "step": 77395
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2619517084260018e-05,
+      "loss": 0.2585,
+      "step": 77400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.261668132209056e-05,
+      "loss": 0.2608,
+      "step": 77405
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2613845590866924e-05,
+      "loss": 0.2521,
+      "step": 77410
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2611009890625924e-05,
+      "loss": 0.2686,
+      "step": 77415
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2608174221404384e-05,
+      "loss": 0.2489,
+      "step": 77420
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2605338583239138e-05,
+      "loss": 0.2762,
+      "step": 77425
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.260250297616699e-05,
+      "loss": 0.249,
+      "step": 77430
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2599667400224748e-05,
+      "loss": 0.2607,
+      "step": 77435
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2596831855449255e-05,
+      "loss": 0.2503,
+      "step": 77440
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.259399634187731e-05,
+      "loss": 0.2347,
+      "step": 77445
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.2591160859545747e-05,
+      "loss": 0.2706,
+      "step": 77450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2588325408491356e-05,
+      "loss": 0.2669,
+      "step": 77455
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2585489988750984e-05,
+      "loss": 0.2725,
+      "step": 77460
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2582654600361433e-05,
+      "loss": 0.2924,
+      "step": 77465
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2579819243359507e-05,
+      "loss": 0.2495,
+      "step": 77470
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2576983917782046e-05,
+      "loss": 0.267,
+      "step": 77475
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2574148623665845e-05,
+      "loss": 0.2731,
+      "step": 77480
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2571313361047734e-05,
+      "loss": 0.2742,
+      "step": 77485
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2568478129964503e-05,
+      "loss": 0.2896,
+      "step": 77490
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.256564293045299e-05,
+      "loss": 0.2554,
+      "step": 77495
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2562807762550006e-05,
+      "loss": 0.2943,
+      "step": 77500
+    },
+    {
+      "epoch": 2.73,
+      "eval_loss": 0.2611740827560425,
+      "eval_runtime": 10.5438,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 77500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.255997262629235e-05,
+      "loss": 0.2799,
+      "step": 77505
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2557137521716834e-05,
+      "loss": 0.2513,
+      "step": 77510
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.255430244886029e-05,
+      "loss": 0.2622,
+      "step": 77515
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2551467407759515e-05,
+      "loss": 0.2908,
+      "step": 77520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2548632398451316e-05,
+      "loss": 0.2826,
+      "step": 77525
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2545797420972514e-05,
+      "loss": 0.2632,
+      "step": 77530
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.254296247535991e-05,
+      "loss": 0.2727,
+      "step": 77535
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2540127561650327e-05,
+      "loss": 0.2617,
+      "step": 77540
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2537292679880567e-05,
+      "loss": 0.2768,
+      "step": 77545
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2534457830087428e-05,
+      "loss": 0.2644,
+      "step": 77550
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.253162301230774e-05,
+      "loss": 0.2605,
+      "step": 77555
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2528788226578292e-05,
+      "loss": 0.2768,
+      "step": 77560
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.252595347293591e-05,
+      "loss": 0.2968,
+      "step": 77565
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.252311875141738e-05,
+      "loss": 0.2626,
+      "step": 77570
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2520284062059527e-05,
+      "loss": 0.2552,
+      "step": 77575
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2517449404899153e-05,
+      "loss": 0.2794,
+      "step": 77580
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2514614779973055e-05,
+      "loss": 0.2895,
+      "step": 77585
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.251178018731805e-05,
+      "loss": 0.2437,
+      "step": 77590
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2508945626970945e-05,
+      "loss": 0.2697,
+      "step": 77595
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2506111098968537e-05,
+      "loss": 0.259,
+      "step": 77600
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2503276603347626e-05,
+      "loss": 0.26,
+      "step": 77605
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.250044214014503e-05,
+      "loss": 0.261,
+      "step": 77610
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2497607709397543e-05,
+      "loss": 0.2581,
+      "step": 77615
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2494773311141974e-05,
+      "loss": 0.2515,
+      "step": 77620
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2491938945415114e-05,
+      "loss": 0.2459,
+      "step": 77625
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2489104612253784e-05,
+      "loss": 0.254,
+      "step": 77630
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2486270311694775e-05,
+      "loss": 0.2882,
+      "step": 77635
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2483436043774884e-05,
+      "loss": 0.274,
+      "step": 77640
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2480601808530914e-05,
+      "loss": 0.2703,
+      "step": 77645
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2477767605999676e-05,
+      "loss": 0.2312,
+      "step": 77650
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2474933436217964e-05,
+      "loss": 0.2711,
+      "step": 77655
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2472099299222583e-05,
+      "loss": 0.2628,
+      "step": 77660
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2469265195050308e-05,
+      "loss": 0.2846,
+      "step": 77665
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2466431123737968e-05,
+      "loss": 0.2754,
+      "step": 77670
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2463597085322352e-05,
+      "loss": 0.2636,
+      "step": 77675
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2460763079840257e-05,
+      "loss": 0.2799,
+      "step": 77680
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2457929107328473e-05,
+      "loss": 0.2776,
+      "step": 77685
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.245509516782381e-05,
+      "loss": 0.2927,
+      "step": 77690
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2452261261363056e-05,
+      "loss": 0.2628,
+      "step": 77695
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2449427387983004e-05,
+      "loss": 0.2494,
+      "step": 77700
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.244659354772047e-05,
+      "loss": 0.2591,
+      "step": 77705
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2443759740612234e-05,
+      "loss": 0.2602,
+      "step": 77710
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2440925966695093e-05,
+      "loss": 0.2782,
+      "step": 77715
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2438092226005832e-05,
+      "loss": 0.267,
+      "step": 77720
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.243525851858126e-05,
+      "loss": 0.2685,
+      "step": 77725
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2432424844458168e-05,
+      "loss": 0.2624,
+      "step": 77730
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.2429591203673353e-05,
+      "loss": 0.2586,
+      "step": 77735
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.242675759626359e-05,
+      "loss": 0.276,
+      "step": 77740
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.242392402226569e-05,
+      "loss": 0.2892,
+      "step": 77745
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.242109048171644e-05,
+      "loss": 0.2714,
+      "step": 77750
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.241825697465263e-05,
+      "loss": 0.2524,
+      "step": 77755
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2415423501111046e-05,
+      "loss": 0.2848,
+      "step": 77760
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.241259006112849e-05,
+      "loss": 0.2867,
+      "step": 77765
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2409756654741753e-05,
+      "loss": 0.2579,
+      "step": 77770
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.240692328198761e-05,
+      "loss": 0.2494,
+      "step": 77775
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2404089942902855e-05,
+      "loss": 0.2742,
+      "step": 77780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.240125663752429e-05,
+      "loss": 0.2523,
+      "step": 77785
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2398423365888697e-05,
+      "loss": 0.2723,
+      "step": 77790
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.239559012803286e-05,
+      "loss": 0.2681,
+      "step": 77795
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.239275692399356e-05,
+      "loss": 0.2579,
+      "step": 77800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.23899237538076e-05,
+      "loss": 0.2741,
+      "step": 77805
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2387090617511763e-05,
+      "loss": 0.2572,
+      "step": 77810
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.238425751514282e-05,
+      "loss": 0.2713,
+      "step": 77815
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2381424446737585e-05,
+      "loss": 0.2708,
+      "step": 77820
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.237859141233282e-05,
+      "loss": 0.2514,
+      "step": 77825
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.237575841196532e-05,
+      "loss": 0.2613,
+      "step": 77830
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2372925445671866e-05,
+      "loss": 0.2606,
+      "step": 77835
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.237009251348924e-05,
+      "loss": 0.2792,
+      "step": 77840
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.236725961545424e-05,
+      "loss": 0.2749,
+      "step": 77845
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2364426751603637e-05,
+      "loss": 0.2729,
+      "step": 77850
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.23615939219742e-05,
+      "loss": 0.2763,
+      "step": 77855
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.235876112660274e-05,
+      "loss": 0.2885,
+      "step": 77860
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.235592836552603e-05,
+      "loss": 0.2678,
+      "step": 77865
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2353095638780845e-05,
+      "loss": 0.288,
+      "step": 77870
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.235026294640396e-05,
+      "loss": 0.2635,
+      "step": 77875
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2347430288432172e-05,
+      "loss": 0.2696,
+      "step": 77880
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2344597664902254e-05,
+      "loss": 0.2782,
+      "step": 77885
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2341765075850985e-05,
+      "loss": 0.2874,
+      "step": 77890
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2338932521315136e-05,
+      "loss": 0.2768,
+      "step": 77895
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2336100001331506e-05,
+      "loss": 0.2596,
+      "step": 77900
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2333267515936863e-05,
+      "loss": 0.2539,
+      "step": 77905
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2330435065167975e-05,
+      "loss": 0.2689,
+      "step": 77910
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.232760264906163e-05,
+      "loss": 0.256,
+      "step": 77915
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2324770267654603e-05,
+      "loss": 0.2695,
+      "step": 77920
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2321937920983675e-05,
+      "loss": 0.2547,
+      "step": 77925
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2319105609085612e-05,
+      "loss": 0.2681,
+      "step": 77930
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2316273331997204e-05,
+      "loss": 0.2799,
+      "step": 77935
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2313441089755217e-05,
+      "loss": 0.2527,
+      "step": 77940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2310608882396425e-05,
+      "loss": 0.2718,
+      "step": 77945
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2307776709957596e-05,
+      "loss": 0.2551,
+      "step": 77950
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2304944572475523e-05,
+      "loss": 0.2872,
+      "step": 77955
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.230211246998697e-05,
+      "loss": 0.2838,
+      "step": 77960
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2299280402528702e-05,
+      "loss": 0.275,
+      "step": 77965
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2296448370137498e-05,
+      "loss": 0.26,
+      "step": 77970
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.229361637285013e-05,
+      "loss": 0.2651,
+      "step": 77975
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.229078441070338e-05,
+      "loss": 0.2758,
+      "step": 77980
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2287952483734004e-05,
+      "loss": 0.277,
+      "step": 77985
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2285120591978766e-05,
+      "loss": 0.2619,
+      "step": 77990
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.228228873547446e-05,
+      "loss": 0.2605,
+      "step": 77995
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.227945691425784e-05,
+      "loss": 0.2698,
+      "step": 78000
+    },
+    {
+      "epoch": 2.74,
+      "eval_loss": 0.26091867685317993,
+      "eval_runtime": 10.5455,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 78000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2276625128365683e-05,
+      "loss": 0.2536,
+      "step": 78005
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2273793377834744e-05,
+      "loss": 0.2584,
+      "step": 78010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.227096166270181e-05,
+      "loss": 0.251,
+      "step": 78015
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.2268129983003643e-05,
+      "loss": 0.2848,
+      "step": 78020
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2265298338777003e-05,
+      "loss": 0.2954,
+      "step": 78025
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2262466730058652e-05,
+      "loss": 0.2684,
+      "step": 78030
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2259635156885384e-05,
+      "loss": 0.2402,
+      "step": 78035
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.225680361929394e-05,
+      "loss": 0.2574,
+      "step": 78040
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2253972117321088e-05,
+      "loss": 0.2703,
+      "step": 78045
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2251140651003608e-05,
+      "loss": 0.2725,
+      "step": 78050
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.224830922037825e-05,
+      "loss": 0.2775,
+      "step": 78055
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2245477825481786e-05,
+      "loss": 0.2936,
+      "step": 78060
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2242646466350966e-05,
+      "loss": 0.2614,
+      "step": 78065
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.223981514302258e-05,
+      "loss": 0.2573,
+      "step": 78070
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.223698385553337e-05,
+      "loss": 0.2834,
+      "step": 78075
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2234152603920102e-05,
+      "loss": 0.266,
+      "step": 78080
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2231321388219532e-05,
+      "loss": 0.2543,
+      "step": 78085
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.222849020846844e-05,
+      "loss": 0.2598,
+      "step": 78090
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.222565906470358e-05,
+      "loss": 0.2721,
+      "step": 78095
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2222827956961707e-05,
+      "loss": 0.2614,
+      "step": 78100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.221999688527957e-05,
+      "loss": 0.2784,
+      "step": 78105
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2217165849693948e-05,
+      "loss": 0.2608,
+      "step": 78110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2214334850241598e-05,
+      "loss": 0.2884,
+      "step": 78115
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2211503886959272e-05,
+      "loss": 0.2564,
+      "step": 78120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2208672959883726e-05,
+      "loss": 0.2679,
+      "step": 78125
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2205842069051727e-05,
+      "loss": 0.2764,
+      "step": 78130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2203011214500026e-05,
+      "loss": 0.2784,
+      "step": 78135
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2200180396265384e-05,
+      "loss": 0.2653,
+      "step": 78140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2197349614384543e-05,
+      "loss": 0.2874,
+      "step": 78145
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2194518868894283e-05,
+      "loss": 0.2918,
+      "step": 78150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2191688159831346e-05,
+      "loss": 0.2566,
+      "step": 78155
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2188857487232477e-05,
+      "loss": 0.2439,
+      "step": 78160
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2186026851134445e-05,
+      "loss": 0.2777,
+      "step": 78165
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2183196251574003e-05,
+      "loss": 0.2715,
+      "step": 78170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2180365688587906e-05,
+      "loss": 0.2752,
+      "step": 78175
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.217753516221289e-05,
+      "loss": 0.2789,
+      "step": 78180
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.217470467248573e-05,
+      "loss": 0.2584,
+      "step": 78185
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2171874219443163e-05,
+      "loss": 0.2661,
+      "step": 78190
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.216904380312195e-05,
+      "loss": 0.2743,
+      "step": 78195
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.216621342355883e-05,
+      "loss": 0.2647,
+      "step": 78200
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2163383080790567e-05,
+      "loss": 0.2367,
+      "step": 78205
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.216055277485391e-05,
+      "loss": 0.2617,
+      "step": 78210
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2157722505785595e-05,
+      "loss": 0.2747,
+      "step": 78215
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2154892273622377e-05,
+      "loss": 0.2815,
+      "step": 78220
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2152062078401018e-05,
+      "loss": 0.2643,
+      "step": 78225
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2149231920158256e-05,
+      "loss": 0.2734,
+      "step": 78230
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.214640179893084e-05,
+      "loss": 0.2741,
+      "step": 78235
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.21435717147555e-05,
+      "loss": 0.29,
+      "step": 78240
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.214074166766901e-05,
+      "loss": 0.2804,
+      "step": 78245
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2137911657708107e-05,
+      "loss": 0.3018,
+      "step": 78250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2135081684909536e-05,
+      "loss": 0.2525,
+      "step": 78255
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2132251749310032e-05,
+      "loss": 0.2621,
+      "step": 78260
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2129421850946357e-05,
+      "loss": 0.2863,
+      "step": 78265
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2126591989855246e-05,
+      "loss": 0.2516,
+      "step": 78270
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2123762166073446e-05,
+      "loss": 0.2547,
+      "step": 78275
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.212093237963769e-05,
+      "loss": 0.2712,
+      "step": 78280
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2118102630584743e-05,
+      "loss": 0.2611,
+      "step": 78285
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2115272918951328e-05,
+      "loss": 0.2859,
+      "step": 78290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2112443244774188e-05,
+      "loss": 0.2507,
+      "step": 78295
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.210961360809007e-05,
+      "loss": 0.2666,
+      "step": 78300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.2106784008935723e-05,
+      "loss": 0.2525,
+      "step": 78305
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.210395444734788e-05,
+      "loss": 0.2667,
+      "step": 78310
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2101124923363267e-05,
+      "loss": 0.2904,
+      "step": 78315
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2098295437018648e-05,
+      "loss": 0.2897,
+      "step": 78320
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2095465988350745e-05,
+      "loss": 0.2825,
+      "step": 78325
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2092636577396306e-05,
+      "loss": 0.29,
+      "step": 78330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2089807204192056e-05,
+      "loss": 0.2703,
+      "step": 78335
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.208697786877475e-05,
+      "loss": 0.2609,
+      "step": 78340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.208414857118112e-05,
+      "loss": 0.2636,
+      "step": 78345
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.208131931144789e-05,
+      "loss": 0.289,
+      "step": 78350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2078490089611808e-05,
+      "loss": 0.2759,
+      "step": 78355
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2075660905709606e-05,
+      "loss": 0.2688,
+      "step": 78360
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2072831759778025e-05,
+      "loss": 0.2762,
+      "step": 78365
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2070002651853797e-05,
+      "loss": 0.2652,
+      "step": 78370
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.206717358197364e-05,
+      "loss": 0.2634,
+      "step": 78375
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.206434455017431e-05,
+      "loss": 0.2625,
+      "step": 78380
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.206151555649253e-05,
+      "loss": 0.2498,
+      "step": 78385
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.205868660096504e-05,
+      "loss": 0.2632,
+      "step": 78390
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2055857683628554e-05,
+      "loss": 0.263,
+      "step": 78395
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2053028804519825e-05,
+      "loss": 0.2532,
+      "step": 78400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2050199963675573e-05,
+      "loss": 0.2502,
+      "step": 78405
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2047371161132523e-05,
+      "loss": 0.2625,
+      "step": 78410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.204454239692742e-05,
+      "loss": 0.2671,
+      "step": 78415
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.204171367109699e-05,
+      "loss": 0.258,
+      "step": 78420
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.203888498367796e-05,
+      "loss": 0.2777,
+      "step": 78425
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2036056334707038e-05,
+      "loss": 0.269,
+      "step": 78430
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2033227724220988e-05,
+      "loss": 0.2586,
+      "step": 78435
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2030399152256516e-05,
+      "loss": 0.2547,
+      "step": 78440
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.202757061885036e-05,
+      "loss": 0.2568,
+      "step": 78445
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2024742124039228e-05,
+      "loss": 0.2524,
+      "step": 78450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.202191366785987e-05,
+      "loss": 0.2505,
+      "step": 78455
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2019085250349e-05,
+      "loss": 0.2746,
+      "step": 78460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.201625687154334e-05,
+      "loss": 0.2663,
+      "step": 78465
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.201342853147961e-05,
+      "loss": 0.2835,
+      "step": 78470
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2010600230194555e-05,
+      "loss": 0.2738,
+      "step": 78475
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2007771967724887e-05,
+      "loss": 0.2533,
+      "step": 78480
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.200494374410732e-05,
+      "loss": 0.257,
+      "step": 78485
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.2002115559378585e-05,
+      "loss": 0.2756,
+      "step": 78490
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1999287413575403e-05,
+      "loss": 0.2491,
+      "step": 78495
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1996459306734503e-05,
+      "loss": 0.2764,
+      "step": 78500
+    },
+    {
+      "epoch": 2.76,
+      "eval_loss": 0.26042640209198,
+      "eval_runtime": 10.5472,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 78500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1993631238892602e-05,
+      "loss": 0.2786,
+      "step": 78505
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.19908032100864e-05,
+      "loss": 0.2912,
+      "step": 78510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1987975220352655e-05,
+      "loss": 0.2569,
+      "step": 78515
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1985147269728056e-05,
+      "loss": 0.2628,
+      "step": 78520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.198231935824933e-05,
+      "loss": 0.2807,
+      "step": 78525
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1979491485953207e-05,
+      "loss": 0.2718,
+      "step": 78530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.19766636528764e-05,
+      "loss": 0.2648,
+      "step": 78535
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1973835859055616e-05,
+      "loss": 0.2811,
+      "step": 78540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1971008104527574e-05,
+      "loss": 0.2712,
+      "step": 78545
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1968180389329e-05,
+      "loss": 0.2597,
+      "step": 78550
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.196535271349661e-05,
+      "loss": 0.2871,
+      "step": 78555
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1962525077067117e-05,
+      "loss": 0.2908,
+      "step": 78560
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.195969748007722e-05,
+      "loss": 0.2651,
+      "step": 78565
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.195686992256366e-05,
+      "loss": 0.2564,
+      "step": 78570
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1954042404563134e-05,
+      "loss": 0.2597,
+      "step": 78575
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1951214926112366e-05,
+      "loss": 0.2854,
+      "step": 78580
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.1948387487248052e-05,
+      "loss": 0.2616,
+      "step": 78585
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1945560088006926e-05,
+      "loss": 0.2823,
+      "step": 78590
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1942732728425685e-05,
+      "loss": 0.261,
+      "step": 78595
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1939905408541046e-05,
+      "loss": 0.2808,
+      "step": 78600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.193707812838971e-05,
+      "loss": 0.2677,
+      "step": 78605
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1934250888008408e-05,
+      "loss": 0.2687,
+      "step": 78610
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1931423687433835e-05,
+      "loss": 0.2635,
+      "step": 78615
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1928596526702706e-05,
+      "loss": 0.2494,
+      "step": 78620
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1925769405851715e-05,
+      "loss": 0.2584,
+      "step": 78625
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1922942324917588e-05,
+      "loss": 0.2525,
+      "step": 78630
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1920115283937033e-05,
+      "loss": 0.2502,
+      "step": 78635
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.191728828294674e-05,
+      "loss": 0.2859,
+      "step": 78640
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1914461321983436e-05,
+      "loss": 0.2472,
+      "step": 78645
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.191163440108382e-05,
+      "loss": 0.2693,
+      "step": 78650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.190880752028459e-05,
+      "loss": 0.2925,
+      "step": 78655
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.190598067962245e-05,
+      "loss": 0.2807,
+      "step": 78660
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1903153879134126e-05,
+      "loss": 0.277,
+      "step": 78665
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1900327118856305e-05,
+      "loss": 0.2835,
+      "step": 78670
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1897500398825695e-05,
+      "loss": 0.2835,
+      "step": 78675
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.189467371907899e-05,
+      "loss": 0.2598,
+      "step": 78680
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1891847079652903e-05,
+      "loss": 0.2599,
+      "step": 78685
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1889020480584137e-05,
+      "loss": 0.2563,
+      "step": 78690
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1886193921909394e-05,
+      "loss": 0.2537,
+      "step": 78695
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1883367403665356e-05,
+      "loss": 0.2813,
+      "step": 78700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.188054092588875e-05,
+      "loss": 0.2766,
+      "step": 78705
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1877714488616262e-05,
+      "loss": 0.2852,
+      "step": 78710
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1874888091884593e-05,
+      "loss": 0.2666,
+      "step": 78715
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1872061735730436e-05,
+      "loss": 0.2682,
+      "step": 78720
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.186923542019051e-05,
+      "loss": 0.2746,
+      "step": 78725
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.186640914530149e-05,
+      "loss": 0.2646,
+      "step": 78730
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1863582911100083e-05,
+      "loss": 0.2447,
+      "step": 78735
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1860756717622978e-05,
+      "loss": 0.2704,
+      "step": 78740
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.185793056490689e-05,
+      "loss": 0.27,
+      "step": 78745
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1855104452988505e-05,
+      "loss": 0.2776,
+      "step": 78750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.18522783819045e-05,
+      "loss": 0.2765,
+      "step": 78755
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.18494523516916e-05,
+      "loss": 0.2562,
+      "step": 78760
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1846626362386478e-05,
+      "loss": 0.3022,
+      "step": 78765
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.184380041402584e-05,
+      "loss": 0.2736,
+      "step": 78770
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1840974506646363e-05,
+      "loss": 0.2568,
+      "step": 78775
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1838148640284763e-05,
+      "loss": 0.2483,
+      "step": 78780
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1835322814977718e-05,
+      "loss": 0.2659,
+      "step": 78785
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1832497030761916e-05,
+      "loss": 0.2842,
+      "step": 78790
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.182967128767405e-05,
+      "loss": 0.2506,
+      "step": 78795
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1826845585750817e-05,
+      "loss": 0.2736,
+      "step": 78800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1824019925028905e-05,
+      "loss": 0.2783,
+      "step": 78805
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1821194305545002e-05,
+      "loss": 0.2795,
+      "step": 78810
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1818368727335784e-05,
+      "loss": 0.2575,
+      "step": 78815
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1815543190437963e-05,
+      "loss": 0.2811,
+      "step": 78820
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1812717694888207e-05,
+      "loss": 0.2559,
+      "step": 78825
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1809892240723224e-05,
+      "loss": 0.2639,
+      "step": 78830
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.180706682797967e-05,
+      "loss": 0.2542,
+      "step": 78835
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1804241456694262e-05,
+      "loss": 0.2483,
+      "step": 78840
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1801416126903667e-05,
+      "loss": 0.2842,
+      "step": 78845
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1798590838644584e-05,
+      "loss": 0.2807,
+      "step": 78850
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1795765591953678e-05,
+      "loss": 0.2798,
+      "step": 78855
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.179294038686765e-05,
+      "loss": 0.2552,
+      "step": 78860
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.179011522342318e-05,
+      "loss": 0.268,
+      "step": 78865
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.1787290101656937e-05,
+      "loss": 0.2761,
+      "step": 78870
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1784465021605627e-05,
+      "loss": 0.276,
+      "step": 78875
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1781639983305917e-05,
+      "loss": 0.2701,
+      "step": 78880
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1778814986794492e-05,
+      "loss": 0.2942,
+      "step": 78885
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.177599003210802e-05,
+      "loss": 0.275,
+      "step": 78890
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1773165119283206e-05,
+      "loss": 0.2897,
+      "step": 78895
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1770340248356714e-05,
+      "loss": 0.285,
+      "step": 78900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.176751541936522e-05,
+      "loss": 0.2668,
+      "step": 78905
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1764690632345407e-05,
+      "loss": 0.2558,
+      "step": 78910
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1761865887333962e-05,
+      "loss": 0.2576,
+      "step": 78915
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1759041184367558e-05,
+      "loss": 0.29,
+      "step": 78920
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.175621652348286e-05,
+      "loss": 0.2788,
+      "step": 78925
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.175339190471655e-05,
+      "loss": 0.2555,
+      "step": 78930
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1750567328105308e-05,
+      "loss": 0.2706,
+      "step": 78935
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1747742793685816e-05,
+      "loss": 0.2622,
+      "step": 78940
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.174491830149474e-05,
+      "loss": 0.2678,
+      "step": 78945
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.174209385156874e-05,
+      "loss": 0.2656,
+      "step": 78950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1739269443944515e-05,
+      "loss": 0.2589,
+      "step": 78955
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1736445078658726e-05,
+      "loss": 0.2621,
+      "step": 78960
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1733620755748054e-05,
+      "loss": 0.2753,
+      "step": 78965
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1730796475249148e-05,
+      "loss": 0.2798,
+      "step": 78970
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1727972237198708e-05,
+      "loss": 0.274,
+      "step": 78975
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1725148041633388e-05,
+      "loss": 0.2559,
+      "step": 78980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1722323888589857e-05,
+      "loss": 0.279,
+      "step": 78985
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1719499778104796e-05,
+      "loss": 0.2681,
+      "step": 78990
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.171667571021487e-05,
+      "loss": 0.2853,
+      "step": 78995
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.171385168495675e-05,
+      "loss": 0.258,
+      "step": 79000
+    },
+    {
+      "epoch": 2.78,
+      "eval_loss": 0.260810911655426,
+      "eval_runtime": 10.5622,
+      "eval_samples_per_second": 9.468,
+      "eval_steps_per_second": 9.468,
+      "step": 79000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1711027702367086e-05,
+      "loss": 0.2571,
+      "step": 79005
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1708203762482574e-05,
+      "loss": 0.2718,
+      "step": 79010
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.170537986533986e-05,
+      "loss": 0.2733,
+      "step": 79015
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1702556010975623e-05,
+      "loss": 0.2939,
+      "step": 79020
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.169973219942651e-05,
+      "loss": 0.2649,
+      "step": 79025
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1696908430729214e-05,
+      "loss": 0.2827,
+      "step": 79030
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.169408470492038e-05,
+      "loss": 0.286,
+      "step": 79035
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1691261022036674e-05,
+      "loss": 0.2421,
+      "step": 79040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.168843738211476e-05,
+      "loss": 0.2687,
+      "step": 79045
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1685613785191312e-05,
+      "loss": 0.2901,
+      "step": 79050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1682790231302988e-05,
+      "loss": 0.2639,
+      "step": 79055
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1679966720486445e-05,
+      "loss": 0.2533,
+      "step": 79060
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1677143252778335e-05,
+      "loss": 0.2734,
+      "step": 79065
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1674319828215337e-05,
+      "loss": 0.2918,
+      "step": 79070
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1671496446834108e-05,
+      "loss": 0.2567,
+      "step": 79075
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1668673108671306e-05,
+      "loss": 0.2541,
+      "step": 79080
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1665849813763576e-05,
+      "loss": 0.2684,
+      "step": 79085
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.16630265621476e-05,
+      "loss": 0.2823,
+      "step": 79090
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.166020335386002e-05,
+      "loss": 0.2759,
+      "step": 79095
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1657380188937494e-05,
+      "loss": 0.284,
+      "step": 79100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.165455706741669e-05,
+      "loss": 0.2545,
+      "step": 79105
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.165173398933426e-05,
+      "loss": 0.2777,
+      "step": 79110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.164891095472686e-05,
+      "loss": 0.2842,
+      "step": 79115
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.164608796363113e-05,
+      "loss": 0.2896,
+      "step": 79120
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.164326501608374e-05,
+      "loss": 0.2724,
+      "step": 79125
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.164044211212135e-05,
+      "loss": 0.2505,
+      "step": 79130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1637619251780604e-05,
+      "loss": 0.2728,
+      "step": 79135
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.163479643509815e-05,
+      "loss": 0.2859,
+      "step": 79140
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.163197366211065e-05,
+      "loss": 0.2734,
+      "step": 79145
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1629150932854752e-05,
+      "loss": 0.2739,
+      "step": 79150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1626328247367112e-05,
+      "loss": 0.2747,
+      "step": 79155
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1623505605684364e-05,
+      "loss": 0.2626,
+      "step": 79160
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1620683007843184e-05,
+      "loss": 0.2693,
+      "step": 79165
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1617860453880205e-05,
+      "loss": 0.2699,
+      "step": 79170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1615037943832076e-05,
+      "loss": 0.2775,
+      "step": 79175
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1612215477735443e-05,
+      "loss": 0.2562,
+      "step": 79180
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.160939305562697e-05,
+      "loss": 0.277,
+      "step": 79185
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1606570677543293e-05,
+      "loss": 0.2747,
+      "step": 79190
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1603748343521062e-05,
+      "loss": 0.2458,
+      "step": 79195
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1600926053596904e-05,
+      "loss": 0.2618,
+      "step": 79200
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.159810380780749e-05,
+      "loss": 0.2707,
+      "step": 79205
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1595281606189464e-05,
+      "loss": 0.2677,
+      "step": 79210
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1592459448779446e-05,
+      "loss": 0.2631,
+      "step": 79215
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1589637335614114e-05,
+      "loss": 0.2719,
+      "step": 79220
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1586815266730088e-05,
+      "loss": 0.2509,
+      "step": 79225
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1583993242164012e-05,
+      "loss": 0.2838,
+      "step": 79230
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1581171261952535e-05,
+      "loss": 0.2725,
+      "step": 79235
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1578349326132293e-05,
+      "loss": 0.2576,
+      "step": 79240
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.157552743473994e-05,
+      "loss": 0.2681,
+      "step": 79245
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.15727055878121e-05,
+      "loss": 0.2652,
+      "step": 79250
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1569883785385413e-05,
+      "loss": 0.2611,
+      "step": 79255
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1567062027496538e-05,
+      "loss": 0.2867,
+      "step": 79260
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.156424031418209e-05,
+      "loss": 0.2822,
+      "step": 79265
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1561418645478722e-05,
+      "loss": 0.2569,
+      "step": 79270
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.155859702142306e-05,
+      "loss": 0.2762,
+      "step": 79275
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1555775442051756e-05,
+      "loss": 0.2956,
+      "step": 79280
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1552953907401433e-05,
+      "loss": 0.2615,
+      "step": 79285
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1550132417508736e-05,
+      "loss": 0.2734,
+      "step": 79290
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1547310972410287e-05,
+      "loss": 0.2429,
+      "step": 79295
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.154448957214274e-05,
+      "loss": 0.2657,
+      "step": 79300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.154166821674272e-05,
+      "loss": 0.2652,
+      "step": 79305
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1538846906246853e-05,
+      "loss": 0.2836,
+      "step": 79310
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1536025640691774e-05,
+      "loss": 0.2665,
+      "step": 79315
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1533204420114122e-05,
+      "loss": 0.2522,
+      "step": 79320
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1530383244550534e-05,
+      "loss": 0.2672,
+      "step": 79325
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.152756211403762e-05,
+      "loss": 0.2577,
+      "step": 79330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1524741028612034e-05,
+      "loss": 0.2838,
+      "step": 79335
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1521919988310392e-05,
+      "loss": 0.287,
+      "step": 79340
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.151909899316933e-05,
+      "loss": 0.2482,
+      "step": 79345
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.151627804322546e-05,
+      "loss": 0.2592,
+      "step": 79350
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.151345713851544e-05,
+      "loss": 0.2645,
+      "step": 79355
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1510636279075878e-05,
+      "loss": 0.2633,
+      "step": 79360
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1507815464943397e-05,
+      "loss": 0.2484,
+      "step": 79365
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1504994696154633e-05,
+      "loss": 0.2675,
+      "step": 79370
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1502173972746215e-05,
+      "loss": 0.2762,
+      "step": 79375
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.149935329475476e-05,
+      "loss": 0.2842,
+      "step": 79380
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.14965326622169e-05,
+      "loss": 0.2683,
+      "step": 79385
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.149371207516924e-05,
+      "loss": 0.2686,
+      "step": 79390
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1490891533648436e-05,
+      "loss": 0.2529,
+      "step": 79395
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.148807103769108e-05,
+      "loss": 0.268,
+      "step": 79400
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1485250587333817e-05,
+      "loss": 0.2525,
+      "step": 79405
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1482430182613248e-05,
+      "loss": 0.278,
+      "step": 79410
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1479609823566013e-05,
+      "loss": 0.2466,
+      "step": 79415
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1476789510228725e-05,
+      "loss": 0.2907,
+      "step": 79420
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1473969242638e-05,
+      "loss": 0.2831,
+      "step": 79425
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1471149020830454e-05,
+      "loss": 0.2742,
+      "step": 79430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1468328844842722e-05,
+      "loss": 0.2506,
+      "step": 79435
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1465508714711413e-05,
+      "loss": 0.2554,
+      "step": 79440
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.146268863047314e-05,
+      "loss": 0.2887,
+      "step": 79445
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.145986859216452e-05,
+      "loss": 0.264,
+      "step": 79450
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.145704859982218e-05,
+      "loss": 0.2591,
+      "step": 79455
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.145422865348273e-05,
+      "loss": 0.2317,
+      "step": 79460
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1451408753182772e-05,
+      "loss": 0.2664,
+      "step": 79465
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1448588898958947e-05,
+      "loss": 0.2619,
+      "step": 79470
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1445769090847854e-05,
+      "loss": 0.2431,
+      "step": 79475
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.14429493288861e-05,
+      "loss": 0.249,
+      "step": 79480
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1440129613110302e-05,
+      "loss": 0.2691,
+      "step": 79485
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.143730994355708e-05,
+      "loss": 0.2644,
+      "step": 79490
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1434490320263048e-05,
+      "loss": 0.2756,
+      "step": 79495
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1431670743264806e-05,
+      "loss": 0.2753,
+      "step": 79500
+    },
+    {
+      "epoch": 2.8,
+      "eval_loss": 0.2603822350502014,
+      "eval_runtime": 10.5518,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 79500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1428851212598958e-05,
+      "loss": 0.2806,
+      "step": 79505
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1426031728302126e-05,
+      "loss": 0.2657,
+      "step": 79510
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1423212290410922e-05,
+      "loss": 0.2625,
+      "step": 79515
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1420392898961956e-05,
+      "loss": 0.2688,
+      "step": 79520
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.141757355399181e-05,
+      "loss": 0.2591,
+      "step": 79525
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1414754255537126e-05,
+      "loss": 0.2542,
+      "step": 79530
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.141193500363449e-05,
+      "loss": 0.2542,
+      "step": 79535
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.140911579832052e-05,
+      "loss": 0.2587,
+      "step": 79540
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1406296639631798e-05,
+      "loss": 0.2797,
+      "step": 79545
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.140347752760496e-05,
+      "loss": 0.2718,
+      "step": 79550
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1400658462276595e-05,
+      "loss": 0.278,
+      "step": 79555
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1397839443683305e-05,
+      "loss": 0.2467,
+      "step": 79560
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1395020471861688e-05,
+      "loss": 0.2569,
+      "step": 79565
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1392201546848368e-05,
+      "loss": 0.2563,
+      "step": 79570
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.138938266867993e-05,
+      "loss": 0.2755,
+      "step": 79575
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.138656383739297e-05,
+      "loss": 0.2489,
+      "step": 79580
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1383745053024106e-05,
+      "loss": 0.2614,
+      "step": 79585
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.138092631560993e-05,
+      "loss": 0.2598,
+      "step": 79590
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.137810762518704e-05,
+      "loss": 0.2725,
+      "step": 79595
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1375288981792026e-05,
+      "loss": 0.2691,
+      "step": 79600
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1372470385461506e-05,
+      "loss": 0.2584,
+      "step": 79605
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.136965183623207e-05,
+      "loss": 0.2645,
+      "step": 79610
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1366833334140302e-05,
+      "loss": 0.2702,
+      "step": 79615
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1364014879222807e-05,
+      "loss": 0.2578,
+      "step": 79620
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.136119647151619e-05,
+      "loss": 0.2612,
+      "step": 79625
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1358378111057046e-05,
+      "loss": 0.2815,
+      "step": 79630
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1355559797881957e-05,
+      "loss": 0.2783,
+      "step": 79635
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1353305181410992e-05,
+      "loss": 0.2755,
+      "step": 79640
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1350486953439432e-05,
+      "loss": 0.2867,
+      "step": 79645
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.134766877285438e-05,
+      "loss": 0.2578,
+      "step": 79650
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1344850639692447e-05,
+      "loss": 0.2673,
+      "step": 79655
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1342032553990228e-05,
+      "loss": 0.2736,
+      "step": 79660
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.13392145157843e-05,
+      "loss": 0.2522,
+      "step": 79665
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1336396525111252e-05,
+      "loss": 0.266,
+      "step": 79670
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.133357858200769e-05,
+      "loss": 0.2527,
+      "step": 79675
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1330760686510188e-05,
+      "loss": 0.2538,
+      "step": 79680
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.132794283865535e-05,
+      "loss": 0.2634,
+      "step": 79685
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1325125038479736e-05,
+      "loss": 0.2837,
+      "step": 79690
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.132230728601997e-05,
+      "loss": 0.2694,
+      "step": 79695
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1319489581312614e-05,
+      "loss": 0.2669,
+      "step": 79700
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.131667192439426e-05,
+      "loss": 0.2794,
+      "step": 79705
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1313854315301486e-05,
+      "loss": 0.2697,
+      "step": 79710
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1311036754070897e-05,
+      "loss": 0.2478,
+      "step": 79715
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.130821924073906e-05,
+      "loss": 0.2492,
+      "step": 79720
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1305401775342566e-05,
+      "loss": 0.2565,
+      "step": 79725
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1302584357917985e-05,
+      "loss": 0.2698,
+      "step": 79730
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1299766988501913e-05,
+      "loss": 0.2696,
+      "step": 79735
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1296949667130935e-05,
+      "loss": 0.278,
+      "step": 79740
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1294132393841627e-05,
+      "loss": 0.2425,
+      "step": 79745
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1291315168670553e-05,
+      "loss": 0.2866,
+      "step": 79750
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1288497991654322e-05,
+      "loss": 0.2734,
+      "step": 79755
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.128568086282949e-05,
+      "loss": 0.2391,
+      "step": 79760
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.128286378223264e-05,
+      "loss": 0.2667,
+      "step": 79765
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1280046749900364e-05,
+      "loss": 0.2573,
+      "step": 79770
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127722976586923e-05,
+      "loss": 0.2598,
+      "step": 79775
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1274412830175813e-05,
+      "loss": 0.264,
+      "step": 79780
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.127159594285668e-05,
+      "loss": 0.2813,
+      "step": 79785
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1268779103948423e-05,
+      "loss": 0.2484,
+      "step": 79790
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1265962313487618e-05,
+      "loss": 0.2816,
+      "step": 79795
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.126314557151083e-05,
+      "loss": 0.2595,
+      "step": 79800
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1260328878054618e-05,
+      "loss": 0.2617,
+      "step": 79805
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1257512233155587e-05,
+      "loss": 0.2855,
+      "step": 79810
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1254695636850288e-05,
+      "loss": 0.2838,
+      "step": 79815
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.12518790891753e-05,
+      "loss": 0.2773,
+      "step": 79820
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1249062590167186e-05,
+      "loss": 0.2413,
+      "step": 79825
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.124624613986253e-05,
+      "loss": 0.2865,
+      "step": 79830
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1243429738297895e-05,
+      "loss": 0.2846,
+      "step": 79835
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.124061338550984e-05,
+      "loss": 0.2647,
+      "step": 79840
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1237797081534948e-05,
+      "loss": 0.2382,
+      "step": 79845
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.123498082640978e-05,
+      "loss": 0.2995,
+      "step": 79850
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1232164620170908e-05,
+      "loss": 0.2746,
+      "step": 79855
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1229348462854898e-05,
+      "loss": 0.2655,
+      "step": 79860
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1226532354498303e-05,
+      "loss": 0.2675,
+      "step": 79865
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.12237162951377e-05,
+      "loss": 0.2652,
+      "step": 79870
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1220900284809657e-05,
+      "loss": 0.2666,
+      "step": 79875
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1218084323550728e-05,
+      "loss": 0.2566,
+      "step": 79880
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.121526841139749e-05,
+      "loss": 0.2406,
+      "step": 79885
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1212452548386497e-05,
+      "loss": 0.2773,
+      "step": 79890
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1209636734554307e-05,
+      "loss": 0.2602,
+      "step": 79895
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1206820969937485e-05,
+      "loss": 0.2835,
+      "step": 79900
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1204005254572594e-05,
+      "loss": 0.2683,
+      "step": 79905
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1201189588496197e-05,
+      "loss": 0.2697,
+      "step": 79910
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.119837397174485e-05,
+      "loss": 0.2709,
+      "step": 79915
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1195558404355104e-05,
+      "loss": 0.2803,
+      "step": 79920
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1192742886363532e-05,
+      "loss": 0.2833,
+      "step": 79925
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1189927417806678e-05,
+      "loss": 0.2837,
+      "step": 79930
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1187111998721114e-05,
+      "loss": 0.2748,
+      "step": 79935
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1184296629143377e-05,
+      "loss": 0.2607,
+      "step": 79940
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1181481309110043e-05,
+      "loss": 0.2764,
+      "step": 79945
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1178666038657653e-05,
+      "loss": 0.2747,
+      "step": 79950
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1175850817822774e-05,
+      "loss": 0.3014,
+      "step": 79955
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1173035646641938e-05,
+      "loss": 0.2642,
+      "step": 79960
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1170220525151724e-05,
+      "loss": 0.2579,
+      "step": 79965
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1167405453388673e-05,
+      "loss": 0.2858,
+      "step": 79970
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1164590431389326e-05,
+      "loss": 0.268,
+      "step": 79975
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.116177545919025e-05,
+      "loss": 0.2737,
+      "step": 79980
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.115896053682799e-05,
+      "loss": 0.2739,
+      "step": 79985
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.11561456643391e-05,
+      "loss": 0.2873,
+      "step": 79990
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1153330841760115e-05,
+      "loss": 0.2966,
+      "step": 79995
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1150516069127603e-05,
+      "loss": 0.2477,
+      "step": 80000
+    },
+    {
+      "epoch": 2.81,
+      "eval_loss": 0.26057764887809753,
+      "eval_runtime": 10.5527,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 80000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.114770134647811e-05,
+      "loss": 0.261,
+      "step": 80005
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1144886673848164e-05,
+      "loss": 0.2834,
+      "step": 80010
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.114207205127432e-05,
+      "loss": 0.2665,
+      "step": 80015
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1139257478793137e-05,
+      "loss": 0.2911,
+      "step": 80020
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1136442956441155e-05,
+      "loss": 0.2903,
+      "step": 80025
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1133628484254906e-05,
+      "loss": 0.2478,
+      "step": 80030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1130814062270944e-05,
+      "loss": 0.2896,
+      "step": 80035
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.112799969052581e-05,
+      "loss": 0.2863,
+      "step": 80040
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.112518536905606e-05,
+      "loss": 0.2933,
+      "step": 80045
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.112237109789822e-05,
+      "loss": 0.2573,
+      "step": 80050
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1119556877088824e-05,
+      "loss": 0.2869,
+      "step": 80055
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.111674270666444e-05,
+      "loss": 0.2613,
+      "step": 80060
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1113928586661583e-05,
+      "loss": 0.2761,
+      "step": 80065
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.111111451711681e-05,
+      "loss": 0.2667,
+      "step": 80070
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1108300498066638e-05,
+      "loss": 0.267,
+      "step": 80075
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1105486529547633e-05,
+      "loss": 0.2622,
+      "step": 80080
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1102672611596318e-05,
+      "loss": 0.2929,
+      "step": 80085
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1099858744249226e-05,
+      "loss": 0.252,
+      "step": 80090
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1097044927542892e-05,
+      "loss": 0.2905,
+      "step": 80095
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.109423116151387e-05,
+      "loss": 0.2739,
+      "step": 80100
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.109141744619868e-05,
+      "loss": 0.2775,
+      "step": 80105
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1088603781633847e-05,
+      "loss": 0.2554,
+      "step": 80110
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.108579016785593e-05,
+      "loss": 0.2898,
+      "step": 80115
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.108297660490144e-05,
+      "loss": 0.2458,
+      "step": 80120
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1080163092806927e-05,
+      "loss": 0.2766,
+      "step": 80125
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1077349631608894e-05,
+      "loss": 0.2797,
+      "step": 80130
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.107453622134391e-05,
+      "loss": 0.2439,
+      "step": 80135
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.107172286204848e-05,
+      "loss": 0.2781,
+      "step": 80140
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1068909553759138e-05,
+      "loss": 0.2773,
+      "step": 80145
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.106609629651241e-05,
+      "loss": 0.2765,
+      "step": 80150
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.106328309034484e-05,
+      "loss": 0.2547,
+      "step": 80155
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.106046993529294e-05,
+      "loss": 0.2779,
+      "step": 80160
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1057656831393246e-05,
+      "loss": 0.2747,
+      "step": 80165
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1054843778682272e-05,
+      "loss": 0.2883,
+      "step": 80170
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.105203077719655e-05,
+      "loss": 0.2727,
+      "step": 80175
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1049217826972617e-05,
+      "loss": 0.2562,
+      "step": 80180
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1046404928046983e-05,
+      "loss": 0.2653,
+      "step": 80185
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1043592080456168e-05,
+      "loss": 0.2759,
+      "step": 80190
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.104077928423671e-05,
+      "loss": 0.2759,
+      "step": 80195
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.103796653942512e-05,
+      "loss": 0.2461,
+      "step": 80200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.103515384605793e-05,
+      "loss": 0.262,
+      "step": 80205
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1032341204171645e-05,
+      "loss": 0.2633,
+      "step": 80210
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.10295286138028e-05,
+      "loss": 0.275,
+      "step": 80215
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1026716074987912e-05,
+      "loss": 0.271,
+      "step": 80220
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.102390358776349e-05,
+      "loss": 0.2527,
+      "step": 80225
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.102109115216606e-05,
+      "loss": 0.2725,
+      "step": 80230
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1018278768232146e-05,
+      "loss": 0.2702,
+      "step": 80235
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.101546643599826e-05,
+      "loss": 0.2668,
+      "step": 80240
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1012654155500904e-05,
+      "loss": 0.2583,
+      "step": 80245
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1009841926776616e-05,
+      "loss": 0.2684,
+      "step": 80250
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1007029749861897e-05,
+      "loss": 0.2692,
+      "step": 80255
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1004217624793272e-05,
+      "loss": 0.2727,
+      "step": 80260
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.1001405551607234e-05,
+      "loss": 0.2685,
+      "step": 80265
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0998593530340322e-05,
+      "loss": 0.2716,
+      "step": 80270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0995781561029035e-05,
+      "loss": 0.2605,
+      "step": 80275
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.099296964370988e-05,
+      "loss": 0.2623,
+      "step": 80280
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0990157778419375e-05,
+      "loss": 0.2637,
+      "step": 80285
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.0987345965194028e-05,
+      "loss": 0.2739,
+      "step": 80290
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0984534204070354e-05,
+      "loss": 0.2722,
+      "step": 80295
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.098172249508486e-05,
+      "loss": 0.2858,
+      "step": 80300
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0978910838274038e-05,
+      "loss": 0.262,
+      "step": 80305
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0976099233674413e-05,
+      "loss": 0.2545,
+      "step": 80310
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0973287681322497e-05,
+      "loss": 0.2886,
+      "step": 80315
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0970476181254784e-05,
+      "loss": 0.2829,
+      "step": 80320
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0967664733507768e-05,
+      "loss": 0.2694,
+      "step": 80325
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0964853338117983e-05,
+      "loss": 0.2816,
+      "step": 80330
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0962041995121912e-05,
+      "loss": 0.28,
+      "step": 80335
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0959230704556066e-05,
+      "loss": 0.2756,
+      "step": 80340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0956419466456945e-05,
+      "loss": 0.2736,
+      "step": 80345
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.095360828086106e-05,
+      "loss": 0.2538,
+      "step": 80350
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0950797147804903e-05,
+      "loss": 0.2854,
+      "step": 80355
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0947986067324962e-05,
+      "loss": 0.2723,
+      "step": 80360
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0945175039457765e-05,
+      "loss": 0.2605,
+      "step": 80365
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0942364064239795e-05,
+      "loss": 0.2816,
+      "step": 80370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.093955314170756e-05,
+      "loss": 0.27,
+      "step": 80375
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0936742271897536e-05,
+      "loss": 0.2619,
+      "step": 80380
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0933931454846245e-05,
+      "loss": 0.2715,
+      "step": 80385
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.093112069059017e-05,
+      "loss": 0.2664,
+      "step": 80390
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0928309979165824e-05,
+      "loss": 0.2681,
+      "step": 80395
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0925499320609674e-05,
+      "loss": 0.2559,
+      "step": 80400
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.092268871495824e-05,
+      "loss": 0.2453,
+      "step": 80405
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0919878162248008e-05,
+      "loss": 0.2544,
+      "step": 80410
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.091706766251546e-05,
+      "loss": 0.2612,
+      "step": 80415
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.09142572157971e-05,
+      "loss": 0.2587,
+      "step": 80420
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.091144682212942e-05,
+      "loss": 0.2638,
+      "step": 80425
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.090863648154891e-05,
+      "loss": 0.2679,
+      "step": 80430
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0905826194092064e-05,
+      "loss": 0.2748,
+      "step": 80435
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.090301595979535e-05,
+      "loss": 0.2592,
+      "step": 80440
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0900205778695285e-05,
+      "loss": 0.2609,
+      "step": 80445
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0897395650828343e-05,
+      "loss": 0.2677,
+      "step": 80450
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0894585576231012e-05,
+      "loss": 0.2443,
+      "step": 80455
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0891775554939792e-05,
+      "loss": 0.2822,
+      "step": 80460
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0888965586991156e-05,
+      "loss": 0.2614,
+      "step": 80465
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0886155672421588e-05,
+      "loss": 0.2685,
+      "step": 80470
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0883345811267574e-05,
+      "loss": 0.2603,
+      "step": 80475
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0880536003565605e-05,
+      "loss": 0.2708,
+      "step": 80480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.087772624935217e-05,
+      "loss": 0.2762,
+      "step": 80485
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0874916548663738e-05,
+      "loss": 0.2803,
+      "step": 80490
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0872106901536786e-05,
+      "loss": 0.2708,
+      "step": 80495
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.086929730800782e-05,
+      "loss": 0.2658,
+      "step": 80500
+    },
+    {
+      "epoch": 2.83,
+      "eval_loss": 0.26011475920677185,
+      "eval_runtime": 10.5627,
+      "eval_samples_per_second": 9.467,
+      "eval_steps_per_second": 9.467,
+      "step": 80500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0866487768113297e-05,
+      "loss": 0.2571,
+      "step": 80505
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0863678281889713e-05,
+      "loss": 0.2468,
+      "step": 80510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0860868849373528e-05,
+      "loss": 0.2418,
+      "step": 80515
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0858059470601243e-05,
+      "loss": 0.242,
+      "step": 80520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.085525014560933e-05,
+      "loss": 0.2763,
+      "step": 80525
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0852440874434256e-05,
+      "loss": 0.2448,
+      "step": 80530
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0849631657112496e-05,
+      "loss": 0.2852,
+      "step": 80535
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0846822493680548e-05,
+      "loss": 0.2857,
+      "step": 80540
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.084401338417487e-05,
+      "loss": 0.3082,
+      "step": 80545
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0841204328631942e-05,
+      "loss": 0.2949,
+      "step": 80550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0838395327088223e-05,
+      "loss": 0.2487,
+      "step": 80555
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0835586379580203e-05,
+      "loss": 0.2753,
+      "step": 80560
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0832777486144357e-05,
+      "loss": 0.275,
+      "step": 80565
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0829968646817135e-05,
+      "loss": 0.277,
+      "step": 80570
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.0827159861635034e-05,
+      "loss": 0.2595,
+      "step": 80575
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0824351130634513e-05,
+      "loss": 0.2627,
+      "step": 80580
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0821542453852032e-05,
+      "loss": 0.2589,
+      "step": 80585
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.081873383132407e-05,
+      "loss": 0.2754,
+      "step": 80590
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.08159252630871e-05,
+      "loss": 0.2601,
+      "step": 80595
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.081311674917758e-05,
+      "loss": 0.2757,
+      "step": 80600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0810308289631984e-05,
+      "loss": 0.2631,
+      "step": 80605
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0807499884486763e-05,
+      "loss": 0.2876,
+      "step": 80610
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0804691533778398e-05,
+      "loss": 0.2869,
+      "step": 80615
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0801883237543355e-05,
+      "loss": 0.2653,
+      "step": 80620
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0799074995818087e-05,
+      "loss": 0.2704,
+      "step": 80625
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0796266808639058e-05,
+      "loss": 0.275,
+      "step": 80630
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0793458676042737e-05,
+      "loss": 0.252,
+      "step": 80635
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.079065059806558e-05,
+      "loss": 0.2659,
+      "step": 80640
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0787842574744058e-05,
+      "loss": 0.265,
+      "step": 80645
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0785034606114613e-05,
+      "loss": 0.2435,
+      "step": 80650
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.078222669221373e-05,
+      "loss": 0.2861,
+      "step": 80655
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0779418833077848e-05,
+      "loss": 0.2312,
+      "step": 80660
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.077661102874343e-05,
+      "loss": 0.2818,
+      "step": 80665
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0773803279246924e-05,
+      "loss": 0.2593,
+      "step": 80670
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0770995584624815e-05,
+      "loss": 0.259,
+      "step": 80675
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0768187944913537e-05,
+      "loss": 0.2681,
+      "step": 80680
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.076538036014955e-05,
+      "loss": 0.2568,
+      "step": 80685
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.07625728303693e-05,
+      "loss": 0.2833,
+      "step": 80690
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.075976535560925e-05,
+      "loss": 0.2708,
+      "step": 80695
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.075695793590586e-05,
+      "loss": 0.2695,
+      "step": 80700
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0754150571295563e-05,
+      "loss": 0.2673,
+      "step": 80705
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0751343261814838e-05,
+      "loss": 0.2859,
+      "step": 80710
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0748536007500116e-05,
+      "loss": 0.2642,
+      "step": 80715
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.074572880838785e-05,
+      "loss": 0.2722,
+      "step": 80720
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0742921664514486e-05,
+      "loss": 0.2735,
+      "step": 80725
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0740114575916484e-05,
+      "loss": 0.2475,
+      "step": 80730
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.073730754263029e-05,
+      "loss": 0.2844,
+      "step": 80735
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.073450056469235e-05,
+      "loss": 0.2879,
+      "step": 80740
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.07316936421391e-05,
+      "loss": 0.2833,
+      "step": 80745
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0728886775006997e-05,
+      "loss": 0.2413,
+      "step": 80750
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.072607996333249e-05,
+      "loss": 0.2701,
+      "step": 80755
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.072327320715202e-05,
+      "loss": 0.2796,
+      "step": 80760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.072046650650202e-05,
+      "loss": 0.249,
+      "step": 80765
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.071765986141895e-05,
+      "loss": 0.2712,
+      "step": 80770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0714853271939235e-05,
+      "loss": 0.2685,
+      "step": 80775
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.071204673809934e-05,
+      "loss": 0.2792,
+      "step": 80780
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0709240259935676e-05,
+      "loss": 0.2435,
+      "step": 80785
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.070643383748471e-05,
+      "loss": 0.2449,
+      "step": 80790
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.070362747078287e-05,
+      "loss": 0.2518,
+      "step": 80795
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0700821159866597e-05,
+      "loss": 0.2533,
+      "step": 80800
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0698014904772323e-05,
+      "loss": 0.2424,
+      "step": 80805
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0695208705536493e-05,
+      "loss": 0.2789,
+      "step": 80810
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0692402562195544e-05,
+      "loss": 0.2933,
+      "step": 80815
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.06895964747859e-05,
+      "loss": 0.2417,
+      "step": 80820
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0686790443344013e-05,
+      "loss": 0.271,
+      "step": 80825
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0683984467906306e-05,
+      "loss": 0.2566,
+      "step": 80830
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0681178548509223e-05,
+      "loss": 0.2505,
+      "step": 80835
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0678372685189178e-05,
+      "loss": 0.2572,
+      "step": 80840
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0675566877982626e-05,
+      "loss": 0.2604,
+      "step": 80845
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.067276112692599e-05,
+      "loss": 0.2816,
+      "step": 80850
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.0669955432055695e-05,
+      "loss": 0.2688,
+      "step": 80855
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.066714979340817e-05,
+      "loss": 0.2857,
+      "step": 80860
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.066434421101985e-05,
+      "loss": 0.2708,
+      "step": 80865
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0661538684927175e-05,
+      "loss": 0.2721,
+      "step": 80870
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0658733215166555e-05,
+      "loss": 0.2632,
+      "step": 80875
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0655927801774412e-05,
+      "loss": 0.2533,
+      "step": 80880
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.06531224447872e-05,
+      "loss": 0.2479,
+      "step": 80885
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.065031714424132e-05,
+      "loss": 0.2733,
+      "step": 80890
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0647511900173208e-05,
+      "loss": 0.2774,
+      "step": 80895
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0644706712619276e-05,
+      "loss": 0.2727,
+      "step": 80900
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.064190158161597e-05,
+      "loss": 0.2752,
+      "step": 80905
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0639096507199693e-05,
+      "loss": 0.2747,
+      "step": 80910
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0636291489406878e-05,
+      "loss": 0.2662,
+      "step": 80915
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0633486528273934e-05,
+      "loss": 0.2834,
+      "step": 80920
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.06306816238373e-05,
+      "loss": 0.2694,
+      "step": 80925
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0627876776133386e-05,
+      "loss": 0.261,
+      "step": 80930
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0625071985198597e-05,
+      "loss": 0.2763,
+      "step": 80935
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.062226725106938e-05,
+      "loss": 0.2674,
+      "step": 80940
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0619462573782132e-05,
+      "loss": 0.2511,
+      "step": 80945
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.061665795337328e-05,
+      "loss": 0.2441,
+      "step": 80950
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0613853389879224e-05,
+      "loss": 0.2774,
+      "step": 80955
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.06110488833364e-05,
+      "loss": 0.2761,
+      "step": 80960
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0608244433781216e-05,
+      "loss": 0.2804,
+      "step": 80965
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0605440041250078e-05,
+      "loss": 0.2761,
+      "step": 80970
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0602635705779403e-05,
+      "loss": 0.2573,
+      "step": 80975
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0599831427405612e-05,
+      "loss": 0.2798,
+      "step": 80980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.059702720616511e-05,
+      "loss": 0.2702,
+      "step": 80985
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0594223042094308e-05,
+      "loss": 0.2639,
+      "step": 80990
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.059141893522961e-05,
+      "loss": 0.2854,
+      "step": 80995
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0588614885607434e-05,
+      "loss": 0.252,
+      "step": 81000
+    },
+    {
+      "epoch": 2.85,
+      "eval_loss": 0.26019665598869324,
+      "eval_runtime": 10.5615,
+      "eval_samples_per_second": 9.468,
+      "eval_steps_per_second": 9.468,
+      "step": 81000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.058581089326419e-05,
+      "loss": 0.284,
+      "step": 81005
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0583006958236283e-05,
+      "loss": 0.2698,
+      "step": 81010
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0580203080560112e-05,
+      "loss": 0.2365,
+      "step": 81015
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.05773992602721e-05,
+      "loss": 0.2765,
+      "step": 81020
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0574595497408637e-05,
+      "loss": 0.2547,
+      "step": 81025
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0571791792006147e-05,
+      "loss": 0.2475,
+      "step": 81030
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0568988144101005e-05,
+      "loss": 0.2706,
+      "step": 81035
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0566184553729645e-05,
+      "loss": 0.276,
+      "step": 81040
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0563381020928458e-05,
+      "loss": 0.2681,
+      "step": 81045
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.056057754573383e-05,
+      "loss": 0.2537,
+      "step": 81050
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0557774128182184e-05,
+      "loss": 0.2597,
+      "step": 81055
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.055497076830992e-05,
+      "loss": 0.2749,
+      "step": 81060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0552167466153428e-05,
+      "loss": 0.2832,
+      "step": 81065
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.05493642217491e-05,
+      "loss": 0.2556,
+      "step": 81070
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.054656103513336e-05,
+      "loss": 0.2678,
+      "step": 81075
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.054375790634258e-05,
+      "loss": 0.2551,
+      "step": 81080
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0540954835413176e-05,
+      "loss": 0.2738,
+      "step": 81085
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0538151822381522e-05,
+      "loss": 0.2672,
+      "step": 81090
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0535348867284036e-05,
+      "loss": 0.247,
+      "step": 81095
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.05325459701571e-05,
+      "loss": 0.2663,
+      "step": 81100
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.052974313103711e-05,
+      "loss": 0.2895,
+      "step": 81105
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0526940349960455e-05,
+      "loss": 0.3062,
+      "step": 81110
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.052413762696354e-05,
+      "loss": 0.2912,
+      "step": 81115
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0521334962082748e-05,
+      "loss": 0.2731,
+      "step": 81120
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.051853235535447e-05,
+      "loss": 0.2838,
+      "step": 81125
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0515729806815084e-05,
+      "loss": 0.2635,
+      "step": 81130
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.0512927316500995e-05,
+      "loss": 0.2871,
+      "step": 81135
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.05101248844486e-05,
+      "loss": 0.2514,
+      "step": 81140
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.050732251069427e-05,
+      "loss": 0.2808,
+      "step": 81145
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0504520195274384e-05,
+      "loss": 0.2765,
+      "step": 81150
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0501717938225355e-05,
+      "loss": 0.2493,
+      "step": 81155
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0498915739583547e-05,
+      "loss": 0.2634,
+      "step": 81160
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0496113599385353e-05,
+      "loss": 0.2576,
+      "step": 81165
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0493311517667153e-05,
+      "loss": 0.2958,
+      "step": 81170
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.049050949446534e-05,
+      "loss": 0.2549,
+      "step": 81175
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.048770752981629e-05,
+      "loss": 0.2545,
+      "step": 81180
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.048490562375637e-05,
+      "loss": 0.2702,
+      "step": 81185
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0482103776321983e-05,
+      "loss": 0.291,
+      "step": 81190
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.047930198754951e-05,
+      "loss": 0.2739,
+      "step": 81195
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.047650025747532e-05,
+      "loss": 0.2484,
+      "step": 81200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0473698586135776e-05,
+      "loss": 0.286,
+      "step": 81205
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.047089697356729e-05,
+      "loss": 0.2554,
+      "step": 81210
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.046809541980621e-05,
+      "loss": 0.303,
+      "step": 81215
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0465293924888936e-05,
+      "loss": 0.2859,
+      "step": 81220
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.046249248885182e-05,
+      "loss": 0.2765,
+      "step": 81225
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0459691111731258e-05,
+      "loss": 0.2501,
+      "step": 81230
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0456889793563617e-05,
+      "loss": 0.2619,
+      "step": 81235
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.045408853438526e-05,
+      "loss": 0.2544,
+      "step": 81240
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0451287334232567e-05,
+      "loss": 0.2714,
+      "step": 81245
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0448486193141914e-05,
+      "loss": 0.2519,
+      "step": 81250
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.044568511114967e-05,
+      "loss": 0.2771,
+      "step": 81255
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0442884088292204e-05,
+      "loss": 0.2701,
+      "step": 81260
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0440083124605873e-05,
+      "loss": 0.2639,
+      "step": 81265
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0437282220127067e-05,
+      "loss": 0.2588,
+      "step": 81270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0434481374892148e-05,
+      "loss": 0.2643,
+      "step": 81275
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.043168058893747e-05,
+      "loss": 0.2891,
+      "step": 81280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0428879862299414e-05,
+      "loss": 0.2688,
+      "step": 81285
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0426079195014347e-05,
+      "loss": 0.2552,
+      "step": 81290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.042327858711862e-05,
+      "loss": 0.2722,
+      "step": 81295
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0420478038648604e-05,
+      "loss": 0.2542,
+      "step": 81300
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0417677549640662e-05,
+      "loss": 0.2934,
+      "step": 81305
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0414877120131163e-05,
+      "loss": 0.2756,
+      "step": 81310
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0412076750156463e-05,
+      "loss": 0.276,
+      "step": 81315
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.040927643975291e-05,
+      "loss": 0.2806,
+      "step": 81320
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0406476188956892e-05,
+      "loss": 0.2896,
+      "step": 81325
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0403675997804748e-05,
+      "loss": 0.2642,
+      "step": 81330
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0400875866332845e-05,
+      "loss": 0.2652,
+      "step": 81335
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0398075794577527e-05,
+      "loss": 0.2599,
+      "step": 81340
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0395275782575172e-05,
+      "loss": 0.2457,
+      "step": 81345
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0392475830362122e-05,
+      "loss": 0.274,
+      "step": 81350
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0389675937974745e-05,
+      "loss": 0.2751,
+      "step": 81355
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0386876105449372e-05,
+      "loss": 0.2649,
+      "step": 81360
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0384076332822387e-05,
+      "loss": 0.2685,
+      "step": 81365
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0381276620130126e-05,
+      "loss": 0.2657,
+      "step": 81370
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0378476967408942e-05,
+      "loss": 0.282,
+      "step": 81375
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0375677374695188e-05,
+      "loss": 0.283,
+      "step": 81380
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0372877842025214e-05,
+      "loss": 0.273,
+      "step": 81385
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0370078369435378e-05,
+      "loss": 0.2746,
+      "step": 81390
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0367278956962012e-05,
+      "loss": 0.2806,
+      "step": 81395
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.036447960464149e-05,
+      "loss": 0.2817,
+      "step": 81400
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.036168031251014e-05,
+      "loss": 0.2665,
+      "step": 81405
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0358881080604313e-05,
+      "loss": 0.2534,
+      "step": 81410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.035608190896035e-05,
+      "loss": 0.2562,
+      "step": 81415
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.035328279761461e-05,
+      "loss": 0.2575,
+      "step": 81420
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0350483746603437e-05,
+      "loss": 0.292,
+      "step": 81425
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.0347684755963164e-05,
+      "loss": 0.2702,
+      "step": 81430
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.034488582573013e-05,
+      "loss": 0.2691,
+      "step": 81435
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.034208695594069e-05,
+      "loss": 0.2523,
+      "step": 81440
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0339288146631185e-05,
+      "loss": 0.3037,
+      "step": 81445
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0336489397837956e-05,
+      "loss": 0.264,
+      "step": 81450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0333690709597326e-05,
+      "loss": 0.2505,
+      "step": 81455
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0330892081945657e-05,
+      "loss": 0.2608,
+      "step": 81460
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.032809351491927e-05,
+      "loss": 0.2412,
+      "step": 81465
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0325295008554517e-05,
+      "loss": 0.2593,
+      "step": 81470
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.032249656288772e-05,
+      "loss": 0.2561,
+      "step": 81475
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0319698177955228e-05,
+      "loss": 0.2668,
+      "step": 81480
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0316899853793374e-05,
+      "loss": 0.2754,
+      "step": 81485
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0314101590438484e-05,
+      "loss": 0.2649,
+      "step": 81490
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0311303387926888e-05,
+      "loss": 0.2693,
+      "step": 81495
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0308505246294942e-05,
+      "loss": 0.2814,
+      "step": 81500
+    },
+    {
+      "epoch": 2.87,
+      "eval_loss": 0.2588658034801483,
+      "eval_runtime": 10.5545,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 9.475,
+      "step": 81500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0305707165578963e-05,
+      "loss": 0.2613,
+      "step": 81505
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.030290914581527e-05,
+      "loss": 0.2867,
+      "step": 81510
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0300111187040222e-05,
+      "loss": 0.2575,
+      "step": 81515
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0297313289290122e-05,
+      "loss": 0.2701,
+      "step": 81520
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0294515452601316e-05,
+      "loss": 0.2571,
+      "step": 81525
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0291717677010115e-05,
+      "loss": 0.2659,
+      "step": 81530
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0288919962552873e-05,
+      "loss": 0.2722,
+      "step": 81535
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0286122309265894e-05,
+      "loss": 0.2492,
+      "step": 81540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0283324717185505e-05,
+      "loss": 0.2694,
+      "step": 81545
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0280527186348035e-05,
+      "loss": 0.2685,
+      "step": 81550
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0277729716789817e-05,
+      "loss": 0.2735,
+      "step": 81555
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.027493230854717e-05,
+      "loss": 0.2833,
+      "step": 81560
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.027213496165641e-05,
+      "loss": 0.3045,
+      "step": 81565
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0269337676153852e-05,
+      "loss": 0.2866,
+      "step": 81570
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.026654045207583e-05,
+      "loss": 0.2906,
+      "step": 81575
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.026374328945867e-05,
+      "loss": 0.2755,
+      "step": 81580
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0260946188338677e-05,
+      "loss": 0.271,
+      "step": 81585
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0258149148752168e-05,
+      "loss": 0.2736,
+      "step": 81590
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0255352170735477e-05,
+      "loss": 0.2743,
+      "step": 81595
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0252555254324905e-05,
+      "loss": 0.2799,
+      "step": 81600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.024975839955678e-05,
+      "loss": 0.2507,
+      "step": 81605
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0246961606467402e-05,
+      "loss": 0.2396,
+      "step": 81610
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0244164875093106e-05,
+      "loss": 0.2552,
+      "step": 81615
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0241368205470194e-05,
+      "loss": 0.2729,
+      "step": 81620
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.023857159763497e-05,
+      "loss": 0.2491,
+      "step": 81625
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0235775051623756e-05,
+      "loss": 0.2673,
+      "step": 81630
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0232978567472875e-05,
+      "loss": 0.2717,
+      "step": 81635
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0230182145218624e-05,
+      "loss": 0.2596,
+      "step": 81640
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.02273857848973e-05,
+      "loss": 0.2645,
+      "step": 81645
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.022458948654524e-05,
+      "loss": 0.2759,
+      "step": 81650
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0221793250198738e-05,
+      "loss": 0.2445,
+      "step": 81655
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.02189970758941e-05,
+      "loss": 0.2694,
+      "step": 81660
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0216200963667624e-05,
+      "loss": 0.2725,
+      "step": 81665
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.021340491355564e-05,
+      "loss": 0.266,
+      "step": 81670
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0210608925594438e-05,
+      "loss": 0.249,
+      "step": 81675
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0207812999820316e-05,
+      "loss": 0.2599,
+      "step": 81680
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0205017136269587e-05,
+      "loss": 0.2606,
+      "step": 81685
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0202221334978548e-05,
+      "loss": 0.2802,
+      "step": 81690
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0199425595983505e-05,
+      "loss": 0.247,
+      "step": 81695
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.019662991932076e-05,
+      "loss": 0.2425,
+      "step": 81700
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0193834305026603e-05,
+      "loss": 0.2506,
+      "step": 81705
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0191038753137344e-05,
+      "loss": 0.2518,
+      "step": 81710
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.0188243263689276e-05,
+      "loss": 0.2504,
+      "step": 81715
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.018544783671871e-05,
+      "loss": 0.2751,
+      "step": 81720
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.018265247226191e-05,
+      "loss": 0.2701,
+      "step": 81725
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0179857170355207e-05,
+      "loss": 0.2809,
+      "step": 81730
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0177061931034875e-05,
+      "loss": 0.2608,
+      "step": 81735
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0174266754337217e-05,
+      "loss": 0.286,
+      "step": 81740
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.017147164029853e-05,
+      "loss": 0.2641,
+      "step": 81745
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.01686765889551e-05,
+      "loss": 0.2808,
+      "step": 81750
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.016588160034322e-05,
+      "loss": 0.2879,
+      "step": 81755
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0163086674499175e-05,
+      "loss": 0.2807,
+      "step": 81760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.016029181145927e-05,
+      "loss": 0.2345,
+      "step": 81765
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0157497011259785e-05,
+      "loss": 0.2527,
+      "step": 81770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.015470227393701e-05,
+      "loss": 0.2662,
+      "step": 81775
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0151907599527225e-05,
+      "loss": 0.2849,
+      "step": 81780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0149112988066734e-05,
+      "loss": 0.2453,
+      "step": 81785
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.014631843959181e-05,
+      "loss": 0.2547,
+      "step": 81790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.014352395413875e-05,
+      "loss": 0.2393,
+      "step": 81795
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0140729531743816e-05,
+      "loss": 0.265,
+      "step": 81800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.013793517244332e-05,
+      "loss": 0.2536,
+      "step": 81805
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.013514087627353e-05,
+      "loss": 0.2445,
+      "step": 81810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0132346643270725e-05,
+      "loss": 0.3005,
+      "step": 81815
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.012955247347119e-05,
+      "loss": 0.2731,
+      "step": 81820
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0126758366911207e-05,
+      "loss": 0.2605,
+      "step": 81825
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.012396432362706e-05,
+      "loss": 0.2733,
+      "step": 81830
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0121170343655024e-05,
+      "loss": 0.2584,
+      "step": 81835
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0118376427031367e-05,
+      "loss": 0.2725,
+      "step": 81840
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0115582573792382e-05,
+      "loss": 0.2574,
+      "step": 81845
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0112788783974335e-05,
+      "loss": 0.2502,
+      "step": 81850
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.01099950576135e-05,
+      "loss": 0.2637,
+      "step": 81855
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0107201394746165e-05,
+      "loss": 0.2621,
+      "step": 81860
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0104407795408596e-05,
+      "loss": 0.2777,
+      "step": 81865
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0101614259637065e-05,
+      "loss": 0.2842,
+      "step": 81870
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0098820787467835e-05,
+      "loss": 0.2921,
+      "step": 81875
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0096027378937193e-05,
+      "loss": 0.2932,
+      "step": 81880
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0093234034081406e-05,
+      "loss": 0.2608,
+      "step": 81885
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0090440752936744e-05,
+      "loss": 0.2765,
+      "step": 81890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.008764753553946e-05,
+      "loss": 0.2695,
+      "step": 81895
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.008485438192585e-05,
+      "loss": 0.2474,
+      "step": 81900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.008206129213216e-05,
+      "loss": 0.2591,
+      "step": 81905
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0079268266194668e-05,
+      "loss": 0.279,
+      "step": 81910
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0076475304149624e-05,
+      "loss": 0.263,
+      "step": 81915
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0073682406033315e-05,
+      "loss": 0.2706,
+      "step": 81920
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0070889571881993e-05,
+      "loss": 0.2994,
+      "step": 81925
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0068096801731915e-05,
+      "loss": 0.2746,
+      "step": 81930
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0065304095619344e-05,
+      "loss": 0.2926,
+      "step": 81935
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.006251145358056e-05,
+      "loss": 0.2821,
+      "step": 81940
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0059718875651812e-05,
+      "loss": 0.2838,
+      "step": 81945
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0056926361869354e-05,
+      "loss": 0.2792,
+      "step": 81950
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0054692397053082e-05,
+      "loss": 0.2797,
+      "step": 81955
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0051899998825325e-05,
+      "loss": 0.2649,
+      "step": 81960
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0049107664845383e-05,
+      "loss": 0.2662,
+      "step": 81965
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0046315395149516e-05,
+      "loss": 0.2687,
+      "step": 81970
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0043523189773983e-05,
+      "loss": 0.2927,
+      "step": 81975
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0040731048755038e-05,
+      "loss": 0.2563,
+      "step": 81980
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.003793897212892e-05,
+      "loss": 0.265,
+      "step": 81985
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0035146959931905e-05,
+      "loss": 0.2949,
+      "step": 81990
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.003235501220023e-05,
+      "loss": 0.2663,
+      "step": 81995
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0029563128970156e-05,
+      "loss": 0.2588,
+      "step": 82000
+    },
+    {
+      "epoch": 2.88,
+      "eval_loss": 0.2587546408176422,
+      "eval_runtime": 10.5627,
+      "eval_samples_per_second": 9.467,
+      "eval_steps_per_second": 9.467,
+      "step": 82000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0026771310277916e-05,
+      "loss": 0.2722,
+      "step": 82005
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.002397955615979e-05,
+      "loss": 0.2583,
+      "step": 82010
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0021187866651997e-05,
+      "loss": 0.2724,
+      "step": 82015
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.001839624179081e-05,
+      "loss": 0.246,
+      "step": 82020
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.001560468161245e-05,
+      "loss": 0.3009,
+      "step": 82025
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.001281318615319e-05,
+      "loss": 0.2928,
+      "step": 82030
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.001002175544927e-05,
+      "loss": 0.2486,
+      "step": 82035
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0007230389536916e-05,
+      "loss": 0.2804,
+      "step": 82040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0004439088452386e-05,
+      "loss": 0.2471,
+      "step": 82045
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 2.0001647852231924e-05,
+      "loss": 0.2752,
+      "step": 82050
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9998856680911777e-05,
+      "loss": 0.2607,
+      "step": 82055
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9996065574528168e-05,
+      "loss": 0.2797,
+      "step": 82060
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9993274533117363e-05,
+      "loss": 0.2755,
+      "step": 82065
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9990483556715587e-05,
+      "loss": 0.2648,
+      "step": 82070
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9987692645359075e-05,
+      "loss": 0.2684,
+      "step": 82075
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9984901799084064e-05,
+      "loss": 0.2473,
+      "step": 82080
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9982111017926812e-05,
+      "loss": 0.2687,
+      "step": 82085
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9979320301923537e-05,
+      "loss": 0.2566,
+      "step": 82090
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9976529651110483e-05,
+      "loss": 0.2715,
+      "step": 82095
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9973739065523865e-05,
+      "loss": 0.2655,
+      "step": 82100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9970948545199943e-05,
+      "loss": 0.2656,
+      "step": 82105
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9968158090174946e-05,
+      "loss": 0.2702,
+      "step": 82110
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.99653677004851e-05,
+      "loss": 0.2788,
+      "step": 82115
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9962577376166622e-05,
+      "loss": 0.2872,
+      "step": 82120
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.995978711725577e-05,
+      "loss": 0.2743,
+      "step": 82125
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9956996923788757e-05,
+      "loss": 0.2629,
+      "step": 82130
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9954206795801815e-05,
+      "loss": 0.2858,
+      "step": 82135
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.995141673333117e-05,
+      "loss": 0.2714,
+      "step": 82140
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9948626736413055e-05,
+      "loss": 0.2657,
+      "step": 82145
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9945836805083695e-05,
+      "loss": 0.2455,
+      "step": 82150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.994304693937931e-05,
+      "loss": 0.2824,
+      "step": 82155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9940257139336122e-05,
+      "loss": 0.2578,
+      "step": 82160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.993746740499037e-05,
+      "loss": 0.2316,
+      "step": 82165
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9934677736378265e-05,
+      "loss": 0.2439,
+      "step": 82170
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9931888133536027e-05,
+      "loss": 0.2705,
+      "step": 82175
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9929098596499887e-05,
+      "loss": 0.2591,
+      "step": 82180
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9926309125306057e-05,
+      "loss": 0.2901,
+      "step": 82185
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9923519719990763e-05,
+      "loss": 0.2306,
+      "step": 82190
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9920730380590212e-05,
+      "loss": 0.2629,
+      "step": 82195
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9917941107140635e-05,
+      "loss": 0.2528,
+      "step": 82200
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.991515189967825e-05,
+      "loss": 0.2852,
+      "step": 82205
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9912362758239258e-05,
+      "loss": 0.2471,
+      "step": 82210
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.990957368285988e-05,
+      "loss": 0.2607,
+      "step": 82215
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.990678467357634e-05,
+      "loss": 0.2535,
+      "step": 82220
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9903995730424846e-05,
+      "loss": 0.2794,
+      "step": 82225
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9901206853441607e-05,
+      "loss": 0.2707,
+      "step": 82230
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.989841804266283e-05,
+      "loss": 0.2502,
+      "step": 82235
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9895629298124734e-05,
+      "loss": 0.2555,
+      "step": 82240
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9892840619863535e-05,
+      "loss": 0.2574,
+      "step": 82245
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9890052007915435e-05,
+      "loss": 0.2736,
+      "step": 82250
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.988726346231663e-05,
+      "loss": 0.2708,
+      "step": 82255
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.988447498310335e-05,
+      "loss": 0.2763,
+      "step": 82260
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9881686570311788e-05,
+      "loss": 0.2744,
+      "step": 82265
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.987889822397816e-05,
+      "loss": 0.2648,
+      "step": 82270
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.987610994413865e-05,
+      "loss": 0.2605,
+      "step": 82275
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.9873321730829486e-05,
+      "loss": 0.264,
+      "step": 82280
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.987053358408686e-05,
+      "loss": 0.2632,
+      "step": 82285
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9867745503946968e-05,
+      "loss": 0.2622,
+      "step": 82290
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9864957490446018e-05,
+      "loss": 0.2775,
+      "step": 82295
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9862169543620222e-05,
+      "loss": 0.2779,
+      "step": 82300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9859381663505763e-05,
+      "loss": 0.2398,
+      "step": 82305
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9856593850138837e-05,
+      "loss": 0.2947,
+      "step": 82310
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9853806103555662e-05,
+      "loss": 0.2575,
+      "step": 82315
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9851018423792416e-05,
+      "loss": 0.2662,
+      "step": 82320
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.984823081088531e-05,
+      "loss": 0.2516,
+      "step": 82325
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9845443264870522e-05,
+      "loss": 0.2661,
+      "step": 82330
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9842655785784265e-05,
+      "loss": 0.2357,
+      "step": 82335
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9839868373662722e-05,
+      "loss": 0.2718,
+      "step": 82340
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9837081028542088e-05,
+      "loss": 0.271,
+      "step": 82345
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.983429375045855e-05,
+      "loss": 0.2681,
+      "step": 82350
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9831506539448302e-05,
+      "loss": 0.2809,
+      "step": 82355
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9828719395547545e-05,
+      "loss": 0.2644,
+      "step": 82360
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9825932318792454e-05,
+      "loss": 0.2748,
+      "step": 82365
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9823145309219213e-05,
+      "loss": 0.2707,
+      "step": 82370
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9820358366864026e-05,
+      "loss": 0.2676,
+      "step": 82375
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9817571491763076e-05,
+      "loss": 0.249,
+      "step": 82380
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9814784683952544e-05,
+      "loss": 0.2791,
+      "step": 82385
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9811997943468605e-05,
+      "loss": 0.2651,
+      "step": 82390
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9809211270347465e-05,
+      "loss": 0.2612,
+      "step": 82395
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.980642466462529e-05,
+      "loss": 0.27,
+      "step": 82400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.980363812633827e-05,
+      "loss": 0.2555,
+      "step": 82405
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.980085165552258e-05,
+      "loss": 0.2816,
+      "step": 82410
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9798065252214415e-05,
+      "loss": 0.2741,
+      "step": 82415
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.979527891644994e-05,
+      "loss": 0.2858,
+      "step": 82420
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.979249264826533e-05,
+      "loss": 0.263,
+      "step": 82425
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9789706447696785e-05,
+      "loss": 0.2587,
+      "step": 82430
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9786920314780462e-05,
+      "loss": 0.2738,
+      "step": 82435
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9784134249552544e-05,
+      "loss": 0.2599,
+      "step": 82440
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9781348252049196e-05,
+      "loss": 0.2548,
+      "step": 82445
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.977856232230661e-05,
+      "loss": 0.2498,
+      "step": 82450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9775776460360952e-05,
+      "loss": 0.2721,
+      "step": 82455
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.97729906662484e-05,
+      "loss": 0.2808,
+      "step": 82460
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9770204940005103e-05,
+      "loss": 0.2835,
+      "step": 82465
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.976741928166726e-05,
+      "loss": 0.2708,
+      "step": 82470
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.976463369127103e-05,
+      "loss": 0.2586,
+      "step": 82475
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9761848168852575e-05,
+      "loss": 0.2569,
+      "step": 82480
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.975906271444807e-05,
+      "loss": 0.2628,
+      "step": 82485
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9756277328093685e-05,
+      "loss": 0.3001,
+      "step": 82490
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9753492009825587e-05,
+      "loss": 0.2645,
+      "step": 82495
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9750706759679934e-05,
+      "loss": 0.2419,
+      "step": 82500
+    },
+    {
+      "epoch": 2.9,
+      "eval_loss": 0.25884824991226196,
+      "eval_runtime": 10.5461,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 82500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9747921577692884e-05,
+      "loss": 0.266,
+      "step": 82505
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9745136463900626e-05,
+      "loss": 0.2801,
+      "step": 82510
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.97423514183393e-05,
+      "loss": 0.2743,
+      "step": 82515
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9739566441045072e-05,
+      "loss": 0.2633,
+      "step": 82520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9736781532054115e-05,
+      "loss": 0.2905,
+      "step": 82525
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9733996691402583e-05,
+      "loss": 0.2409,
+      "step": 82530
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9731211919126634e-05,
+      "loss": 0.275,
+      "step": 82535
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9728427215262415e-05,
+      "loss": 0.2609,
+      "step": 82540
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.97256425798461e-05,
+      "loss": 0.272,
+      "step": 82545
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9722858012913843e-05,
+      "loss": 0.2581,
+      "step": 82550
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9720073514501802e-05,
+      "loss": 0.2868,
+      "step": 82555
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.971728908464611e-05,
+      "loss": 0.2457,
+      "step": 82560
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.9714504723382952e-05,
+      "loss": 0.2821,
+      "step": 82565
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9711720430748466e-05,
+      "loss": 0.2635,
+      "step": 82570
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9708936206778803e-05,
+      "loss": 0.2541,
+      "step": 82575
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9706152051510106e-05,
+      "loss": 0.2585,
+      "step": 82580
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.970336796497855e-05,
+      "loss": 0.2627,
+      "step": 82585
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9700583947220268e-05,
+      "loss": 0.2771,
+      "step": 82590
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.969779999827141e-05,
+      "loss": 0.2966,
+      "step": 82595
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9695016118168117e-05,
+      "loss": 0.2555,
+      "step": 82600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.969223230694655e-05,
+      "loss": 0.2502,
+      "step": 82605
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9689448564642855e-05,
+      "loss": 0.238,
+      "step": 82610
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9686664891293167e-05,
+      "loss": 0.275,
+      "step": 82615
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9683881286933625e-05,
+      "loss": 0.2512,
+      "step": 82620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9681097751600387e-05,
+      "loss": 0.2571,
+      "step": 82625
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.967831428532959e-05,
+      "loss": 0.2597,
+      "step": 82630
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9675530888157366e-05,
+      "loss": 0.2538,
+      "step": 82635
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9672747560119877e-05,
+      "loss": 0.2731,
+      "step": 82640
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9669964301253252e-05,
+      "loss": 0.2618,
+      "step": 82645
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9667181111593618e-05,
+      "loss": 0.2696,
+      "step": 82650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.966439799117712e-05,
+      "loss": 0.2296,
+      "step": 82655
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.966161494003991e-05,
+      "loss": 0.263,
+      "step": 82660
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9658831958218107e-05,
+      "loss": 0.2833,
+      "step": 82665
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9656049045747853e-05,
+      "loss": 0.2817,
+      "step": 82670
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.965326620266527e-05,
+      "loss": 0.2714,
+      "step": 82675
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9650483429006512e-05,
+      "loss": 0.2619,
+      "step": 82680
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.96477007248077e-05,
+      "loss": 0.2632,
+      "step": 82685
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9644918090104968e-05,
+      "loss": 0.2504,
+      "step": 82690
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9642135524934434e-05,
+      "loss": 0.2637,
+      "step": 82695
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9639353029332255e-05,
+      "loss": 0.2801,
+      "step": 82700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9636570603334537e-05,
+      "loss": 0.2835,
+      "step": 82705
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9633788246977415e-05,
+      "loss": 0.2811,
+      "step": 82710
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.963100596029701e-05,
+      "loss": 0.2644,
+      "step": 82715
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9628223743329466e-05,
+      "loss": 0.2563,
+      "step": 82720
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9625441596110894e-05,
+      "loss": 0.2486,
+      "step": 82725
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9622659518677416e-05,
+      "loss": 0.2612,
+      "step": 82730
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9619877511065157e-05,
+      "loss": 0.2785,
+      "step": 82735
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.961709557331025e-05,
+      "loss": 0.2814,
+      "step": 82740
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9614313705448814e-05,
+      "loss": 0.2843,
+      "step": 82745
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.961153190751695e-05,
+      "loss": 0.2801,
+      "step": 82750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9608750179550806e-05,
+      "loss": 0.2575,
+      "step": 82755
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.960596852158648e-05,
+      "loss": 0.272,
+      "step": 82760
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.960318693366011e-05,
+      "loss": 0.2707,
+      "step": 82765
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9600405415807786e-05,
+      "loss": 0.2527,
+      "step": 82770
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9597623968065647e-05,
+      "loss": 0.2772,
+      "step": 82775
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9594842590469807e-05,
+      "loss": 0.2358,
+      "step": 82780
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9592061283056363e-05,
+      "loss": 0.2604,
+      "step": 82785
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.958928004586144e-05,
+      "loss": 0.2802,
+      "step": 82790
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9586498878921153e-05,
+      "loss": 0.2797,
+      "step": 82795
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9583717782271614e-05,
+      "loss": 0.2408,
+      "step": 82800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9580936755948932e-05,
+      "loss": 0.2661,
+      "step": 82805
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.95781557999892e-05,
+      "loss": 0.2735,
+      "step": 82810
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9575374914428548e-05,
+      "loss": 0.2756,
+      "step": 82815
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9572594099303086e-05,
+      "loss": 0.2808,
+      "step": 82820
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.956981335464891e-05,
+      "loss": 0.2507,
+      "step": 82825
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.956703268050212e-05,
+      "loss": 0.2484,
+      "step": 82830
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9564252076898842e-05,
+      "loss": 0.2662,
+      "step": 82835
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.956147154387516e-05,
+      "loss": 0.2683,
+      "step": 82840
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9558691081467193e-05,
+      "loss": 0.2644,
+      "step": 82845
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9555910689711025e-05,
+      "loss": 0.256,
+      "step": 82850
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.955313036864278e-05,
+      "loss": 0.2545,
+      "step": 82855
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.955035011829855e-05,
+      "loss": 0.2803,
+      "step": 82860
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9547569938714415e-05,
+      "loss": 0.2853,
+      "step": 82865
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9544789829926507e-05,
+      "loss": 0.2946,
+      "step": 82870
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.95420097919709e-05,
+      "loss": 0.2517,
+      "step": 82875
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.953922982488371e-05,
+      "loss": 0.2771,
+      "step": 82880
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9536449928701007e-05,
+      "loss": 0.2789,
+      "step": 82885
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9533670103458912e-05,
+      "loss": 0.2807,
+      "step": 82890
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9530890349193502e-05,
+      "loss": 0.2674,
+      "step": 82895
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9528110665940882e-05,
+      "loss": 0.2566,
+      "step": 82900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9525331053737128e-05,
+      "loss": 0.2781,
+      "step": 82905
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9522551512618354e-05,
+      "loss": 0.2592,
+      "step": 82910
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9519772042620637e-05,
+      "loss": 0.2896,
+      "step": 82915
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9516992643780068e-05,
+      "loss": 0.2845,
+      "step": 82920
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9514213316132732e-05,
+      "loss": 0.256,
+      "step": 82925
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.951143405971472e-05,
+      "loss": 0.2472,
+      "step": 82930
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9508654874562127e-05,
+      "loss": 0.2716,
+      "step": 82935
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.950587576071103e-05,
+      "loss": 0.2728,
+      "step": 82940
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9503096718197507e-05,
+      "loss": 0.279,
+      "step": 82945
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9500317747057662e-05,
+      "loss": 0.2568,
+      "step": 82950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9497538847327557e-05,
+      "loss": 0.2763,
+      "step": 82955
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9494760019043292e-05,
+      "loss": 0.2828,
+      "step": 82960
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.949198126224093e-05,
+      "loss": 0.3088,
+      "step": 82965
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.948920257695657e-05,
+      "loss": 0.253,
+      "step": 82970
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9486423963226284e-05,
+      "loss": 0.2667,
+      "step": 82975
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9483645421086137e-05,
+      "loss": 0.2419,
+      "step": 82980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9480866950572227e-05,
+      "loss": 0.2721,
+      "step": 82985
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9478088551720626e-05,
+      "loss": 0.2733,
+      "step": 82990
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9475310224567405e-05,
+      "loss": 0.2441,
+      "step": 82995
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9472531969148632e-05,
+      "loss": 0.2591,
+      "step": 83000
+    },
+    {
+      "epoch": 2.92,
+      "eval_loss": 0.25857260823249817,
+      "eval_runtime": 10.5923,
+      "eval_samples_per_second": 9.441,
+      "eval_steps_per_second": 9.441,
+      "step": 83000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9469753785500397e-05,
+      "loss": 0.2793,
+      "step": 83005
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.946697567365876e-05,
+      "loss": 0.2883,
+      "step": 83010
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9464197633659802e-05,
+      "loss": 0.2625,
+      "step": 83015
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9461419665539582e-05,
+      "loss": 0.2913,
+      "step": 83020
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9458641769334184e-05,
+      "loss": 0.2658,
+      "step": 83025
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.945586394507967e-05,
+      "loss": 0.2692,
+      "step": 83030
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.945308619281211e-05,
+      "loss": 0.2514,
+      "step": 83035
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9450308512567562e-05,
+      "loss": 0.2617,
+      "step": 83040
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9447530904382108e-05,
+      "loss": 0.2585,
+      "step": 83045
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9444753368291808e-05,
+      "loss": 0.2485,
+      "step": 83050
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9441975904332727e-05,
+      "loss": 0.2593,
+      "step": 83055
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9439198512540913e-05,
+      "loss": 0.2562,
+      "step": 83060
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9436421192952445e-05,
+      "loss": 0.2747,
+      "step": 83065
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.943364394560339e-05,
+      "loss": 0.2603,
+      "step": 83070
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.94308667705298e-05,
+      "loss": 0.2622,
+      "step": 83075
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9428089667767722e-05,
+      "loss": 0.2696,
+      "step": 83080
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.942531263735324e-05,
+      "loss": 0.2567,
+      "step": 83085
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.942253567932239e-05,
+      "loss": 0.2608,
+      "step": 83090
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.941975879371124e-05,
+      "loss": 0.2979,
+      "step": 83095
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.941698198055585e-05,
+      "loss": 0.2658,
+      "step": 83100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.941420523989227e-05,
+      "loss": 0.2573,
+      "step": 83105
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.941142857175656e-05,
+      "loss": 0.2538,
+      "step": 83110
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.940865197618475e-05,
+      "loss": 0.2743,
+      "step": 83115
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.940587545321291e-05,
+      "loss": 0.2432,
+      "step": 83120
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.9403099002877108e-05,
+      "loss": 0.2777,
+      "step": 83125
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.940032262521337e-05,
+      "loss": 0.2655,
+      "step": 83130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.939754632025774e-05,
+      "loss": 0.2827,
+      "step": 83135
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9394770088046294e-05,
+      "loss": 0.265,
+      "step": 83140
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.939199392861506e-05,
+      "loss": 0.2621,
+      "step": 83145
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9389217842000088e-05,
+      "loss": 0.2781,
+      "step": 83150
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.938644182823742e-05,
+      "loss": 0.2644,
+      "step": 83155
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9383665887363113e-05,
+      "loss": 0.2719,
+      "step": 83160
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.93808900194132e-05,
+      "loss": 0.2746,
+      "step": 83165
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9378114224423726e-05,
+      "loss": 0.2669,
+      "step": 83170
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.937533850243073e-05,
+      "loss": 0.2738,
+      "step": 83175
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9372562853470263e-05,
+      "loss": 0.2719,
+      "step": 83180
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9369787277578365e-05,
+      "loss": 0.2708,
+      "step": 83185
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9367011774791062e-05,
+      "loss": 0.2549,
+      "step": 83190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9364236345144393e-05,
+      "loss": 0.2744,
+      "step": 83195
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.93614609886744e-05,
+      "loss": 0.3034,
+      "step": 83200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9358685705417133e-05,
+      "loss": 0.2712,
+      "step": 83205
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.935591049540861e-05,
+      "loss": 0.2648,
+      "step": 83210
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.935313535868486e-05,
+      "loss": 0.2597,
+      "step": 83215
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9350360295281937e-05,
+      "loss": 0.2538,
+      "step": 83220
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9347585305235858e-05,
+      "loss": 0.2593,
+      "step": 83225
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.934481038858266e-05,
+      "loss": 0.2686,
+      "step": 83230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9342035545358366e-05,
+      "loss": 0.2657,
+      "step": 83235
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.933926077559902e-05,
+      "loss": 0.2594,
+      "step": 83240
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9336486079340643e-05,
+      "loss": 0.2823,
+      "step": 83245
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.933371145661925e-05,
+      "loss": 0.2676,
+      "step": 83250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9330936907470886e-05,
+      "loss": 0.2641,
+      "step": 83255
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9328162431931575e-05,
+      "loss": 0.2608,
+      "step": 83260
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.932538803003734e-05,
+      "loss": 0.2673,
+      "step": 83265
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9322613701824188e-05,
+      "loss": 0.2695,
+      "step": 83270
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9319839447328167e-05,
+      "loss": 0.2779,
+      "step": 83275
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9317065266585283e-05,
+      "loss": 0.2721,
+      "step": 83280
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9314291159631566e-05,
+      "loss": 0.2656,
+      "step": 83285
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.931151712650302e-05,
+      "loss": 0.2559,
+      "step": 83290
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9308743167235683e-05,
+      "loss": 0.2672,
+      "step": 83295
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9305969281865568e-05,
+      "loss": 0.2542,
+      "step": 83300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9303195470428687e-05,
+      "loss": 0.2666,
+      "step": 83305
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9300421732961055e-05,
+      "loss": 0.2709,
+      "step": 83310
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9297648069498692e-05,
+      "loss": 0.2806,
+      "step": 83315
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9294874480077613e-05,
+      "loss": 0.2487,
+      "step": 83320
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9292100964733835e-05,
+      "loss": 0.2662,
+      "step": 83325
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.928932752350335e-05,
+      "loss": 0.2892,
+      "step": 83330
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9286554156422188e-05,
+      "loss": 0.2582,
+      "step": 83335
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.928378086352636e-05,
+      "loss": 0.2557,
+      "step": 83340
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.928100764485186e-05,
+      "loss": 0.269,
+      "step": 83345
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.927823450043472e-05,
+      "loss": 0.2733,
+      "step": 83350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.927546143031093e-05,
+      "loss": 0.267,
+      "step": 83355
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.92726884345165e-05,
+      "loss": 0.2704,
+      "step": 83360
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9269915513087437e-05,
+      "loss": 0.277,
+      "step": 83365
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.926714266605974e-05,
+      "loss": 0.2227,
+      "step": 83370
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9264369893469424e-05,
+      "loss": 0.2612,
+      "step": 83375
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9261597195352486e-05,
+      "loss": 0.2579,
+      "step": 83380
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9258824571744917e-05,
+      "loss": 0.2713,
+      "step": 83385
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9256052022682738e-05,
+      "loss": 0.2766,
+      "step": 83390
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.925327954820193e-05,
+      "loss": 0.27,
+      "step": 83395
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.925050714833851e-05,
+      "loss": 0.2737,
+      "step": 83400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.924773482312845e-05,
+      "loss": 0.2652,
+      "step": 83405
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9244962572607776e-05,
+      "loss": 0.2849,
+      "step": 83410
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9242190396812472e-05,
+      "loss": 0.2707,
+      "step": 83415
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9239418295778518e-05,
+      "loss": 0.2855,
+      "step": 83420
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.923664626954192e-05,
+      "loss": 0.2633,
+      "step": 83425
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9233874318138688e-05,
+      "loss": 0.2519,
+      "step": 83430
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9231102441604793e-05,
+      "loss": 0.256,
+      "step": 83435
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9228330639976223e-05,
+      "loss": 0.2413,
+      "step": 83440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9225558913288982e-05,
+      "loss": 0.2396,
+      "step": 83445
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.922278726157905e-05,
+      "loss": 0.2843,
+      "step": 83450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.922001568488242e-05,
+      "loss": 0.256,
+      "step": 83455
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9217244183235073e-05,
+      "loss": 0.2692,
+      "step": 83460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9214472756673008e-05,
+      "loss": 0.2827,
+      "step": 83465
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.92117014052322e-05,
+      "loss": 0.2706,
+      "step": 83470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.920893012894863e-05,
+      "loss": 0.2947,
+      "step": 83475
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.920615892785828e-05,
+      "loss": 0.278,
+      "step": 83480
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9203387801997148e-05,
+      "loss": 0.2691,
+      "step": 83485
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9200616751401207e-05,
+      "loss": 0.2807,
+      "step": 83490
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.919784577610643e-05,
+      "loss": 0.2716,
+      "step": 83495
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.91950748761488e-05,
+      "loss": 0.2561,
+      "step": 83500
+    },
+    {
+      "epoch": 2.94,
+      "eval_loss": 0.2586362063884735,
+      "eval_runtime": 10.5563,
+      "eval_samples_per_second": 9.473,
+      "eval_steps_per_second": 9.473,
+      "step": 83500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9192304051564296e-05,
+      "loss": 0.2662,
+      "step": 83505
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9189533302388906e-05,
+      "loss": 0.2656,
+      "step": 83510
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9186762628658594e-05,
+      "loss": 0.2826,
+      "step": 83515
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9183992030409333e-05,
+      "loss": 0.2547,
+      "step": 83520
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9181221507677104e-05,
+      "loss": 0.2424,
+      "step": 83525
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.917845106049788e-05,
+      "loss": 0.2857,
+      "step": 83530
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.917568068890764e-05,
+      "loss": 0.2393,
+      "step": 83535
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.917291039294233e-05,
+      "loss": 0.2903,
+      "step": 83540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9170140172637953e-05,
+      "loss": 0.2738,
+      "step": 83545
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.916737002803046e-05,
+      "loss": 0.2613,
+      "step": 83550
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9164599959155823e-05,
+      "loss": 0.2761,
+      "step": 83555
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9161829966050003e-05,
+      "loss": 0.27,
+      "step": 83560
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9159060048748984e-05,
+      "loss": 0.2818,
+      "step": 83565
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9156290207288723e-05,
+      "loss": 0.2836,
+      "step": 83570
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9153520441705168e-05,
+      "loss": 0.2686,
+      "step": 83575
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.915075075203431e-05,
+      "loss": 0.2763,
+      "step": 83580
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9147981138312093e-05,
+      "loss": 0.2754,
+      "step": 83585
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.914521160057449e-05,
+      "loss": 0.2634,
+      "step": 83590
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.914244213885744e-05,
+      "loss": 0.2856,
+      "step": 83595
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9139672753196936e-05,
+      "loss": 0.2785,
+      "step": 83600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9136903443628916e-05,
+      "loss": 0.255,
+      "step": 83605
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.913413421018934e-05,
+      "loss": 0.2541,
+      "step": 83610
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9131365052914154e-05,
+      "loss": 0.2687,
+      "step": 83615
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.912859597183934e-05,
+      "loss": 0.2711,
+      "step": 83620
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9125826967000836e-05,
+      "loss": 0.2686,
+      "step": 83625
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.91230580384346e-05,
+      "loss": 0.2434,
+      "step": 83630
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.912028918617657e-05,
+      "loss": 0.3043,
+      "step": 83635
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.911752041026271e-05,
+      "loss": 0.2653,
+      "step": 83640
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9114751710728983e-05,
+      "loss": 0.2592,
+      "step": 83645
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9111983087611328e-05,
+      "loss": 0.2862,
+      "step": 83650
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9109214540945677e-05,
+      "loss": 0.2584,
+      "step": 83655
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9106446070768007e-05,
+      "loss": 0.2845,
+      "step": 83660
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9103677677114244e-05,
+      "loss": 0.264,
+      "step": 83665
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9100909360020343e-05,
+      "loss": 0.2732,
+      "step": 83670
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.909814111952224e-05,
+      "loss": 0.2679,
+      "step": 83675
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9095372955655894e-05,
+      "loss": 0.267,
+      "step": 83680
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9092604868457238e-05,
+      "loss": 0.2704,
+      "step": 83685
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9089836857962203e-05,
+      "loss": 0.2713,
+      "step": 83690
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.908706892420675e-05,
+      "loss": 0.2662,
+      "step": 83695
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9084301067226815e-05,
+      "loss": 0.2738,
+      "step": 83700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9081533287058328e-05,
+      "loss": 0.2564,
+      "step": 83705
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9078765583737223e-05,
+      "loss": 0.2507,
+      "step": 83710
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9075997957299456e-05,
+      "loss": 0.2706,
+      "step": 83715
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9073230407780947e-05,
+      "loss": 0.2618,
+      "step": 83720
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9070462935217638e-05,
+      "loss": 0.277,
+      "step": 83725
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9067695539645453e-05,
+      "loss": 0.2748,
+      "step": 83730
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.906492822110034e-05,
+      "loss": 0.2872,
+      "step": 83735
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9062160979618224e-05,
+      "loss": 0.256,
+      "step": 83740
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9059393815235026e-05,
+      "loss": 0.2739,
+      "step": 83745
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9056626727986686e-05,
+      "loss": 0.2743,
+      "step": 83750
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9053859717909132e-05,
+      "loss": 0.2674,
+      "step": 83755
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9051092785038297e-05,
+      "loss": 0.2634,
+      "step": 83760
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9048325929410103e-05,
+      "loss": 0.2678,
+      "step": 83765
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.904555915106046e-05,
+      "loss": 0.2741,
+      "step": 83770
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9042792450025313e-05,
+      "loss": 0.2709,
+      "step": 83775
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.904002582634059e-05,
+      "loss": 0.2552,
+      "step": 83780
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9037259280042197e-05,
+      "loss": 0.2854,
+      "step": 83785
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9034492811166056e-05,
+      "loss": 0.2707,
+      "step": 83790
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9031726419748103e-05,
+      "loss": 0.2613,
+      "step": 83795
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9028960105824245e-05,
+      "loss": 0.27,
+      "step": 83800
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9026193869430404e-05,
+      "loss": 0.2898,
+      "step": 83805
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9023427710602504e-05,
+      "loss": 0.261,
+      "step": 83810
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9020661629376456e-05,
+      "loss": 0.2876,
+      "step": 83815
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9017895625788178e-05,
+      "loss": 0.2652,
+      "step": 83820
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9015129699873573e-05,
+      "loss": 0.2626,
+      "step": 83825
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9012363851668576e-05,
+      "loss": 0.2728,
+      "step": 83830
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.900959808120908e-05,
+      "loss": 0.2711,
+      "step": 83835
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.900683238853101e-05,
+      "loss": 0.2751,
+      "step": 83840
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9004066773670265e-05,
+      "loss": 0.2988,
+      "step": 83845
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.900130123666277e-05,
+      "loss": 0.2605,
+      "step": 83850
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8998535777544424e-05,
+      "loss": 0.2607,
+      "step": 83855
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8995770396351134e-05,
+      "loss": 0.2659,
+      "step": 83860
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.89930050931188e-05,
+      "loss": 0.2775,
+      "step": 83865
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8990239867883348e-05,
+      "loss": 0.2859,
+      "step": 83870
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.898747472068067e-05,
+      "loss": 0.2673,
+      "step": 83875
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8984709651546667e-05,
+      "loss": 0.2706,
+      "step": 83880
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8981944660517243e-05,
+      "loss": 0.27,
+      "step": 83885
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.89791797476283e-05,
+      "loss": 0.2736,
+      "step": 83890
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8976414912915748e-05,
+      "loss": 0.2677,
+      "step": 83895
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.897365015641548e-05,
+      "loss": 0.2639,
+      "step": 83900
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.897088547816338e-05,
+      "loss": 0.2631,
+      "step": 83905
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.896812087819537e-05,
+      "loss": 0.2578,
+      "step": 83910
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.896535635654733e-05,
+      "loss": 0.2558,
+      "step": 83915
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8962591913255158e-05,
+      "loss": 0.2719,
+      "step": 83920
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.895982754835476e-05,
+      "loss": 0.2821,
+      "step": 83925
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.895706326188202e-05,
+      "loss": 0.2708,
+      "step": 83930
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.895429905387283e-05,
+      "loss": 0.282,
+      "step": 83935
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.895153492436307e-05,
+      "loss": 0.2764,
+      "step": 83940
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8948770873388654e-05,
+      "loss": 0.2581,
+      "step": 83945
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.894600690098546e-05,
+      "loss": 0.2626,
+      "step": 83950
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.894324300718938e-05,
+      "loss": 0.2693,
+      "step": 83955
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8940479192036286e-05,
+      "loss": 0.2833,
+      "step": 83960
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.893771545556209e-05,
+      "loss": 0.2659,
+      "step": 83965
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8934951797802654e-05,
+      "loss": 0.2795,
+      "step": 83970
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.893218821879388e-05,
+      "loss": 0.2528,
+      "step": 83975
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.892942471857163e-05,
+      "loss": 0.2465,
+      "step": 83980
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.8926661297171816e-05,
+      "loss": 0.2652,
+      "step": 83985
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8923897954630298e-05,
+      "loss": 0.2615,
+      "step": 83990
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8921134690982957e-05,
+      "loss": 0.2867,
+      "step": 83995
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8918371506265666e-05,
+      "loss": 0.2851,
+      "step": 84000
+    },
+    {
+      "epoch": 2.96,
+      "eval_loss": 0.2586404085159302,
+      "eval_runtime": 10.5437,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 84000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.891560840051433e-05,
+      "loss": 0.2446,
+      "step": 84005
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8912845373764804e-05,
+      "loss": 0.2668,
+      "step": 84010
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8910082426052973e-05,
+      "loss": 0.2509,
+      "step": 84015
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8907319557414696e-05,
+      "loss": 0.2543,
+      "step": 84020
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8904556767885865e-05,
+      "loss": 0.2615,
+      "step": 84025
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.890179405750235e-05,
+      "loss": 0.2538,
+      "step": 84030
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.889903142630001e-05,
+      "loss": 0.2779,
+      "step": 84035
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8896268874314743e-05,
+      "loss": 0.2719,
+      "step": 84040
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8893506401582397e-05,
+      "loss": 0.261,
+      "step": 84045
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.889074400813884e-05,
+      "loss": 0.2607,
+      "step": 84050
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8887981694019946e-05,
+      "loss": 0.2538,
+      "step": 84055
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8885219459261588e-05,
+      "loss": 0.2449,
+      "step": 84060
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8882457303899626e-05,
+      "loss": 0.2915,
+      "step": 84065
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8879695227969923e-05,
+      "loss": 0.2925,
+      "step": 84070
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8876933231508337e-05,
+      "loss": 0.2451,
+      "step": 84075
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.887417131455074e-05,
+      "loss": 0.272,
+      "step": 84080
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8871409477132997e-05,
+      "loss": 0.265,
+      "step": 84085
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8868647719290966e-05,
+      "loss": 0.2835,
+      "step": 84090
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.886588604106049e-05,
+      "loss": 0.2811,
+      "step": 84095
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8863124442477453e-05,
+      "loss": 0.2649,
+      "step": 84100
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8860362923577694e-05,
+      "loss": 0.2909,
+      "step": 84105
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8857601484397082e-05,
+      "loss": 0.2592,
+      "step": 84110
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.885484012497146e-05,
+      "loss": 0.2774,
+      "step": 84115
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8852078845336694e-05,
+      "loss": 0.2713,
+      "step": 84120
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8849317645528636e-05,
+      "loss": 0.2646,
+      "step": 84125
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8846556525583127e-05,
+      "loss": 0.2542,
+      "step": 84130
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8843795485536027e-05,
+      "loss": 0.2764,
+      "step": 84135
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.884103452542319e-05,
+      "loss": 0.2772,
+      "step": 84140
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.883827364528046e-05,
+      "loss": 0.281,
+      "step": 84145
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.883551284514368e-05,
+      "loss": 0.2541,
+      "step": 84150
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8832752125048713e-05,
+      "loss": 0.2678,
+      "step": 84155
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8829991485031387e-05,
+      "loss": 0.2587,
+      "step": 84160
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8827230925127566e-05,
+      "loss": 0.2556,
+      "step": 84165
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8824470445373065e-05,
+      "loss": 0.2801,
+      "step": 84170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.882171004580376e-05,
+      "loss": 0.2753,
+      "step": 84175
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.881894972645548e-05,
+      "loss": 0.2691,
+      "step": 84180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8816189487364056e-05,
+      "loss": 0.2618,
+      "step": 84185
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8813429328565342e-05,
+      "loss": 0.244,
+      "step": 84190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8810669250095163e-05,
+      "loss": 0.2633,
+      "step": 84195
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8807909251989376e-05,
+      "loss": 0.2525,
+      "step": 84200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8805149334283807e-05,
+      "loss": 0.2616,
+      "step": 84205
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.880238949701428e-05,
+      "loss": 0.2629,
+      "step": 84210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8799629740216643e-05,
+      "loss": 0.2652,
+      "step": 84215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8796870063926733e-05,
+      "loss": 0.2691,
+      "step": 84220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8794110468180383e-05,
+      "loss": 0.2754,
+      "step": 84225
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8791350953013402e-05,
+      "loss": 0.281,
+      "step": 84230
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8788591518461647e-05,
+      "loss": 0.2688,
+      "step": 84235
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8785832164560937e-05,
+      "loss": 0.2692,
+      "step": 84240
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8783072891347105e-05,
+      "loss": 0.2761,
+      "step": 84245
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8780313698855964e-05,
+      "loss": 0.2574,
+      "step": 84250
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.877755458712336e-05,
+      "loss": 0.2711,
+      "step": 84255
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.877479555618511e-05,
+      "loss": 0.2504,
+      "step": 84260
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8772036606077025e-05,
+      "loss": 0.2722,
+      "step": 84265
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.8769277736834954e-05,
+      "loss": 0.2722,
+      "step": 84270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.87665189484947e-05,
+      "loss": 0.2855,
+      "step": 84275
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.876376024109209e-05,
+      "loss": 0.2642,
+      "step": 84280
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8761001614662938e-05,
+      "loss": 0.2801,
+      "step": 84285
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8758243069243077e-05,
+      "loss": 0.2736,
+      "step": 84290
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8755484604868315e-05,
+      "loss": 0.2919,
+      "step": 84295
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8752726221574466e-05,
+      "loss": 0.2704,
+      "step": 84300
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8749967919397342e-05,
+      "loss": 0.2514,
+      "step": 84305
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.874720969837278e-05,
+      "loss": 0.2576,
+      "step": 84310
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.874445155853658e-05,
+      "loss": 0.2626,
+      "step": 84315
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8741693499924546e-05,
+      "loss": 0.269,
+      "step": 84320
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8738935522572498e-05,
+      "loss": 0.2554,
+      "step": 84325
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8736177626516246e-05,
+      "loss": 0.266,
+      "step": 84330
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8733419811791604e-05,
+      "loss": 0.2702,
+      "step": 84335
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.873066207843438e-05,
+      "loss": 0.2851,
+      "step": 84340
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8727904426480362e-05,
+      "loss": 0.2812,
+      "step": 84345
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8725146855965385e-05,
+      "loss": 0.2573,
+      "step": 84350
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8722389366925232e-05,
+      "loss": 0.2504,
+      "step": 84355
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.871963195939572e-05,
+      "loss": 0.2567,
+      "step": 84360
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8716874633412647e-05,
+      "loss": 0.2856,
+      "step": 84365
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.871411738901182e-05,
+      "loss": 0.2759,
+      "step": 84370
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8711360226229035e-05,
+      "loss": 0.2461,
+      "step": 84375
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8708603145100085e-05,
+      "loss": 0.2644,
+      "step": 84380
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8705846145660782e-05,
+      "loss": 0.2525,
+      "step": 84385
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8703089227946924e-05,
+      "loss": 0.274,
+      "step": 84390
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.87003323919943e-05,
+      "loss": 0.2611,
+      "step": 84395
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8697575637838703e-05,
+      "loss": 0.2785,
+      "step": 84400
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8694818965515938e-05,
+      "loss": 0.2823,
+      "step": 84405
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.869206237506179e-05,
+      "loss": 0.2583,
+      "step": 84410
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8689305866512063e-05,
+      "loss": 0.2773,
+      "step": 84415
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8686549439902528e-05,
+      "loss": 0.2528,
+      "step": 84420
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8683793095268997e-05,
+      "loss": 0.2625,
+      "step": 84425
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8681036832647253e-05,
+      "loss": 0.274,
+      "step": 84430
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8678280652073074e-05,
+      "loss": 0.2494,
+      "step": 84435
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.867552455358225e-05,
+      "loss": 0.272,
+      "step": 84440
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8672768537210583e-05,
+      "loss": 0.2845,
+      "step": 84445
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8670012602993843e-05,
+      "loss": 0.2485,
+      "step": 84450
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.866725675096782e-05,
+      "loss": 0.267,
+      "step": 84455
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8664500981168286e-05,
+      "loss": 0.3005,
+      "step": 84460
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8661745293631036e-05,
+      "loss": 0.2624,
+      "step": 84465
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.865898968839185e-05,
+      "loss": 0.255,
+      "step": 84470
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8656234165486506e-05,
+      "loss": 0.27,
+      "step": 84475
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.865347872495077e-05,
+      "loss": 0.2593,
+      "step": 84480
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.865072336682044e-05,
+      "loss": 0.2904,
+      "step": 84485
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8647968091131273e-05,
+      "loss": 0.2726,
+      "step": 84490
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8645212897919067e-05,
+      "loss": 0.261,
+      "step": 84495
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.864245778721957e-05,
+      "loss": 0.2694,
+      "step": 84500
+    },
+    {
+      "epoch": 2.97,
+      "eval_loss": 0.2578907310962677,
+      "eval_runtime": 10.553,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 84500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8639702759068578e-05,
+      "loss": 0.2505,
+      "step": 84505
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8636947813501853e-05,
+      "loss": 0.2453,
+      "step": 84510
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8634192950555163e-05,
+      "loss": 0.2729,
+      "step": 84515
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.863143817026428e-05,
+      "loss": 0.2606,
+      "step": 84520
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8628683472664988e-05,
+      "loss": 0.2258,
+      "step": 84525
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8625928857793034e-05,
+      "loss": 0.2844,
+      "step": 84530
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8623725225483194e-05,
+      "loss": 0.2798,
+      "step": 84535
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8620970759610597e-05,
+      "loss": 0.2855,
+      "step": 84540
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.861821637656549e-05,
+      "loss": 0.2655,
+      "step": 84545
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8615462076383633e-05,
+      "loss": 0.279,
+      "step": 84550
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.8612707859100796e-05,
+      "loss": 0.2723,
+      "step": 84555
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8609953724752745e-05,
+      "loss": 0.2618,
+      "step": 84560
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.860719967337523e-05,
+      "loss": 0.2769,
+      "step": 84565
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.860444570500401e-05,
+      "loss": 0.276,
+      "step": 84570
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8601691819674847e-05,
+      "loss": 0.2715,
+      "step": 84575
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8598938017423505e-05,
+      "loss": 0.2972,
+      "step": 84580
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8596184298285725e-05,
+      "loss": 0.2717,
+      "step": 84585
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.859343066229728e-05,
+      "loss": 0.2735,
+      "step": 84590
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8590677109493918e-05,
+      "loss": 0.2694,
+      "step": 84595
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.858792363991139e-05,
+      "loss": 0.2491,
+      "step": 84600
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8585170253585436e-05,
+      "loss": 0.2665,
+      "step": 84605
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8582416950551823e-05,
+      "loss": 0.2561,
+      "step": 84610
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8579663730846305e-05,
+      "loss": 0.2626,
+      "step": 84615
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.857691059450462e-05,
+      "loss": 0.2642,
+      "step": 84620
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.857415754156251e-05,
+      "loss": 0.2656,
+      "step": 84625
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.857140457205574e-05,
+      "loss": 0.2666,
+      "step": 84630
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8568651686020037e-05,
+      "loss": 0.288,
+      "step": 84635
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.856589888349116e-05,
+      "loss": 0.2713,
+      "step": 84640
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.856314616450483e-05,
+      "loss": 0.2631,
+      "step": 84645
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8560393529096827e-05,
+      "loss": 0.2697,
+      "step": 84650
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.855764097730286e-05,
+      "loss": 0.2666,
+      "step": 84655
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.855488850915868e-05,
+      "loss": 0.2862,
+      "step": 84660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8552136124700014e-05,
+      "loss": 0.264,
+      "step": 84665
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8549383823962625e-05,
+      "loss": 0.2594,
+      "step": 84670
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.854663160698224e-05,
+      "loss": 0.2513,
+      "step": 84675
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8543879473794584e-05,
+      "loss": 0.2686,
+      "step": 84680
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.854112742443539e-05,
+      "loss": 0.2723,
+      "step": 84685
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8538375458940408e-05,
+      "loss": 0.2528,
+      "step": 84690
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8535623577345364e-05,
+      "loss": 0.285,
+      "step": 84695
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.853287177968598e-05,
+      "loss": 0.2968,
+      "step": 84700
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8530120065998004e-05,
+      "loss": 0.2695,
+      "step": 84705
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.852736843631715e-05,
+      "loss": 0.2702,
+      "step": 84710
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.852461689067915e-05,
+      "loss": 0.2808,
+      "step": 84715
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.852186542911973e-05,
+      "loss": 0.2427,
+      "step": 84720
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8519114051674626e-05,
+      "loss": 0.2538,
+      "step": 84725
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8516362758379552e-05,
+      "loss": 0.2875,
+      "step": 84730
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.851361154927024e-05,
+      "loss": 0.2583,
+      "step": 84735
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8510860424382392e-05,
+      "loss": 0.2762,
+      "step": 84740
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8508109383751754e-05,
+      "loss": 0.2963,
+      "step": 84745
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8505358427414037e-05,
+      "loss": 0.258,
+      "step": 84750
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8502607555404966e-05,
+      "loss": 0.2605,
+      "step": 84755
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8499856767760243e-05,
+      "loss": 0.2667,
+      "step": 84760
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8497106064515605e-05,
+      "loss": 0.2566,
+      "step": 84765
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8494355445706755e-05,
+      "loss": 0.2565,
+      "step": 84770
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8491604911369418e-05,
+      "loss": 0.2795,
+      "step": 84775
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.848885446153929e-05,
+      "loss": 0.2754,
+      "step": 84780
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8486104096252104e-05,
+      "loss": 0.2819,
+      "step": 84785
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.848335381554357e-05,
+      "loss": 0.255,
+      "step": 84790
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8480603619449383e-05,
+      "loss": 0.2634,
+      "step": 84795
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8477853508005255e-05,
+      "loss": 0.2802,
+      "step": 84800
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8475103481246912e-05,
+      "loss": 0.277,
+      "step": 84805
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8472353539210052e-05,
+      "loss": 0.2462,
+      "step": 84810
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.846960368193037e-05,
+      "loss": 0.2716,
+      "step": 84815
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8466853909443588e-05,
+      "loss": 0.2653,
+      "step": 84820
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.84641042217854e-05,
+      "loss": 0.281,
+      "step": 84825
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.846135461899151e-05,
+      "loss": 0.2615,
+      "step": 84830
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8458605101097618e-05,
+      "loss": 0.2775,
+      "step": 84835
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8455855668139434e-05,
+      "loss": 0.2839,
+      "step": 84840
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8453106320152654e-05,
+      "loss": 0.2763,
+      "step": 84845
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8450357057172967e-05,
+      "loss": 0.2809,
+      "step": 84850
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8447607879236078e-05,
+      "loss": 0.2501,
+      "step": 84855
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8444858786377682e-05,
+      "loss": 0.2753,
+      "step": 84860
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.844210977863348e-05,
+      "loss": 0.2645,
+      "step": 84865
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8439360856039163e-05,
+      "loss": 0.3029,
+      "step": 84870
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.843661201863041e-05,
+      "loss": 0.2619,
+      "step": 84875
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8433863266442937e-05,
+      "loss": 0.2809,
+      "step": 84880
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8431114599512412e-05,
+      "loss": 0.2983,
+      "step": 84885
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.842836601787454e-05,
+      "loss": 0.2685,
+      "step": 84890
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8425617521564998e-05,
+      "loss": 0.2727,
+      "step": 84895
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.842286911061949e-05,
+      "loss": 0.2634,
+      "step": 84900
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8420120785073685e-05,
+      "loss": 0.2866,
+      "step": 84905
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.841737254496328e-05,
+      "loss": 0.2641,
+      "step": 84910
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8414624390323946e-05,
+      "loss": 0.2716,
+      "step": 84915
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8411876321191384e-05,
+      "loss": 0.2535,
+      "step": 84920
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8409128337601267e-05,
+      "loss": 0.2594,
+      "step": 84925
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8406380439589262e-05,
+      "loss": 0.2455,
+      "step": 84930
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8403632627191074e-05,
+      "loss": 0.2776,
+      "step": 84935
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.840088490044236e-05,
+      "loss": 0.2514,
+      "step": 84940
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8398137259378816e-05,
+      "loss": 0.2845,
+      "step": 84945
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8395389704036096e-05,
+      "loss": 0.2742,
+      "step": 84950
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8392642234449903e-05,
+      "loss": 0.2583,
+      "step": 84955
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8389894850655895e-05,
+      "loss": 0.2654,
+      "step": 84960
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8387147552689737e-05,
+      "loss": 0.2815,
+      "step": 84965
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.838440034058711e-05,
+      "loss": 0.2674,
+      "step": 84970
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8381653214383692e-05,
+      "loss": 0.2691,
+      "step": 84975
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8378906174115145e-05,
+      "loss": 0.2662,
+      "step": 84980
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8376159219817135e-05,
+      "loss": 0.2625,
+      "step": 84985
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.837341235152533e-05,
+      "loss": 0.2713,
+      "step": 84990
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.83706655692754e-05,
+      "loss": 0.2785,
+      "step": 84995
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8367918873103012e-05,
+      "loss": 0.2897,
+      "step": 85000
+    },
+    {
+      "epoch": 2.99,
+      "eval_loss": 0.2575870752334595,
+      "eval_runtime": 10.5655,
+      "eval_samples_per_second": 9.465,
+      "eval_steps_per_second": 9.465,
+      "step": 85000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8365172263043827e-05,
+      "loss": 0.2663,
+      "step": 85005
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.83624257391335e-05,
+      "loss": 0.2902,
+      "step": 85010
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8359679301407705e-05,
+      "loss": 0.2739,
+      "step": 85015
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.83569329499021e-05,
+      "loss": 0.2879,
+      "step": 85020
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8354186684652343e-05,
+      "loss": 0.2705,
+      "step": 85025
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8351440505694083e-05,
+      "loss": 0.2706,
+      "step": 85030
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8348694413063e-05,
+      "loss": 0.2817,
+      "step": 85035
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8345948406794733e-05,
+      "loss": 0.2729,
+      "step": 85040
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8343202486924933e-05,
+      "loss": 0.2621,
+      "step": 85045
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8340456653489264e-05,
+      "loss": 0.2582,
+      "step": 85050
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8337710906523384e-05,
+      "loss": 0.2626,
+      "step": 85055
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8334965246062936e-05,
+      "loss": 0.2544,
+      "step": 85060
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.833221967214356e-05,
+      "loss": 0.2565,
+      "step": 85065
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8329474184800932e-05,
+      "loss": 0.2458,
+      "step": 85070
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8326728784070676e-05,
+      "loss": 0.2674,
+      "step": 85075
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8323983469988458e-05,
+      "loss": 0.2548,
+      "step": 85080
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.83212382425899e-05,
+      "loss": 0.279,
+      "step": 85085
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8318493101910676e-05,
+      "loss": 0.2629,
+      "step": 85090
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8315748047986415e-05,
+      "loss": 0.2786,
+      "step": 85095
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8313003080852755e-05,
+      "loss": 0.2567,
+      "step": 85100
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8310258200545338e-05,
+      "loss": 0.2676,
+      "step": 85105
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.830751340709982e-05,
+      "loss": 0.2571,
+      "step": 85110
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.830476870055183e-05,
+      "loss": 0.2583,
+      "step": 85115
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8302024080937008e-05,
+      "loss": 0.2619,
+      "step": 85120
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8299279548290975e-05,
+      "loss": 0.2721,
+      "step": 85125
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8296535102649388e-05,
+      "loss": 0.2721,
+      "step": 85130
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.829379074404788e-05,
+      "loss": 0.2777,
+      "step": 85135
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.829104647252208e-05,
+      "loss": 0.2647,
+      "step": 85140
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8288302288107613e-05,
+      "loss": 0.2806,
+      "step": 85145
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8285558190840123e-05,
+      "loss": 0.2671,
+      "step": 85150
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.828281418075523e-05,
+      "loss": 0.2809,
+      "step": 85155
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8280070257888565e-05,
+      "loss": 0.2558,
+      "step": 85160
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8277326422275767e-05,
+      "loss": 0.2798,
+      "step": 85165
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8274582673952458e-05,
+      "loss": 0.2723,
+      "step": 85170
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8271839012954258e-05,
+      "loss": 0.2397,
+      "step": 85175
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8269095439316786e-05,
+      "loss": 0.259,
+      "step": 85180
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8266351953075676e-05,
+      "loss": 0.2452,
+      "step": 85185
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.826360855426656e-05,
+      "loss": 0.2612,
+      "step": 85190
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8260865242925042e-05,
+      "loss": 0.2567,
+      "step": 85195
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8258122019086736e-05,
+      "loss": 0.2513,
+      "step": 85200
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.825537888278729e-05,
+      "loss": 0.2751,
+      "step": 85205
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8252635834062292e-05,
+      "loss": 0.2635,
+      "step": 85210
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.824989287294738e-05,
+      "loss": 0.2698,
+      "step": 85215
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.824714999947815e-05,
+      "loss": 0.2512,
+      "step": 85220
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8244407213690234e-05,
+      "loss": 0.287,
+      "step": 85225
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8241664515619245e-05,
+      "loss": 0.2691,
+      "step": 85230
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8238921905300777e-05,
+      "loss": 0.2627,
+      "step": 85235
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8236179382770448e-05,
+      "loss": 0.2863,
+      "step": 85240
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8233436948063885e-05,
+      "loss": 0.2653,
+      "step": 85245
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.823069460121668e-05,
+      "loss": 0.2855,
+      "step": 85250
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.822795234226445e-05,
+      "loss": 0.289,
+      "step": 85255
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8225210171242778e-05,
+      "loss": 0.2794,
+      "step": 85260
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.82224680881873e-05,
+      "loss": 0.2918,
+      "step": 85265
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.821972609313361e-05,
+      "loss": 0.2634,
+      "step": 85270
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8216984186117297e-05,
+      "loss": 0.249,
+      "step": 85275
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8214242367173984e-05,
+      "loss": 0.2576,
+      "step": 85280
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.821150063633926e-05,
+      "loss": 0.2529,
+      "step": 85285
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.820875899364873e-05,
+      "loss": 0.2693,
+      "step": 85290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.820601743913798e-05,
+      "loss": 0.2434,
+      "step": 85295
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.820327597284262e-05,
+      "loss": 0.2495,
+      "step": 85300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8200534594798244e-05,
+      "loss": 0.247,
+      "step": 85305
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.819779330504045e-05,
+      "loss": 0.2358,
+      "step": 85310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.819505210360481e-05,
+      "loss": 0.2602,
+      "step": 85315
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.819231099052695e-05,
+      "loss": 0.2637,
+      "step": 85320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.818956996584244e-05,
+      "loss": 0.2622,
+      "step": 85325
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.818682902958688e-05,
+      "loss": 0.2495,
+      "step": 85330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8184088181795844e-05,
+      "loss": 0.2399,
+      "step": 85335
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.818134742250494e-05,
+      "loss": 0.2366,
+      "step": 85340
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8178606751749742e-05,
+      "loss": 0.2724,
+      "step": 85345
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.817586616956585e-05,
+      "loss": 0.2627,
+      "step": 85350
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8173125675988822e-05,
+      "loss": 0.2537,
+      "step": 85355
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.817038527105427e-05,
+      "loss": 0.2662,
+      "step": 85360
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8167644954797763e-05,
+      "loss": 0.2653,
+      "step": 85365
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8164904727254883e-05,
+      "loss": 0.2675,
+      "step": 85370
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8162164588461207e-05,
+      "loss": 0.2528,
+      "step": 85375
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.815942453845232e-05,
+      "loss": 0.265,
+      "step": 85380
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.81566845772638e-05,
+      "loss": 0.2637,
+      "step": 85385
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.815394470493121e-05,
+      "loss": 0.2915,
+      "step": 85390
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8151204921490146e-05,
+      "loss": 0.2486,
+      "step": 85395
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8148465226976175e-05,
+      "loss": 0.2617,
+      "step": 85400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.814572562142486e-05,
+      "loss": 0.2724,
+      "step": 85405
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.814298610487178e-05,
+      "loss": 0.2499,
+      "step": 85410
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8140246677352514e-05,
+      "loss": 0.2662,
+      "step": 85415
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8137507338902625e-05,
+      "loss": 0.2476,
+      "step": 85420
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8134768089557675e-05,
+      "loss": 0.2447,
+      "step": 85425
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8132028929353235e-05,
+      "loss": 0.2333,
+      "step": 85430
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8129289858324875e-05,
+      "loss": 0.271,
+      "step": 85435
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8126550876508163e-05,
+      "loss": 0.2516,
+      "step": 85440
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.812381198393866e-05,
+      "loss": 0.2651,
+      "step": 85445
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8121073180651915e-05,
+      "loss": 0.2639,
+      "step": 85450
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.811833446668351e-05,
+      "loss": 0.2613,
+      "step": 85455
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8115595842068995e-05,
+      "loss": 0.2641,
+      "step": 85460
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8112857306843933e-05,
+      "loss": 0.2608,
+      "step": 85465
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.811011886104387e-05,
+      "loss": 0.2452,
+      "step": 85470
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8107380504704386e-05,
+      "loss": 0.2699,
+      "step": 85475
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.810464223786102e-05,
+      "loss": 0.2613,
+      "step": 85480
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8101904060549328e-05,
+      "loss": 0.271,
+      "step": 85485
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.809916597280486e-05,
+      "loss": 0.2667,
+      "step": 85490
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8096427974663183e-05,
+      "loss": 0.2599,
+      "step": 85495
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.809369006615984e-05,
+      "loss": 0.2713,
+      "step": 85500
+    },
+    {
+      "epoch": 3.01,
+      "eval_loss": 0.2571566104888916,
+      "eval_runtime": 10.5571,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 9.472,
+      "step": 85500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8090952247330374e-05,
+      "loss": 0.2468,
+      "step": 85505
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8088214518210345e-05,
+      "loss": 0.2664,
+      "step": 85510
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.808547687883529e-05,
+      "loss": 0.2476,
+      "step": 85515
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.808273932924077e-05,
+      "loss": 0.2681,
+      "step": 85520
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8080001869462308e-05,
+      "loss": 0.2499,
+      "step": 85525
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.807726449953547e-05,
+      "loss": 0.2518,
+      "step": 85530
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8074527219495793e-05,
+      "loss": 0.2667,
+      "step": 85535
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.807179002937881e-05,
+      "loss": 0.268,
+      "step": 85540
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8069052929220064e-05,
+      "loss": 0.2868,
+      "step": 85545
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8066315919055106e-05,
+      "loss": 0.2774,
+      "step": 85550
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8063578998919467e-05,
+      "loss": 0.2597,
+      "step": 85555
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8060842168848686e-05,
+      "loss": 0.269,
+      "step": 85560
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.805810542887828e-05,
+      "loss": 0.2514,
+      "step": 85565
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.805536877904381e-05,
+      "loss": 0.2565,
+      "step": 85570
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8052632219380806e-05,
+      "loss": 0.2527,
+      "step": 85575
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.804989574992479e-05,
+      "loss": 0.2334,
+      "step": 85580
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.804715937071129e-05,
+      "loss": 0.2542,
+      "step": 85585
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8044423081775853e-05,
+      "loss": 0.2429,
+      "step": 85590
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8041686883153992e-05,
+      "loss": 0.2496,
+      "step": 85595
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8038950774881245e-05,
+      "loss": 0.2554,
+      "step": 85600
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.803621475699313e-05,
+      "loss": 0.239,
+      "step": 85605
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8033478829525185e-05,
+      "loss": 0.2629,
+      "step": 85610
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8030742992512924e-05,
+      "loss": 0.258,
+      "step": 85615
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8028007245991863e-05,
+      "loss": 0.2504,
+      "step": 85620
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.802527158999754e-05,
+      "loss": 0.2538,
+      "step": 85625
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8022536024565474e-05,
+      "loss": 0.251,
+      "step": 85630
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.801980054973118e-05,
+      "loss": 0.2519,
+      "step": 85635
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8017065165530162e-05,
+      "loss": 0.2593,
+      "step": 85640
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8014329871997963e-05,
+      "loss": 0.3128,
+      "step": 85645
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8011594669170078e-05,
+      "loss": 0.2533,
+      "step": 85650
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.800885955708204e-05,
+      "loss": 0.2519,
+      "step": 85655
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8006124535769345e-05,
+      "loss": 0.2527,
+      "step": 85660
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.800338960526752e-05,
+      "loss": 0.2399,
+      "step": 85665
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.8000654765612074e-05,
+      "loss": 0.2729,
+      "step": 85670
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7997920016838504e-05,
+      "loss": 0.2545,
+      "step": 85675
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7995185358982323e-05,
+      "loss": 0.2372,
+      "step": 85680
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7992450792079056e-05,
+      "loss": 0.2608,
+      "step": 85685
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7989716316164198e-05,
+      "loss": 0.2648,
+      "step": 85690
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.7986981931273254e-05,
+      "loss": 0.2433,
+      "step": 85695
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7984247637441714e-05,
+      "loss": 0.2697,
+      "step": 85700
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.79815134347051e-05,
+      "loss": 0.2455,
+      "step": 85705
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.797877932309892e-05,
+      "loss": 0.2586,
+      "step": 85710
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7976045302658657e-05,
+      "loss": 0.282,
+      "step": 85715
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.797331137341981e-05,
+      "loss": 0.2418,
+      "step": 85720
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7970577535417892e-05,
+      "loss": 0.2651,
+      "step": 85725
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7967843788688388e-05,
+      "loss": 0.2649,
+      "step": 85730
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7965110133266806e-05,
+      "loss": 0.2598,
+      "step": 85735
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7962376569188617e-05,
+      "loss": 0.2503,
+      "step": 85740
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7959643096489342e-05,
+      "loss": 0.2473,
+      "step": 85745
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.795690971520447e-05,
+      "loss": 0.2483,
+      "step": 85750
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7954176425369463e-05,
+      "loss": 0.2506,
+      "step": 85755
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7951443227019843e-05,
+      "loss": 0.2643,
+      "step": 85760
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7948710120191086e-05,
+      "loss": 0.257,
+      "step": 85765
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.794597710491869e-05,
+      "loss": 0.2715,
+      "step": 85770
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.794324418123811e-05,
+      "loss": 0.2561,
+      "step": 85775
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7940511349184873e-05,
+      "loss": 0.2645,
+      "step": 85780
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7937778608794437e-05,
+      "loss": 0.2468,
+      "step": 85785
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7935045960102298e-05,
+      "loss": 0.2588,
+      "step": 85790
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.793231340314392e-05,
+      "loss": 0.2506,
+      "step": 85795
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7929580937954803e-05,
+      "loss": 0.2624,
+      "step": 85800
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.792684856457042e-05,
+      "loss": 0.2708,
+      "step": 85805
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7924116283026244e-05,
+      "loss": 0.2593,
+      "step": 85810
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7921384093357753e-05,
+      "loss": 0.2432,
+      "step": 85815
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7918651995600423e-05,
+      "loss": 0.2573,
+      "step": 85820
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7915919989789733e-05,
+      "loss": 0.2652,
+      "step": 85825
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7913188075961156e-05,
+      "loss": 0.2361,
+      "step": 85830
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7910456254150152e-05,
+      "loss": 0.242,
+      "step": 85835
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.790772452439221e-05,
+      "loss": 0.2871,
+      "step": 85840
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7904992886722787e-05,
+      "loss": 0.2457,
+      "step": 85845
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7902261341177363e-05,
+      "loss": 0.2623,
+      "step": 85850
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7899529887791386e-05,
+      "loss": 0.2535,
+      "step": 85855
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7896798526600345e-05,
+      "loss": 0.2648,
+      "step": 85860
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7894067257639686e-05,
+      "loss": 0.2569,
+      "step": 85865
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7891336080944882e-05,
+      "loss": 0.2785,
+      "step": 85870
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7888604996551394e-05,
+      "loss": 0.2462,
+      "step": 85875
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7885874004494692e-05,
+      "loss": 0.265,
+      "step": 85880
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7883143104810223e-05,
+      "loss": 0.2593,
+      "step": 85885
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7880412297533444e-05,
+      "loss": 0.2675,
+      "step": 85890
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7877681582699827e-05,
+      "loss": 0.2416,
+      "step": 85895
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7874950960344818e-05,
+      "loss": 0.2507,
+      "step": 85900
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7872220430503878e-05,
+      "loss": 0.2235,
+      "step": 85905
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7869489993212445e-05,
+      "loss": 0.2617,
+      "step": 85910
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7866759648506e-05,
+      "loss": 0.2712,
+      "step": 85915
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.786402939641998e-05,
+      "loss": 0.2656,
+      "step": 85920
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7861299236989832e-05,
+      "loss": 0.2583,
+      "step": 85925
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7858569170251e-05,
+      "loss": 0.2673,
+      "step": 85930
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7855839196238956e-05,
+      "loss": 0.2758,
+      "step": 85935
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7853109314989125e-05,
+      "loss": 0.2585,
+      "step": 85940
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7850379526536958e-05,
+      "loss": 0.2618,
+      "step": 85945
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.78476498309179e-05,
+      "loss": 0.2605,
+      "step": 85950
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7844920228167397e-05,
+      "loss": 0.2557,
+      "step": 85955
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7842190718320896e-05,
+      "loss": 0.2569,
+      "step": 85960
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.783946130141383e-05,
+      "loss": 0.2488,
+      "step": 85965
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7836731977481628e-05,
+      "loss": 0.2511,
+      "step": 85970
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.7834002746559754e-05,
+      "loss": 0.2645,
+      "step": 85975
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.783127360868362e-05,
+      "loss": 0.2593,
+      "step": 85980
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7828544563888675e-05,
+      "loss": 0.2706,
+      "step": 85985
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.782581561221036e-05,
+      "loss": 0.2762,
+      "step": 85990
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7823086753684103e-05,
+      "loss": 0.2745,
+      "step": 85995
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7820357988345336e-05,
+      "loss": 0.2397,
+      "step": 86000
+    },
+    {
+      "epoch": 3.03,
+      "eval_loss": 0.2575805187225342,
+      "eval_runtime": 10.5481,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 86000
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7817629316229474e-05,
+      "loss": 0.2688,
+      "step": 86005
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.781490073737197e-05,
+      "loss": 0.2693,
+      "step": 86010
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.781217225180825e-05,
+      "loss": 0.2262,
+      "step": 86015
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7809443859573738e-05,
+      "loss": 0.2576,
+      "step": 86020
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7806715560703845e-05,
+      "loss": 0.2587,
+      "step": 86025
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7803987355234023e-05,
+      "loss": 0.2747,
+      "step": 86030
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7801259243199676e-05,
+      "loss": 0.2475,
+      "step": 86035
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.779853122463624e-05,
+      "loss": 0.2707,
+      "step": 86040
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7795803299579116e-05,
+      "loss": 0.2519,
+      "step": 86045
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7793075468063755e-05,
+      "loss": 0.2443,
+      "step": 86050
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7790347730125555e-05,
+      "loss": 0.2526,
+      "step": 86055
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7787620085799935e-05,
+      "loss": 0.2438,
+      "step": 86060
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7784892535122312e-05,
+      "loss": 0.2435,
+      "step": 86065
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.778216507812811e-05,
+      "loss": 0.2692,
+      "step": 86070
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.777943771485274e-05,
+      "loss": 0.2492,
+      "step": 86075
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7776710445331613e-05,
+      "loss": 0.2531,
+      "step": 86080
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.777398326960013e-05,
+      "loss": 0.248,
+      "step": 86085
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.777125618769372e-05,
+      "loss": 0.2593,
+      "step": 86090
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7768529199647784e-05,
+      "loss": 0.2566,
+      "step": 86095
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7765802305497724e-05,
+      "loss": 0.2718,
+      "step": 86100
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7763075505278962e-05,
+      "loss": 0.2584,
+      "step": 86105
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7760348799026896e-05,
+      "loss": 0.267,
+      "step": 86110
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7757622186776922e-05,
+      "loss": 0.2841,
+      "step": 86115
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7754895668564452e-05,
+      "loss": 0.2695,
+      "step": 86120
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7752169244424894e-05,
+      "loss": 0.2629,
+      "step": 86125
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7749442914393644e-05,
+      "loss": 0.2655,
+      "step": 86130
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7746716678506094e-05,
+      "loss": 0.2585,
+      "step": 86135
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.774399053679764e-05,
+      "loss": 0.2798,
+      "step": 86140
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7741264489303694e-05,
+      "loss": 0.2878,
+      "step": 86145
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.773853853605965e-05,
+      "loss": 0.2479,
+      "step": 86150
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7735812677100896e-05,
+      "loss": 0.2513,
+      "step": 86155
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7733086912462817e-05,
+      "loss": 0.2343,
+      "step": 86160
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7730361242180827e-05,
+      "loss": 0.2601,
+      "step": 86165
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7727635666290298e-05,
+      "loss": 0.2608,
+      "step": 86170
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.772491018482664e-05,
+      "loss": 0.2794,
+      "step": 86175
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.772218479782521e-05,
+      "loss": 0.2484,
+      "step": 86180
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7719459505321425e-05,
+      "loss": 0.2363,
+      "step": 86185
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7716734307350664e-05,
+      "loss": 0.2372,
+      "step": 86190
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.77140092039483e-05,
+      "loss": 0.2774,
+      "step": 86195
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.771128419514972e-05,
+      "loss": 0.2725,
+      "step": 86200
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7708559280990322e-05,
+      "loss": 0.2564,
+      "step": 86205
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7705834461505477e-05,
+      "loss": 0.2536,
+      "step": 86210
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7703109736730556e-05,
+      "loss": 0.2554,
+      "step": 86215
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.770038510670095e-05,
+      "loss": 0.2723,
+      "step": 86220
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7697660571452034e-05,
+      "loss": 0.2475,
+      "step": 86225
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7694936131019186e-05,
+      "loss": 0.2406,
+      "step": 86230
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7692211785437768e-05,
+      "loss": 0.259,
+      "step": 86235
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7689487534743177e-05,
+      "loss": 0.2834,
+      "step": 86240
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7686763378970767e-05,
+      "loss": 0.289,
+      "step": 86245
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7684039318155914e-05,
+      "loss": 0.2726,
+      "step": 86250
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7681315352333987e-05,
+      "loss": 0.2704,
+      "step": 86255
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7678591481540357e-05,
+      "loss": 0.2619,
+      "step": 86260
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.76758677058104e-05,
+      "loss": 0.2641,
+      "step": 86265
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7673144025179473e-05,
+      "loss": 0.2632,
+      "step": 86270
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.767042043968293e-05,
+      "loss": 0.26,
+      "step": 86275
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7667696949356156e-05,
+      "loss": 0.2451,
+      "step": 86280
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7664973554234505e-05,
+      "loss": 0.2431,
+      "step": 86285
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7662250254353337e-05,
+      "loss": 0.2845,
+      "step": 86290
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.765952704974801e-05,
+      "loss": 0.2732,
+      "step": 86295
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7656803940453893e-05,
+      "loss": 0.227,
+      "step": 86300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7654080926506335e-05,
+      "loss": 0.2505,
+      "step": 86305
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7651358007940694e-05,
+      "loss": 0.2697,
+      "step": 86310
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.764863518479232e-05,
+      "loss": 0.2797,
+      "step": 86315
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7645912457096586e-05,
+      "loss": 0.2775,
+      "step": 86320
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7643189824888827e-05,
+      "loss": 0.2449,
+      "step": 86325
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.764046728820439e-05,
+      "loss": 0.2385,
+      "step": 86330
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7637744847078652e-05,
+      "loss": 0.2674,
+      "step": 86335
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7635022501546938e-05,
+      "loss": 0.2579,
+      "step": 86340
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7632300251644606e-05,
+      "loss": 0.2655,
+      "step": 86345
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.762957809740699e-05,
+      "loss": 0.2744,
+      "step": 86350
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7626856038869456e-05,
+      "loss": 0.247,
+      "step": 86355
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7624134076067336e-05,
+      "loss": 0.268,
+      "step": 86360
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.762141220903597e-05,
+      "loss": 0.2387,
+      "step": 86365
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7618690437810697e-05,
+      "loss": 0.2421,
+      "step": 86370
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7615968762426876e-05,
+      "loss": 0.2817,
+      "step": 86375
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7613247182919836e-05,
+      "loss": 0.2491,
+      "step": 86380
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7610525699324908e-05,
+      "loss": 0.2615,
+      "step": 86385
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7607804311677427e-05,
+      "loss": 0.2524,
+      "step": 86390
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7605083020012744e-05,
+      "loss": 0.2461,
+      "step": 86395
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7602361824366186e-05,
+      "loss": 0.2614,
+      "step": 86400
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7599640724773085e-05,
+      "loss": 0.2538,
+      "step": 86405
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.759691972126876e-05,
+      "loss": 0.2505,
+      "step": 86410
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7594198813888568e-05,
+      "loss": 0.2782,
+      "step": 86415
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7591478002667817e-05,
+      "loss": 0.2593,
+      "step": 86420
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7588757287641844e-05,
+      "loss": 0.2685,
+      "step": 86425
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7586036668845966e-05,
+      "loss": 0.282,
+      "step": 86430
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7583316146315525e-05,
+      "loss": 0.2636,
+      "step": 86435
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7580595720085835e-05,
+      "loss": 0.2661,
+      "step": 86440
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.757787539019221e-05,
+      "loss": 0.2676,
+      "step": 86445
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7575155156669993e-05,
+      "loss": 0.257,
+      "step": 86450
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.757243501955449e-05,
+      "loss": 0.2919,
+      "step": 86455
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.756971497888103e-05,
+      "loss": 0.2632,
+      "step": 86460
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7566995034684908e-05,
+      "loss": 0.2783,
+      "step": 86465
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7564275187001472e-05,
+      "loss": 0.2767,
+      "step": 86470
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7561555435866012e-05,
+      "loss": 0.2671,
+      "step": 86475
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.755883578131386e-05,
+      "loss": 0.2403,
+      "step": 86480
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7556116223380315e-05,
+      "loss": 0.2492,
+      "step": 86485
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7553396762100702e-05,
+      "loss": 0.257,
+      "step": 86490
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7550677397510324e-05,
+      "loss": 0.2787,
+      "step": 86495
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.754795812964449e-05,
+      "loss": 0.2483,
+      "step": 86500
+    },
+    {
+      "epoch": 3.04,
+      "eval_loss": 0.2569468021392822,
+      "eval_runtime": 10.5367,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 86500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7545238958538496e-05,
+      "loss": 0.2703,
+      "step": 86505
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7542519884227677e-05,
+      "loss": 0.2612,
+      "step": 86510
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.753980090674732e-05,
+      "loss": 0.2541,
+      "step": 86515
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7537082026132734e-05,
+      "loss": 0.2626,
+      "step": 86520
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7534363242419208e-05,
+      "loss": 0.2525,
+      "step": 86525
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.753164455564206e-05,
+      "loss": 0.2516,
+      "step": 86530
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.752892596583659e-05,
+      "loss": 0.2643,
+      "step": 86535
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7526207473038096e-05,
+      "loss": 0.2558,
+      "step": 86540
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.752348907728186e-05,
+      "loss": 0.2617,
+      "step": 86545
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7520770778603203e-05,
+      "loss": 0.2511,
+      "step": 86550
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7518052577037404e-05,
+      "loss": 0.271,
+      "step": 86555
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.751533447261975e-05,
+      "loss": 0.2645,
+      "step": 86560
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7512616465385562e-05,
+      "loss": 0.2534,
+      "step": 86565
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7509898555370113e-05,
+      "loss": 0.2519,
+      "step": 86570
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7507180742608695e-05,
+      "loss": 0.2461,
+      "step": 86575
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7504463027136585e-05,
+      "loss": 0.249,
+      "step": 86580
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.750174540898909e-05,
+      "loss": 0.2673,
+      "step": 86585
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7499027888201496e-05,
+      "loss": 0.2567,
+      "step": 86590
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.749631046480908e-05,
+      "loss": 0.2565,
+      "step": 86595
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7493593138847113e-05,
+      "loss": 0.2932,
+      "step": 86600
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.749087591035091e-05,
+      "loss": 0.2615,
+      "step": 86605
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7488158779355724e-05,
+      "loss": 0.2538,
+      "step": 86610
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7485441745896853e-05,
+      "loss": 0.2511,
+      "step": 86615
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.748272481000956e-05,
+      "loss": 0.2427,
+      "step": 86620
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.748000797172914e-05,
+      "loss": 0.2435,
+      "step": 86625
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7477291231090857e-05,
+      "loss": 0.242,
+      "step": 86630
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7474574588129987e-05,
+      "loss": 0.2656,
+      "step": 86635
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7471858042881805e-05,
+      "loss": 0.2405,
+      "step": 86640
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7469141595381594e-05,
+      "loss": 0.2518,
+      "step": 86645
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7466425245664614e-05,
+      "loss": 0.2752,
+      "step": 86650
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7463708993766137e-05,
+      "loss": 0.2634,
+      "step": 86655
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7460992839721422e-05,
+      "loss": 0.2472,
+      "step": 86660
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7458276783565754e-05,
+      "loss": 0.2352,
+      "step": 86665
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.74555608253344e-05,
+      "loss": 0.2356,
+      "step": 86670
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7452844965062598e-05,
+      "loss": 0.2326,
+      "step": 86675
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7450129202785647e-05,
+      "loss": 0.2642,
+      "step": 86680
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.744741353853879e-05,
+      "loss": 0.2588,
+      "step": 86685
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7444697972357287e-05,
+      "loss": 0.2801,
+      "step": 86690
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.74419825042764e-05,
+      "loss": 0.2571,
+      "step": 86695
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7439267134331393e-05,
+      "loss": 0.2463,
+      "step": 86700
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.743655186255752e-05,
+      "loss": 0.2573,
+      "step": 86705
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.743383668899004e-05,
+      "loss": 0.2593,
+      "step": 86710
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7431121613664193e-05,
+      "loss": 0.257,
+      "step": 86715
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7428406636615252e-05,
+      "loss": 0.2623,
+      "step": 86720
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.742569175787846e-05,
+      "loss": 0.2597,
+      "step": 86725
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.742297697748907e-05,
+      "loss": 0.2608,
+      "step": 86730
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.742026229548232e-05,
+      "loss": 0.2469,
+      "step": 86735
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7417547711893485e-05,
+      "loss": 0.2675,
+      "step": 86740
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7414833226757786e-05,
+      "loss": 0.2315,
+      "step": 86745
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7412118840110484e-05,
+      "loss": 0.2394,
+      "step": 86750
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.740940455198681e-05,
+      "loss": 0.267,
+      "step": 86755
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7406690362422023e-05,
+      "loss": 0.2702,
+      "step": 86760
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7403976271451358e-05,
+      "loss": 0.2547,
+      "step": 86765
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7401805069686268e-05,
+      "loss": 0.2618,
+      "step": 86770
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7399091156273828e-05,
+      "loss": 0.2611,
+      "step": 86775
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7396377341554176e-05,
+      "loss": 0.2334,
+      "step": 86780
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7393663625562544e-05,
+      "loss": 0.2733,
+      "step": 86785
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7390950008334194e-05,
+      "loss": 0.2609,
+      "step": 86790
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.738823648990434e-05,
+      "loss": 0.2555,
+      "step": 86795
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7385523070308218e-05,
+      "loss": 0.2658,
+      "step": 86800
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.738280974958105e-05,
+      "loss": 0.2902,
+      "step": 86805
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.738009652775809e-05,
+      "loss": 0.2581,
+      "step": 86810
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.737738340487455e-05,
+      "loss": 0.2536,
+      "step": 86815
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7374670380965668e-05,
+      "loss": 0.2576,
+      "step": 86820
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7371957456066657e-05,
+      "loss": 0.2596,
+      "step": 86825
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7369244630212757e-05,
+      "loss": 0.2531,
+      "step": 86830
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7366531903439184e-05,
+      "loss": 0.2439,
+      "step": 86835
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7363819275781168e-05,
+      "loss": 0.2591,
+      "step": 86840
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7361106747273918e-05,
+      "loss": 0.2785,
+      "step": 86845
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7358394317952668e-05,
+      "loss": 0.2839,
+      "step": 86850
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.735568198785264e-05,
+      "loss": 0.2684,
+      "step": 86855
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7352969757009025e-05,
+      "loss": 0.2733,
+      "step": 86860
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.735025762545707e-05,
+      "loss": 0.2549,
+      "step": 86865
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.734754559323198e-05,
+      "loss": 0.2776,
+      "step": 86870
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7344833660368967e-05,
+      "loss": 0.2678,
+      "step": 86875
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7342121826903235e-05,
+      "loss": 0.2734,
+      "step": 86880
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7339410092870013e-05,
+      "loss": 0.2364,
+      "step": 86885
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7336698458304497e-05,
+      "loss": 0.2606,
+      "step": 86890
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7333986923241906e-05,
+      "loss": 0.2835,
+      "step": 86895
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7331275487717437e-05,
+      "loss": 0.239,
+      "step": 86900
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.732856415176631e-05,
+      "loss": 0.2545,
+      "step": 86905
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.732585291542372e-05,
+      "loss": 0.2363,
+      "step": 86910
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.732314177872487e-05,
+      "loss": 0.2535,
+      "step": 86915
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7320430741704962e-05,
+      "loss": 0.266,
+      "step": 86920
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.73177198043992e-05,
+      "loss": 0.2702,
+      "step": 86925
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7315008966842788e-05,
+      "loss": 0.2557,
+      "step": 86930
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.731229822907092e-05,
+      "loss": 0.2587,
+      "step": 86935
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7309587591118787e-05,
+      "loss": 0.2599,
+      "step": 86940
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7306877053021602e-05,
+      "loss": 0.2537,
+      "step": 86945
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7304166614814538e-05,
+      "loss": 0.2791,
+      "step": 86950
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7301456276532808e-05,
+      "loss": 0.2666,
+      "step": 86955
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.729874603821158e-05,
+      "loss": 0.2499,
+      "step": 86960
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.729603589988607e-05,
+      "loss": 0.2402,
+      "step": 86965
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7293325861591456e-05,
+      "loss": 0.2629,
+      "step": 86970
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.729061592336293e-05,
+      "loss": 0.2554,
+      "step": 86975
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7287906085235662e-05,
+      "loss": 0.2523,
+      "step": 86980
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7285196347244868e-05,
+      "loss": 0.2578,
+      "step": 86985
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.728248670942571e-05,
+      "loss": 0.2509,
+      "step": 86990
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7279777171813366e-05,
+      "loss": 0.2498,
+      "step": 86995
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7277067734443038e-05,
+      "loss": 0.2589,
+      "step": 87000
+    },
+    {
+      "epoch": 3.06,
+      "eval_loss": 0.25700682401657104,
+      "eval_runtime": 10.5666,
+      "eval_samples_per_second": 9.464,
+      "eval_steps_per_second": 9.464,
+      "step": 87000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7274358397349895e-05,
+      "loss": 0.2571,
+      "step": 87005
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.727164916056912e-05,
+      "loss": 0.2686,
+      "step": 87010
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.726894002413587e-05,
+      "loss": 0.2568,
+      "step": 87015
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.726623098808536e-05,
+      "loss": 0.2401,
+      "step": 87020
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7263522052452734e-05,
+      "loss": 0.2597,
+      "step": 87025
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7260813217273178e-05,
+      "loss": 0.2436,
+      "step": 87030
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7258104482581854e-05,
+      "loss": 0.2491,
+      "step": 87035
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7255395848413954e-05,
+      "loss": 0.25,
+      "step": 87040
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.725268731480463e-05,
+      "loss": 0.2528,
+      "step": 87045
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7249978881789053e-05,
+      "loss": 0.2766,
+      "step": 87050
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7247270549402393e-05,
+      "loss": 0.2714,
+      "step": 87055
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7244562317679816e-05,
+      "loss": 0.2443,
+      "step": 87060
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7241854186656493e-05,
+      "loss": 0.2568,
+      "step": 87065
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.723914615636758e-05,
+      "loss": 0.2646,
+      "step": 87070
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7236438226848224e-05,
+      "loss": 0.2628,
+      "step": 87075
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7233730398133616e-05,
+      "loss": 0.2443,
+      "step": 87080
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7231022670258894e-05,
+      "loss": 0.2564,
+      "step": 87085
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.722831504325923e-05,
+      "loss": 0.2679,
+      "step": 87090
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.722560751716976e-05,
+      "loss": 0.2628,
+      "step": 87095
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7222900092025668e-05,
+      "loss": 0.2385,
+      "step": 87100
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7220192767862085e-05,
+      "loss": 0.2463,
+      "step": 87105
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7217485544714167e-05,
+      "loss": 0.2827,
+      "step": 87110
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7214778422617077e-05,
+      "loss": 0.2391,
+      "step": 87115
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7212071401605962e-05,
+      "loss": 0.2367,
+      "step": 87120
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7209364481715968e-05,
+      "loss": 0.2658,
+      "step": 87125
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7206657662982227e-05,
+      "loss": 0.245,
+      "step": 87130
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7203950945439917e-05,
+      "loss": 0.261,
+      "step": 87135
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.720124432912416e-05,
+      "loss": 0.2542,
+      "step": 87140
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.719853781407011e-05,
+      "loss": 0.2666,
+      "step": 87145
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.719583140031289e-05,
+      "loss": 0.2677,
+      "step": 87150
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7193125087887673e-05,
+      "loss": 0.27,
+      "step": 87155
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7190418876829577e-05,
+      "loss": 0.2322,
+      "step": 87160
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7187712767173747e-05,
+      "loss": 0.2537,
+      "step": 87165
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.718500675895531e-05,
+      "loss": 0.2604,
+      "step": 87170
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7182300852209416e-05,
+      "loss": 0.2583,
+      "step": 87175
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7179595046971196e-05,
+      "loss": 0.2391,
+      "step": 87180
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.717688934327578e-05,
+      "loss": 0.2704,
+      "step": 87185
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.717418374115829e-05,
+      "loss": 0.2694,
+      "step": 87190
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.717147824065387e-05,
+      "loss": 0.2692,
+      "step": 87195
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7168772841797658e-05,
+      "loss": 0.2604,
+      "step": 87200
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7166067544624765e-05,
+      "loss": 0.2707,
+      "step": 87205
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.716336234917031e-05,
+      "loss": 0.2606,
+      "step": 87210
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7160657255469442e-05,
+      "loss": 0.236,
+      "step": 87215
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.715795226355727e-05,
+      "loss": 0.2382,
+      "step": 87220
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7155247373468912e-05,
+      "loss": 0.2898,
+      "step": 87225
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.715254258523951e-05,
+      "loss": 0.2667,
+      "step": 87230
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.714983789890417e-05,
+      "loss": 0.2648,
+      "step": 87235
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7147133314498008e-05,
+      "loss": 0.2587,
+      "step": 87240
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.714442883205614e-05,
+      "loss": 0.2615,
+      "step": 87245
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.714172445161369e-05,
+      "loss": 0.2606,
+      "step": 87250
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.713902017320577e-05,
+      "loss": 0.2385,
+      "step": 87255
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7136315996867497e-05,
+      "loss": 0.243,
+      "step": 87260
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.713361192263397e-05,
+      "loss": 0.2508,
+      "step": 87265
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.713090795054031e-05,
+      "loss": 0.2614,
+      "step": 87270
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7128204080621627e-05,
+      "loss": 0.2454,
+      "step": 87275
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.712550031291303e-05,
+      "loss": 0.2599,
+      "step": 87280
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.712279664744961e-05,
+      "loss": 0.2705,
+      "step": 87285
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7120093084266493e-05,
+      "loss": 0.2621,
+      "step": 87290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7117389623398776e-05,
+      "loss": 0.242,
+      "step": 87295
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7114686264881557e-05,
+      "loss": 0.2857,
+      "step": 87300
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7111983008749937e-05,
+      "loss": 0.2446,
+      "step": 87305
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7109279855039023e-05,
+      "loss": 0.2541,
+      "step": 87310
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7106576803783913e-05,
+      "loss": 0.2635,
+      "step": 87315
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.71038738550197e-05,
+      "loss": 0.2522,
+      "step": 87320
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.710117100878147e-05,
+      "loss": 0.2472,
+      "step": 87325
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7098468265104336e-05,
+      "loss": 0.2591,
+      "step": 87330
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.709576562402339e-05,
+      "loss": 0.2769,
+      "step": 87335
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.709306308557371e-05,
+      "loss": 0.285,
+      "step": 87340
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7090360649790398e-05,
+      "loss": 0.2444,
+      "step": 87345
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.708765831670855e-05,
+      "loss": 0.2723,
+      "step": 87350
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7084956086363228e-05,
+      "loss": 0.2508,
+      "step": 87355
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7082253958789545e-05,
+      "loss": 0.2719,
+      "step": 87360
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.707955193402257e-05,
+      "loss": 0.2668,
+      "step": 87365
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7076850012097405e-05,
+      "loss": 0.2523,
+      "step": 87370
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7074148193049116e-05,
+      "loss": 0.2612,
+      "step": 87375
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7071446476912776e-05,
+      "loss": 0.2518,
+      "step": 87380
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7068744863723496e-05,
+      "loss": 0.2548,
+      "step": 87385
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.706604335351633e-05,
+      "loss": 0.2838,
+      "step": 87390
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7063341946326366e-05,
+      "loss": 0.2316,
+      "step": 87395
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7060640642188664e-05,
+      "loss": 0.2461,
+      "step": 87400
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7057939441138328e-05,
+      "loss": 0.2472,
+      "step": 87405
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.70552383432104e-05,
+      "loss": 0.2563,
+      "step": 87410
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7052537348439977e-05,
+      "loss": 0.2647,
+      "step": 87415
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.704983645686211e-05,
+      "loss": 0.2284,
+      "step": 87420
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.704713566851188e-05,
+      "loss": 0.2526,
+      "step": 87425
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7044434983424363e-05,
+      "loss": 0.2665,
+      "step": 87430
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.70417344016346e-05,
+      "loss": 0.2559,
+      "step": 87435
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7039033923177676e-05,
+      "loss": 0.255,
+      "step": 87440
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7036333548088652e-05,
+      "loss": 0.2554,
+      "step": 87445
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.703363327640259e-05,
+      "loss": 0.2613,
+      "step": 87450
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7030933108154544e-05,
+      "loss": 0.2755,
+      "step": 87455
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7028233043379584e-05,
+      "loss": 0.2605,
+      "step": 87460
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7025533082112768e-05,
+      "loss": 0.2295,
+      "step": 87465
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7022833224389144e-05,
+      "loss": 0.2576,
+      "step": 87470
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7020133470243768e-05,
+      "loss": 0.27,
+      "step": 87475
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7017433819711707e-05,
+      "loss": 0.2392,
+      "step": 87480
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7014734272828016e-05,
+      "loss": 0.2587,
+      "step": 87485
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7012034829627727e-05,
+      "loss": 0.252,
+      "step": 87490
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.70093354901459e-05,
+      "loss": 0.2711,
+      "step": 87495
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7006636254417583e-05,
+      "loss": 0.275,
+      "step": 87500
+    },
+    {
+      "epoch": 3.08,
+      "eval_loss": 0.25675198435783386,
+      "eval_runtime": 10.5938,
+      "eval_samples_per_second": 9.439,
+      "eval_steps_per_second": 9.439,
+      "step": 87500
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.700393712247784e-05,
+      "loss": 0.2602,
+      "step": 87505
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.70012380943617e-05,
+      "loss": 0.2764,
+      "step": 87510
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.69985391701042e-05,
+      "loss": 0.2733,
+      "step": 87515
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.699584034974041e-05,
+      "loss": 0.2469,
+      "step": 87520
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.699314163330535e-05,
+      "loss": 0.2683,
+      "step": 87525
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6990443020834073e-05,
+      "loss": 0.2443,
+      "step": 87530
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6987744512361607e-05,
+      "loss": 0.2555,
+      "step": 87535
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.698504610792301e-05,
+      "loss": 0.2549,
+      "step": 87540
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.69823478075533e-05,
+      "loss": 0.268,
+      "step": 87545
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.697964961128752e-05,
+      "loss": 0.2675,
+      "step": 87550
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.69769515191607e-05,
+      "loss": 0.2432,
+      "step": 87555
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6974253531207886e-05,
+      "loss": 0.2515,
+      "step": 87560
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6971555647464103e-05,
+      "loss": 0.2636,
+      "step": 87565
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6968857867964367e-05,
+      "loss": 0.2557,
+      "step": 87570
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6966160192743727e-05,
+      "loss": 0.2648,
+      "step": 87575
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.69634626218372e-05,
+      "loss": 0.2296,
+      "step": 87580
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6960765155279818e-05,
+      "loss": 0.2627,
+      "step": 87585
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.695806779310659e-05,
+      "loss": 0.2599,
+      "step": 87590
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6955370535352566e-05,
+      "loss": 0.2587,
+      "step": 87595
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.695267338205275e-05,
+      "loss": 0.2623,
+      "step": 87600
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.694997633324217e-05,
+      "loss": 0.2587,
+      "step": 87605
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.694727938895583e-05,
+      "loss": 0.2652,
+      "step": 87610
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.694458254922877e-05,
+      "loss": 0.2439,
+      "step": 87615
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6941885814096004e-05,
+      "loss": 0.245,
+      "step": 87620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6939189183592537e-05,
+      "loss": 0.2677,
+      "step": 87625
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6936492657753373e-05,
+      "loss": 0.266,
+      "step": 87630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.693379623661354e-05,
+      "loss": 0.253,
+      "step": 87635
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6931099920208058e-05,
+      "loss": 0.2574,
+      "step": 87640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6928403708571924e-05,
+      "loss": 0.2643,
+      "step": 87645
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6925707601740136e-05,
+      "loss": 0.2571,
+      "step": 87650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.692301159974773e-05,
+      "loss": 0.2617,
+      "step": 87655
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6920315702629685e-05,
+      "loss": 0.2567,
+      "step": 87660
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6917619910421025e-05,
+      "loss": 0.2675,
+      "step": 87665
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6914924223156727e-05,
+      "loss": 0.2791,
+      "step": 87670
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6912228640871823e-05,
+      "loss": 0.26,
+      "step": 87675
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6909533163601305e-05,
+      "loss": 0.2729,
+      "step": 87680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6906837791380152e-05,
+      "loss": 0.2604,
+      "step": 87685
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6904142524243383e-05,
+      "loss": 0.2677,
+      "step": 87690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6901447362225992e-05,
+      "loss": 0.2759,
+      "step": 87695
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.689875230536297e-05,
+      "loss": 0.2578,
+      "step": 87700
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6896057353689305e-05,
+      "loss": 0.2391,
+      "step": 87705
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6893362507239998e-05,
+      "loss": 0.252,
+      "step": 87710
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6890667766050032e-05,
+      "loss": 0.2659,
+      "step": 87715
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.688797313015441e-05,
+      "loss": 0.2584,
+      "step": 87720
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.68852785995881e-05,
+      "loss": 0.2694,
+      "step": 87725
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6882584174386106e-05,
+      "loss": 0.2415,
+      "step": 87730
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.687988985458341e-05,
+      "loss": 0.2396,
+      "step": 87735
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6877195640214987e-05,
+      "loss": 0.2591,
+      "step": 87740
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6874501531315828e-05,
+      "loss": 0.2512,
+      "step": 87745
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6871807527920907e-05,
+      "loss": 0.2601,
+      "step": 87750
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6869113630065217e-05,
+      "loss": 0.2553,
+      "step": 87755
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6866419837783727e-05,
+      "loss": 0.2651,
+      "step": 87760
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6863726151111403e-05,
+      "loss": 0.244,
+      "step": 87765
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.686103257008324e-05,
+      "loss": 0.2777,
+      "step": 87770
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6858339094734212e-05,
+      "loss": 0.2576,
+      "step": 87775
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6855645725099283e-05,
+      "loss": 0.2737,
+      "step": 87780
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6852952461213416e-05,
+      "loss": 0.2445,
+      "step": 87785
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6850259303111606e-05,
+      "loss": 0.2697,
+      "step": 87790
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.68475662508288e-05,
+      "loss": 0.247,
+      "step": 87795
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6844873304399977e-05,
+      "loss": 0.2542,
+      "step": 87800
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6842180463860095e-05,
+      "loss": 0.2503,
+      "step": 87805
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.683948772924413e-05,
+      "loss": 0.2418,
+      "step": 87810
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.683679510058704e-05,
+      "loss": 0.2551,
+      "step": 87815
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6834102577923777e-05,
+      "loss": 0.275,
+      "step": 87820
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6831410161289317e-05,
+      "loss": 0.2568,
+      "step": 87825
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.682871785071861e-05,
+      "loss": 0.2469,
+      "step": 87830
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6826025646246623e-05,
+      "loss": 0.2618,
+      "step": 87835
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6823333547908292e-05,
+      "loss": 0.2836,
+      "step": 87840
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.68206415557386e-05,
+      "loss": 0.2508,
+      "step": 87845
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6817949669772483e-05,
+      "loss": 0.246,
+      "step": 87850
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6815257890044898e-05,
+      "loss": 0.2478,
+      "step": 87855
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6812566216590786e-05,
+      "loss": 0.259,
+      "step": 87860
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6809874649445116e-05,
+      "loss": 0.2453,
+      "step": 87865
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6807183188642827e-05,
+      "loss": 0.249,
+      "step": 87870
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6804491834218864e-05,
+      "loss": 0.2558,
+      "step": 87875
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6801800586208168e-05,
+      "loss": 0.2504,
+      "step": 87880
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.679910944464569e-05,
+      "loss": 0.2629,
+      "step": 87885
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.679641840956638e-05,
+      "loss": 0.2806,
+      "step": 87890
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6793727481005168e-05,
+      "loss": 0.2629,
+      "step": 87895
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6791036658996984e-05,
+      "loss": 0.2511,
+      "step": 87900
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.678834594357679e-05,
+      "loss": 0.2563,
+      "step": 87905
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.678565533477951e-05,
+      "loss": 0.2638,
+      "step": 87910
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6782964832640075e-05,
+      "loss": 0.2509,
+      "step": 87915
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6780274437193435e-05,
+      "loss": 0.2488,
+      "step": 87920
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6777584148474518e-05,
+      "loss": 0.2512,
+      "step": 87925
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6774893966518242e-05,
+      "loss": 0.2355,
+      "step": 87930
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.677220389135955e-05,
+      "loss": 0.2639,
+      "step": 87935
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.676951392303337e-05,
+      "loss": 0.2628,
+      "step": 87940
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6766824061574632e-05,
+      "loss": 0.2853,
+      "step": 87945
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.676413430701826e-05,
+      "loss": 0.2306,
+      "step": 87950
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6761444659399163e-05,
+      "loss": 0.2469,
+      "step": 87955
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6758755118752284e-05,
+      "loss": 0.2654,
+      "step": 87960
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6756065685112538e-05,
+      "loss": 0.2666,
+      "step": 87965
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6753376358514852e-05,
+      "loss": 0.2589,
+      "step": 87970
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6750687138994125e-05,
+      "loss": 0.2552,
+      "step": 87975
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6747998026585306e-05,
+      "loss": 0.2464,
+      "step": 87980
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6745309021323287e-05,
+      "loss": 0.2492,
+      "step": 87985
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.674262012324299e-05,
+      "loss": 0.2546,
+      "step": 87990
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.673993133237932e-05,
+      "loss": 0.2565,
+      "step": 87995
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6737242648767215e-05,
+      "loss": 0.2637,
+      "step": 88000
+    },
+    {
+      "epoch": 3.1,
+      "eval_loss": 0.25597599148750305,
+      "eval_runtime": 10.5513,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 88000
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6734554072441563e-05,
+      "loss": 0.2452,
+      "step": 88005
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.673186560343728e-05,
+      "loss": 0.2522,
+      "step": 88010
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.672917724178926e-05,
+      "loss": 0.2383,
+      "step": 88015
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6726488987532434e-05,
+      "loss": 0.2508,
+      "step": 88020
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6723800840701702e-05,
+      "loss": 0.266,
+      "step": 88025
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6721112801331946e-05,
+      "loss": 0.2572,
+      "step": 88030
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.67184248694581e-05,
+      "loss": 0.2678,
+      "step": 88035
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6715737045115043e-05,
+      "loss": 0.2583,
+      "step": 88040
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6713049328337683e-05,
+      "loss": 0.2586,
+      "step": 88045
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6710361719160905e-05,
+      "loss": 0.2745,
+      "step": 88050
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.670767421761963e-05,
+      "loss": 0.2549,
+      "step": 88055
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.670498682374874e-05,
+      "loss": 0.2587,
+      "step": 88060
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6702299537583126e-05,
+      "loss": 0.2475,
+      "step": 88065
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.669961235915768e-05,
+      "loss": 0.2702,
+      "step": 88070
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6696925288507297e-05,
+      "loss": 0.2693,
+      "step": 88075
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.669423832566688e-05,
+      "loss": 0.2599,
+      "step": 88080
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6691551470671296e-05,
+      "loss": 0.2832,
+      "step": 88085
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6688864723555436e-05,
+      "loss": 0.255,
+      "step": 88090
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6686178084354198e-05,
+      "loss": 0.2701,
+      "step": 88095
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6683491553102454e-05,
+      "loss": 0.2571,
+      "step": 88100
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.66808051298351e-05,
+      "loss": 0.2642,
+      "step": 88105
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.667811881458699e-05,
+      "loss": 0.2762,
+      "step": 88110
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.667543260739304e-05,
+      "loss": 0.2425,
+      "step": 88115
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.667274650828811e-05,
+      "loss": 0.2539,
+      "step": 88120
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6670060517307072e-05,
+      "loss": 0.2692,
+      "step": 88125
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6667374634484806e-05,
+      "loss": 0.2809,
+      "step": 88130
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6664688859856196e-05,
+      "loss": 0.2754,
+      "step": 88135
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.666200319345611e-05,
+      "loss": 0.2544,
+      "step": 88140
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.665931763531942e-05,
+      "loss": 0.2284,
+      "step": 88145
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.665663218548098e-05,
+      "loss": 0.2623,
+      "step": 88150
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.665394684397568e-05,
+      "loss": 0.246,
+      "step": 88155
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6651261610838383e-05,
+      "loss": 0.2496,
+      "step": 88160
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.664857648610394e-05,
+      "loss": 0.2721,
+      "step": 88165
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6645891469807242e-05,
+      "loss": 0.2511,
+      "step": 88170
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6643206561983138e-05,
+      "loss": 0.2559,
+      "step": 88175
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6640521762666482e-05,
+      "loss": 0.2766,
+      "step": 88180
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.663783707189214e-05,
+      "loss": 0.2513,
+      "step": 88185
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6635152489694976e-05,
+      "loss": 0.2542,
+      "step": 88190
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6632468016109848e-05,
+      "loss": 0.26,
+      "step": 88195
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.662978365117161e-05,
+      "loss": 0.2627,
+      "step": 88200
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.66270993949151e-05,
+      "loss": 0.275,
+      "step": 88205
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.66244152473752e-05,
+      "loss": 0.2463,
+      "step": 88210
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.662173120858675e-05,
+      "loss": 0.2575,
+      "step": 88215
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6619047278584596e-05,
+      "loss": 0.2504,
+      "step": 88220
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.661636345740358e-05,
+      "loss": 0.2562,
+      "step": 88225
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.661367974507857e-05,
+      "loss": 0.2457,
+      "step": 88230
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6610996141644397e-05,
+      "loss": 0.2446,
+      "step": 88235
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6608312647135917e-05,
+      "loss": 0.2849,
+      "step": 88240
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6605629261587957e-05,
+      "loss": 0.2788,
+      "step": 88245
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.660294598503538e-05,
+      "loss": 0.2552,
+      "step": 88250
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.660026281751301e-05,
+      "loss": 0.2568,
+      "step": 88255
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6597579759055688e-05,
+      "loss": 0.2444,
+      "step": 88260
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.659489680969826e-05,
+      "loss": 0.2542,
+      "step": 88265
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6592213969475557e-05,
+      "loss": 0.2692,
+      "step": 88270
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6589531238422416e-05,
+      "loss": 0.2584,
+      "step": 88275
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.658684861657366e-05,
+      "loss": 0.2641,
+      "step": 88280
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6584166103964143e-05,
+      "loss": 0.2619,
+      "step": 88285
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6581483700628676e-05,
+      "loss": 0.2526,
+      "step": 88290
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.65788014066021e-05,
+      "loss": 0.2417,
+      "step": 88295
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6576119221919228e-05,
+      "loss": 0.2669,
+      "step": 88300
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6573437146614906e-05,
+      "loss": 0.2453,
+      "step": 88305
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.657075518072395e-05,
+      "loss": 0.2621,
+      "step": 88310
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6568073324281178e-05,
+      "loss": 0.2774,
+      "step": 88315
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6565391577321415e-05,
+      "loss": 0.2645,
+      "step": 88320
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6562709939879486e-05,
+      "loss": 0.245,
+      "step": 88325
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6560028411990213e-05,
+      "loss": 0.2613,
+      "step": 88330
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6557346993688412e-05,
+      "loss": 0.2542,
+      "step": 88335
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.655466568500888e-05,
+      "loss": 0.2499,
+      "step": 88340
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.655198448598646e-05,
+      "loss": 0.2386,
+      "step": 88345
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6549303396655953e-05,
+      "loss": 0.2671,
+      "step": 88350
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6546622417052174e-05,
+      "loss": 0.2524,
+      "step": 88355
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6543941547209923e-05,
+      "loss": 0.2415,
+      "step": 88360
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6541260787164026e-05,
+      "loss": 0.2601,
+      "step": 88365
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.653858013694928e-05,
+      "loss": 0.2599,
+      "step": 88370
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.65358995966005e-05,
+      "loss": 0.2621,
+      "step": 88375
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6533219166152473e-05,
+      "loss": 0.2534,
+      "step": 88380
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.653053884564003e-05,
+      "loss": 0.25,
+      "step": 88385
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6527858635097955e-05,
+      "loss": 0.2317,
+      "step": 88390
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.652517853456104e-05,
+      "loss": 0.2635,
+      "step": 88395
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6522498544064113e-05,
+      "loss": 0.2467,
+      "step": 88400
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.651981866364195e-05,
+      "loss": 0.2529,
+      "step": 88405
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6517138893329355e-05,
+      "loss": 0.2823,
+      "step": 88410
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.651445923316111e-05,
+      "loss": 0.2919,
+      "step": 88415
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.651177968317203e-05,
+      "loss": 0.2554,
+      "step": 88420
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.65091002433969e-05,
+      "loss": 0.233,
+      "step": 88425
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.65064209138705e-05,
+      "loss": 0.2487,
+      "step": 88430
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.650374169462762e-05,
+      "loss": 0.2457,
+      "step": 88435
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.650106258570307e-05,
+      "loss": 0.254,
+      "step": 88440
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.649838358713162e-05,
+      "loss": 0.2363,
+      "step": 88445
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6495704698948057e-05,
+      "loss": 0.2572,
+      "step": 88450
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.649302592118715e-05,
+      "loss": 0.259,
+      "step": 88455
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.64903472538837e-05,
+      "loss": 0.2359,
+      "step": 88460
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6487668697072493e-05,
+      "loss": 0.2649,
+      "step": 88465
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6484990250788293e-05,
+      "loss": 0.2425,
+      "step": 88470
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6482311915065874e-05,
+      "loss": 0.2795,
+      "step": 88475
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6479633689940032e-05,
+      "loss": 0.2461,
+      "step": 88480
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.647695557544553e-05,
+      "loss": 0.2561,
+      "step": 88485
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6474277571617145e-05,
+      "loss": 0.2542,
+      "step": 88490
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6471599678489634e-05,
+      "loss": 0.2431,
+      "step": 88495
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6468921896097793e-05,
+      "loss": 0.2672,
+      "step": 88500
+    },
+    {
+      "epoch": 3.11,
+      "eval_loss": 0.25621533393859863,
+      "eval_runtime": 10.5456,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 88500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.646624422447638e-05,
+      "loss": 0.2621,
+      "step": 88505
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6463566663660153e-05,
+      "loss": 0.2563,
+      "step": 88510
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6460889213683892e-05,
+      "loss": 0.2306,
+      "step": 88515
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.645821187458236e-05,
+      "loss": 0.2811,
+      "step": 88520
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.645553464639032e-05,
+      "loss": 0.2564,
+      "step": 88525
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6452857529142518e-05,
+      "loss": 0.2592,
+      "step": 88530
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.645018052287374e-05,
+      "loss": 0.2565,
+      "step": 88535
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6447503627618728e-05,
+      "loss": 0.2429,
+      "step": 88540
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6444826843412254e-05,
+      "loss": 0.2409,
+      "step": 88545
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.644215017028905e-05,
+      "loss": 0.2629,
+      "step": 88550
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6439473608283894e-05,
+      "loss": 0.2676,
+      "step": 88555
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.643679715743154e-05,
+      "loss": 0.2427,
+      "step": 88560
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6434120817766723e-05,
+      "loss": 0.2589,
+      "step": 88565
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6431444589324193e-05,
+      "loss": 0.2767,
+      "step": 88570
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6428768472138723e-05,
+      "loss": 0.2622,
+      "step": 88575
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6426092466245044e-05,
+      "loss": 0.2695,
+      "step": 88580
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6423416571677904e-05,
+      "loss": 0.2686,
+      "step": 88585
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.642074078847204e-05,
+      "loss": 0.2474,
+      "step": 88590
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6418065116662202e-05,
+      "loss": 0.2619,
+      "step": 88595
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6415389556283144e-05,
+      "loss": 0.2584,
+      "step": 88600
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6412714107369593e-05,
+      "loss": 0.2847,
+      "step": 88605
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6410038769956277e-05,
+      "loss": 0.2579,
+      "step": 88610
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.640736354407796e-05,
+      "loss": 0.2679,
+      "step": 88615
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6404688429769362e-05,
+      "loss": 0.2467,
+      "step": 88620
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6402013427065215e-05,
+      "loss": 0.2191,
+      "step": 88625
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6399338536000258e-05,
+      "loss": 0.2667,
+      "step": 88630
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.639666375660923e-05,
+      "loss": 0.2494,
+      "step": 88635
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.639398908892685e-05,
+      "loss": 0.2712,
+      "step": 88640
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6391314532987844e-05,
+      "loss": 0.2481,
+      "step": 88645
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6388640088826946e-05,
+      "loss": 0.2413,
+      "step": 88650
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6385965756478887e-05,
+      "loss": 0.2479,
+      "step": 88655
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.638329153597839e-05,
+      "loss": 0.2568,
+      "step": 88660
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6380617427360164e-05,
+      "loss": 0.2788,
+      "step": 88665
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6377943430658942e-05,
+      "loss": 0.2436,
+      "step": 88670
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6375269545909445e-05,
+      "loss": 0.2451,
+      "step": 88675
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.637259577314639e-05,
+      "loss": 0.2385,
+      "step": 88680
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6369922112404488e-05,
+      "loss": 0.2284,
+      "step": 88685
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6367248563718468e-05,
+      "loss": 0.2568,
+      "step": 88690
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6364575127123034e-05,
+      "loss": 0.2702,
+      "step": 88695
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6361901802652896e-05,
+      "loss": 0.2529,
+      "step": 88700
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6359228590342774e-05,
+      "loss": 0.2551,
+      "step": 88705
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.635655549022737e-05,
+      "loss": 0.2418,
+      "step": 88710
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6353882502341404e-05,
+      "loss": 0.2548,
+      "step": 88715
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6351209626719578e-05,
+      "loss": 0.2475,
+      "step": 88720
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.634853686339658e-05,
+      "loss": 0.2506,
+      "step": 88725
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6345864212407136e-05,
+      "loss": 0.2439,
+      "step": 88730
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6343191673785953e-05,
+      "loss": 0.2548,
+      "step": 88735
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6340519247567702e-05,
+      "loss": 0.2592,
+      "step": 88740
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6337846933787116e-05,
+      "loss": 0.2338,
+      "step": 88745
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.633517473247888e-05,
+      "loss": 0.2454,
+      "step": 88750
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6332502643677683e-05,
+      "loss": 0.2575,
+      "step": 88755
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6329830667418228e-05,
+      "loss": 0.2732,
+      "step": 88760
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.632715880373521e-05,
+      "loss": 0.2445,
+      "step": 88765
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.632448705266332e-05,
+      "loss": 0.2622,
+      "step": 88770
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.632181541423725e-05,
+      "loss": 0.2481,
+      "step": 88775
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6319143888491675e-05,
+      "loss": 0.2467,
+      "step": 88780
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.631647247546131e-05,
+      "loss": 0.2521,
+      "step": 88785
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6313801175180822e-05,
+      "loss": 0.2684,
+      "step": 88790
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6311129987684902e-05,
+      "loss": 0.2602,
+      "step": 88795
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6308458913008228e-05,
+      "loss": 0.2543,
+      "step": 88800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6305787951185494e-05,
+      "loss": 0.2528,
+      "step": 88805
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.630311710225137e-05,
+      "loss": 0.2517,
+      "step": 88810
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6300446366240543e-05,
+      "loss": 0.2709,
+      "step": 88815
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6297775743187677e-05,
+      "loss": 0.2588,
+      "step": 88820
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6295105233127468e-05,
+      "loss": 0.2637,
+      "step": 88825
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6292434836094584e-05,
+      "loss": 0.2487,
+      "step": 88830
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6289764552123688e-05,
+      "loss": 0.2458,
+      "step": 88835
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6287094381249457e-05,
+      "loss": 0.2593,
+      "step": 88840
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6284424323506564e-05,
+      "loss": 0.237,
+      "step": 88845
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6281754378929684e-05,
+      "loss": 0.2576,
+      "step": 88850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6279084547553465e-05,
+      "loss": 0.2742,
+      "step": 88855
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.62764148294126e-05,
+      "loss": 0.2643,
+      "step": 88860
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6273745224541735e-05,
+      "loss": 0.2551,
+      "step": 88865
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.627107573297554e-05,
+      "loss": 0.2677,
+      "step": 88870
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6268406354748662e-05,
+      "loss": 0.2393,
+      "step": 88875
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6265737089895783e-05,
+      "loss": 0.2798,
+      "step": 88880
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6263067938451553e-05,
+      "loss": 0.2618,
+      "step": 88885
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.626039890045063e-05,
+      "loss": 0.2606,
+      "step": 88890
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6257729975927656e-05,
+      "loss": 0.2465,
+      "step": 88895
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.62550611649173e-05,
+      "loss": 0.2683,
+      "step": 88900
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.625239246745422e-05,
+      "loss": 0.2552,
+      "step": 88905
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.624972388357306e-05,
+      "loss": 0.2637,
+      "step": 88910
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6247055413308455e-05,
+      "loss": 0.2644,
+      "step": 88915
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.624438705669508e-05,
+      "loss": 0.2455,
+      "step": 88920
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.624171881376756e-05,
+      "loss": 0.2699,
+      "step": 88925
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6239050684560557e-05,
+      "loss": 0.2428,
+      "step": 88930
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6236382669108696e-05,
+      "loss": 0.2529,
+      "step": 88935
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6233714767446645e-05,
+      "loss": 0.2753,
+      "step": 88940
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6231046979609026e-05,
+      "loss": 0.2709,
+      "step": 88945
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.622837930563048e-05,
+      "loss": 0.2455,
+      "step": 88950
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.622571174554564e-05,
+      "loss": 0.2726,
+      "step": 88955
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6223044299389163e-05,
+      "loss": 0.2659,
+      "step": 88960
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6220376967195673e-05,
+      "loss": 0.2429,
+      "step": 88965
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6217709748999787e-05,
+      "loss": 0.2912,
+      "step": 88970
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.621504264483617e-05,
+      "loss": 0.2627,
+      "step": 88975
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.621237565473942e-05,
+      "loss": 0.2782,
+      "step": 88980
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6210242144813453e-05,
+      "loss": 0.27,
+      "step": 88985
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.620757536012436e-05,
+      "loss": 0.2398,
+      "step": 88990
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6204908689599114e-05,
+      "loss": 0.2631,
+      "step": 88995
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6202242133272327e-05,
+      "loss": 0.2362,
+      "step": 89000
+    },
+    {
+      "epoch": 3.13,
+      "eval_loss": 0.25561562180519104,
+      "eval_runtime": 10.5484,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 89000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.619957569117862e-05,
+      "loss": 0.2543,
+      "step": 89005
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6196909363352637e-05,
+      "loss": 0.2474,
+      "step": 89010
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6194243149828974e-05,
+      "loss": 0.2468,
+      "step": 89015
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6191577050642267e-05,
+      "loss": 0.2666,
+      "step": 89020
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6188911065827118e-05,
+      "loss": 0.2764,
+      "step": 89025
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6186245195418162e-05,
+      "loss": 0.2577,
+      "step": 89030
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6183579439449998e-05,
+      "loss": 0.2554,
+      "step": 89035
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6180913797957246e-05,
+      "loss": 0.2567,
+      "step": 89040
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6178248270974515e-05,
+      "loss": 0.273,
+      "step": 89045
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.617558285853642e-05,
+      "loss": 0.2565,
+      "step": 89050
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.617291756067757e-05,
+      "loss": 0.2484,
+      "step": 89055
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6170252377432554e-05,
+      "loss": 0.2745,
+      "step": 89060
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6167587308836007e-05,
+      "loss": 0.2313,
+      "step": 89065
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6164922354922507e-05,
+      "loss": 0.2444,
+      "step": 89070
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6162257515726678e-05,
+      "loss": 0.2472,
+      "step": 89075
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6159592791283097e-05,
+      "loss": 0.2551,
+      "step": 89080
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.615692818162639e-05,
+      "loss": 0.2704,
+      "step": 89085
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6154263686791143e-05,
+      "loss": 0.266,
+      "step": 89090
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6151599306811946e-05,
+      "loss": 0.2748,
+      "step": 89095
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6148935041723396e-05,
+      "loss": 0.2489,
+      "step": 89100
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.61462708915601e-05,
+      "loss": 0.24,
+      "step": 89105
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6143606856356644e-05,
+      "loss": 0.2735,
+      "step": 89110
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6140942936147615e-05,
+      "loss": 0.2626,
+      "step": 89115
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6138279130967594e-05,
+      "loss": 0.2559,
+      "step": 89120
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6135615440851185e-05,
+      "loss": 0.2752,
+      "step": 89125
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6132951865832968e-05,
+      "loss": 0.2571,
+      "step": 89130
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6130288405947534e-05,
+      "loss": 0.2491,
+      "step": 89135
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6127625061229447e-05,
+      "loss": 0.2669,
+      "step": 89140
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.612496183171331e-05,
+      "loss": 0.2549,
+      "step": 89145
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.612229871743369e-05,
+      "loss": 0.27,
+      "step": 89150
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6119635718425163e-05,
+      "loss": 0.2503,
+      "step": 89155
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6116972834722328e-05,
+      "loss": 0.2405,
+      "step": 89160
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6114310066359746e-05,
+      "loss": 0.2495,
+      "step": 89165
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6111647413371992e-05,
+      "loss": 0.2459,
+      "step": 89170
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6108984875793626e-05,
+      "loss": 0.2613,
+      "step": 89175
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.610632245365924e-05,
+      "loss": 0.2814,
+      "step": 89180
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6103660147003402e-05,
+      "loss": 0.2423,
+      "step": 89185
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.610099795586067e-05,
+      "loss": 0.2627,
+      "step": 89190
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6098335880265608e-05,
+      "loss": 0.2506,
+      "step": 89195
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6095673920252795e-05,
+      "loss": 0.2539,
+      "step": 89200
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6093012075856787e-05,
+      "loss": 0.2608,
+      "step": 89205
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6090350347112148e-05,
+      "loss": 0.267,
+      "step": 89210
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.608768873405343e-05,
+      "loss": 0.2623,
+      "step": 89215
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6085027236715212e-05,
+      "loss": 0.2672,
+      "step": 89220
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.608236585513204e-05,
+      "loss": 0.2301,
+      "step": 89225
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6079704589338463e-05,
+      "loss": 0.2536,
+      "step": 89230
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6077043439369037e-05,
+      "loss": 0.2361,
+      "step": 89235
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6074382405258337e-05,
+      "loss": 0.273,
+      "step": 89240
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6071721487040897e-05,
+      "loss": 0.2442,
+      "step": 89245
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.606906068475127e-05,
+      "loss": 0.2665,
+      "step": 89250
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.606639999842399e-05,
+      "loss": 0.2583,
+      "step": 89255
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6063739428093626e-05,
+      "loss": 0.2467,
+      "step": 89260
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6061078973794725e-05,
+      "loss": 0.2623,
+      "step": 89265
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6058418635561805e-05,
+      "loss": 0.2491,
+      "step": 89270
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.605575841342944e-05,
+      "loss": 0.2509,
+      "step": 89275
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6053098307432162e-05,
+      "loss": 0.2567,
+      "step": 89280
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6050438317604496e-05,
+      "loss": 0.2441,
+      "step": 89285
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6047778443980992e-05,
+      "loss": 0.2644,
+      "step": 89290
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6045118686596182e-05,
+      "loss": 0.2314,
+      "step": 89295
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6042459045484614e-05,
+      "loss": 0.2554,
+      "step": 89300
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.603979952068081e-05,
+      "loss": 0.2759,
+      "step": 89305
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6037140112219294e-05,
+      "loss": 0.2714,
+      "step": 89310
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6034480820134615e-05,
+      "loss": 0.2457,
+      "step": 89315
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6031821644461296e-05,
+      "loss": 0.2447,
+      "step": 89320
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6029162585233866e-05,
+      "loss": 0.2676,
+      "step": 89325
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6026503642486835e-05,
+      "loss": 0.2476,
+      "step": 89330
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.602384481625475e-05,
+      "loss": 0.2467,
+      "step": 89335
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.602118610657212e-05,
+      "loss": 0.2636,
+      "step": 89340
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6018527513473476e-05,
+      "loss": 0.2591,
+      "step": 89345
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6015869036993326e-05,
+      "loss": 0.2523,
+      "step": 89350
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.60132106771662e-05,
+      "loss": 0.2732,
+      "step": 89355
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6010552434026615e-05,
+      "loss": 0.2625,
+      "step": 89360
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6007894307609078e-05,
+      "loss": 0.2702,
+      "step": 89365
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6005236297948107e-05,
+      "loss": 0.2705,
+      "step": 89370
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.600257840507821e-05,
+      "loss": 0.2774,
+      "step": 89375
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5999920629033914e-05,
+      "loss": 0.2673,
+      "step": 89380
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5997262969849712e-05,
+      "loss": 0.25,
+      "step": 89385
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.5994605427560112e-05,
+      "loss": 0.2511,
+      "step": 89390
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.599194800219963e-05,
+      "loss": 0.2466,
+      "step": 89395
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.598929069380277e-05,
+      "loss": 0.2701,
+      "step": 89400
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5986633502404024e-05,
+      "loss": 0.2589,
+      "step": 89405
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.598397642803791e-05,
+      "loss": 0.2441,
+      "step": 89410
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.598131947073892e-05,
+      "loss": 0.2767,
+      "step": 89415
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.597866263054155e-05,
+      "loss": 0.2721,
+      "step": 89420
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5976005907480303e-05,
+      "loss": 0.2415,
+      "step": 89425
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5973349301589668e-05,
+      "loss": 0.276,
+      "step": 89430
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5970692812904148e-05,
+      "loss": 0.264,
+      "step": 89435
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5968036441458233e-05,
+      "loss": 0.244,
+      "step": 89440
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.59653801872864e-05,
+      "loss": 0.2484,
+      "step": 89445
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5962724050423168e-05,
+      "loss": 0.2579,
+      "step": 89450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5960068030902997e-05,
+      "loss": 0.2605,
+      "step": 89455
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.595741212876039e-05,
+      "loss": 0.254,
+      "step": 89460
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.595475634402982e-05,
+      "loss": 0.2555,
+      "step": 89465
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5952100676745792e-05,
+      "loss": 0.2482,
+      "step": 89470
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.594944512694276e-05,
+      "loss": 0.2741,
+      "step": 89475
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5946789694655228e-05,
+      "loss": 0.2761,
+      "step": 89480
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5944134379917657e-05,
+      "loss": 0.2633,
+      "step": 89485
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5941479182764542e-05,
+      "loss": 0.236,
+      "step": 89490
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5938824103230344e-05,
+      "loss": 0.2626,
+      "step": 89495
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5936169141349548e-05,
+      "loss": 0.2246,
+      "step": 89500
+    },
+    {
+      "epoch": 3.15,
+      "eval_loss": 0.25540173053741455,
+      "eval_runtime": 10.5366,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 89500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5933514297156614e-05,
+      "loss": 0.2681,
+      "step": 89505
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.593085957068603e-05,
+      "loss": 0.2602,
+      "step": 89510
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5928204961972257e-05,
+      "loss": 0.2618,
+      "step": 89515
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5925550471049756e-05,
+      "loss": 0.2692,
+      "step": 89520
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.592289609795301e-05,
+      "loss": 0.2481,
+      "step": 89525
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5920241842716477e-05,
+      "loss": 0.2424,
+      "step": 89530
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5917587705374614e-05,
+      "loss": 0.2458,
+      "step": 89535
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5914933685961886e-05,
+      "loss": 0.266,
+      "step": 89540
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5912279784512767e-05,
+      "loss": 0.2782,
+      "step": 89545
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5909626001061707e-05,
+      "loss": 0.2647,
+      "step": 89550
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.590697233564316e-05,
+      "loss": 0.2916,
+      "step": 89555
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5904318788291577e-05,
+      "loss": 0.2814,
+      "step": 89560
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5901665359041425e-05,
+      "loss": 0.2549,
+      "step": 89565
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5899012047927155e-05,
+      "loss": 0.2596,
+      "step": 89570
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5896358854983216e-05,
+      "loss": 0.2437,
+      "step": 89575
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.589370578024405e-05,
+      "loss": 0.2533,
+      "step": 89580
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.589105282374412e-05,
+      "loss": 0.2313,
+      "step": 89585
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5888399985517863e-05,
+      "loss": 0.2523,
+      "step": 89590
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.588574726559973e-05,
+      "loss": 0.2657,
+      "step": 89595
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5883094664024155e-05,
+      "loss": 0.2559,
+      "step": 89600
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.58804421808256e-05,
+      "loss": 0.2689,
+      "step": 89605
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.587778981603849e-05,
+      "loss": 0.25,
+      "step": 89610
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.587513756969726e-05,
+      "loss": 0.2579,
+      "step": 89615
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5872485441836353e-05,
+      "loss": 0.2435,
+      "step": 89620
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.586983343249022e-05,
+      "loss": 0.2647,
+      "step": 89625
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5867181541693278e-05,
+      "loss": 0.2701,
+      "step": 89630
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.586452976947996e-05,
+      "loss": 0.2476,
+      "step": 89635
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.586187811588471e-05,
+      "loss": 0.2859,
+      "step": 89640
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5859226580941945e-05,
+      "loss": 0.2638,
+      "step": 89645
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5856575164686104e-05,
+      "loss": 0.2614,
+      "step": 89650
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5853923867151596e-05,
+      "loss": 0.243,
+      "step": 89655
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.585127268837287e-05,
+      "loss": 0.2492,
+      "step": 89660
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.584862162838434e-05,
+      "loss": 0.2499,
+      "step": 89665
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.5845970687220424e-05,
+      "loss": 0.2493,
+      "step": 89670
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.584331986491554e-05,
+      "loss": 0.2587,
+      "step": 89675
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5840669161504125e-05,
+      "loss": 0.2524,
+      "step": 89680
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.583801857702058e-05,
+      "loss": 0.2829,
+      "step": 89685
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5835368111499328e-05,
+      "loss": 0.2596,
+      "step": 89690
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5832717764974772e-05,
+      "loss": 0.2582,
+      "step": 89695
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5830067537481337e-05,
+      "loss": 0.2688,
+      "step": 89700
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.582741742905344e-05,
+      "loss": 0.243,
+      "step": 89705
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.582476743972548e-05,
+      "loss": 0.2272,
+      "step": 89710
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.582211756953186e-05,
+      "loss": 0.2525,
+      "step": 89715
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5819467818507005e-05,
+      "loss": 0.2644,
+      "step": 89720
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5816818186685306e-05,
+      "loss": 0.2635,
+      "step": 89725
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5814168674101174e-05,
+      "loss": 0.2387,
+      "step": 89730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5811519280789e-05,
+      "loss": 0.2879,
+      "step": 89735
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.58088700067832e-05,
+      "loss": 0.2594,
+      "step": 89740
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5806220852118166e-05,
+      "loss": 0.2604,
+      "step": 89745
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5803571816828287e-05,
+      "loss": 0.2514,
+      "step": 89750
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5800922900947973e-05,
+      "loss": 0.2494,
+      "step": 89755
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5798274104511614e-05,
+      "loss": 0.2555,
+      "step": 89760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5795625427553606e-05,
+      "loss": 0.2639,
+      "step": 89765
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5792976870108324e-05,
+      "loss": 0.2454,
+      "step": 89770
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5790328432210176e-05,
+      "loss": 0.253,
+      "step": 89775
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5787680113893544e-05,
+      "loss": 0.2621,
+      "step": 89780
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5785031915192818e-05,
+      "loss": 0.2478,
+      "step": 89785
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.578238383614237e-05,
+      "loss": 0.2364,
+      "step": 89790
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5779735876776604e-05,
+      "loss": 0.2658,
+      "step": 89795
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5777088037129888e-05,
+      "loss": 0.2406,
+      "step": 89800
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5774440317236606e-05,
+      "loss": 0.2387,
+      "step": 89805
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5771792717131133e-05,
+      "loss": 0.2783,
+      "step": 89810
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.576914523684785e-05,
+      "loss": 0.2633,
+      "step": 89815
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.576649787642114e-05,
+      "loss": 0.2534,
+      "step": 89820
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5763850635885367e-05,
+      "loss": 0.2456,
+      "step": 89825
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5761203515274898e-05,
+      "loss": 0.2555,
+      "step": 89830
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5758556514624118e-05,
+      "loss": 0.2793,
+      "step": 89835
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.575590963396739e-05,
+      "loss": 0.259,
+      "step": 89840
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5753262873339093e-05,
+      "loss": 0.2493,
+      "step": 89845
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.575061623277357e-05,
+      "loss": 0.2773,
+      "step": 89850
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.574796971230521e-05,
+      "loss": 0.2326,
+      "step": 89855
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.574532331196836e-05,
+      "loss": 0.2391,
+      "step": 89860
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5742677031797388e-05,
+      "loss": 0.2515,
+      "step": 89865
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5740030871826656e-05,
+      "loss": 0.2605,
+      "step": 89870
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5737384832090523e-05,
+      "loss": 0.2575,
+      "step": 89875
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.573473891262334e-05,
+      "loss": 0.2595,
+      "step": 89880
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5732093113459466e-05,
+      "loss": 0.242,
+      "step": 89885
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.572944743463326e-05,
+      "loss": 0.2755,
+      "step": 89890
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5726801876179066e-05,
+      "loss": 0.2453,
+      "step": 89895
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.572415643813124e-05,
+      "loss": 0.2467,
+      "step": 89900
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5721511120524117e-05,
+      "loss": 0.2782,
+      "step": 89905
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5718865923392073e-05,
+      "loss": 0.2259,
+      "step": 89910
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5716220846769434e-05,
+      "loss": 0.2757,
+      "step": 89915
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5713575890690553e-05,
+      "loss": 0.2559,
+      "step": 89920
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.571093105518976e-05,
+      "loss": 0.2471,
+      "step": 89925
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.570828634030141e-05,
+      "loss": 0.2502,
+      "step": 89930
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5705641746059843e-05,
+      "loss": 0.2415,
+      "step": 89935
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5702997272499387e-05,
+      "loss": 0.2449,
+      "step": 89940
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5700352919654386e-05,
+      "loss": 0.2644,
+      "step": 89945
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.569770868755917e-05,
+      "loss": 0.2435,
+      "step": 89950
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.5695064576248088e-05,
+      "loss": 0.2489,
+      "step": 89955
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.569242058575545e-05,
+      "loss": 0.2684,
+      "step": 89960
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5689776716115594e-05,
+      "loss": 0.2492,
+      "step": 89965
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.568713296736286e-05,
+      "loss": 0.266,
+      "step": 89970
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5684489339531563e-05,
+      "loss": 0.2666,
+      "step": 89975
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.568184583265603e-05,
+      "loss": 0.2632,
+      "step": 89980
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.567920244677059e-05,
+      "loss": 0.2733,
+      "step": 89985
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.567655918190957e-05,
+      "loss": 0.2579,
+      "step": 89990
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5673916038107283e-05,
+      "loss": 0.2433,
+      "step": 89995
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5671273015398042e-05,
+      "loss": 0.2469,
+      "step": 90000
+    },
+    {
+      "epoch": 3.17,
+      "eval_loss": 0.2552323043346405,
+      "eval_runtime": 10.5438,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 90000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5668630113816173e-05,
+      "loss": 0.2563,
+      "step": 90005
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5665987333396003e-05,
+      "loss": 0.256,
+      "step": 90010
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5663344674171832e-05,
+      "loss": 0.2445,
+      "step": 90015
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5660702136177968e-05,
+      "loss": 0.253,
+      "step": 90020
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5658059719448745e-05,
+      "loss": 0.2588,
+      "step": 90025
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5655417424018455e-05,
+      "loss": 0.2653,
+      "step": 90030
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5652775249921413e-05,
+      "loss": 0.2451,
+      "step": 90035
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5650133197191914e-05,
+      "loss": 0.2793,
+      "step": 90040
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5647491265864287e-05,
+      "loss": 0.2702,
+      "step": 90045
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.564484945597282e-05,
+      "loss": 0.2383,
+      "step": 90050
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5642207767551814e-05,
+      "loss": 0.2614,
+      "step": 90055
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5639566200635565e-05,
+      "loss": 0.2682,
+      "step": 90060
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5636924755258396e-05,
+      "loss": 0.2693,
+      "step": 90065
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5634283431454582e-05,
+      "loss": 0.2385,
+      "step": 90070
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.563164222925843e-05,
+      "loss": 0.2838,
+      "step": 90075
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5629001148704218e-05,
+      "loss": 0.263,
+      "step": 90080
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5626360189826257e-05,
+      "loss": 0.2606,
+      "step": 90085
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5623719352658833e-05,
+      "loss": 0.2521,
+      "step": 90090
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5621078637236224e-05,
+      "loss": 0.239,
+      "step": 90095
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5618438043592744e-05,
+      "loss": 0.2759,
+      "step": 90100
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5615797571762658e-05,
+      "loss": 0.2607,
+      "step": 90105
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.561315722178025e-05,
+      "loss": 0.2598,
+      "step": 90110
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.561051699367981e-05,
+      "loss": 0.2588,
+      "step": 90115
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5607876887495628e-05,
+      "loss": 0.2403,
+      "step": 90120
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.560523690326197e-05,
+      "loss": 0.232,
+      "step": 90125
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5602597041013122e-05,
+      "loss": 0.2458,
+      "step": 90130
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5599957300783346e-05,
+      "loss": 0.2349,
+      "step": 90135
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.559731768260694e-05,
+      "loss": 0.2434,
+      "step": 90140
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5594678186518168e-05,
+      "loss": 0.2512,
+      "step": 90145
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5592038812551303e-05,
+      "loss": 0.2682,
+      "step": 90150
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5589399560740603e-05,
+      "loss": 0.2573,
+      "step": 90155
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5586760431120358e-05,
+      "loss": 0.2615,
+      "step": 90160
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.558412142372482e-05,
+      "loss": 0.2723,
+      "step": 90165
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5581482538588265e-05,
+      "loss": 0.2662,
+      "step": 90170
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5578843775744945e-05,
+      "loss": 0.2276,
+      "step": 90175
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5576205135229137e-05,
+      "loss": 0.2404,
+      "step": 90180
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5573566617075096e-05,
+      "loss": 0.2428,
+      "step": 90185
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5570928221317075e-05,
+      "loss": 0.2637,
+      "step": 90190
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.556828994798933e-05,
+      "loss": 0.276,
+      "step": 90195
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5565651797126137e-05,
+      "loss": 0.2541,
+      "step": 90200
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5563013768761735e-05,
+      "loss": 0.2626,
+      "step": 90205
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5560375862930376e-05,
+      "loss": 0.2614,
+      "step": 90210
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.555773807966632e-05,
+      "loss": 0.2798,
+      "step": 90215
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.555510041900381e-05,
+      "loss": 0.2506,
+      "step": 90220
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5552462880977098e-05,
+      "loss": 0.2327,
+      "step": 90225
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.554982546562042e-05,
+      "loss": 0.2532,
+      "step": 90230
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5547188172968048e-05,
+      "loss": 0.2477,
+      "step": 90235
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.55445510030542e-05,
+      "loss": 0.2468,
+      "step": 90240
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5541913955913124e-05,
+      "loss": 0.2551,
+      "step": 90245
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5539277031579058e-05,
+      "loss": 0.263,
+      "step": 90250
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.553664023008625e-05,
+      "loss": 0.2437,
+      "step": 90255
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5534003551468934e-05,
+      "loss": 0.2818,
+      "step": 90260
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5531366995761347e-05,
+      "loss": 0.2727,
+      "step": 90265
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5528730562997705e-05,
+      "loss": 0.2619,
+      "step": 90270
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5526094253212264e-05,
+      "loss": 0.2655,
+      "step": 90275
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5523458066439248e-05,
+      "loss": 0.2708,
+      "step": 90280
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5520822002712883e-05,
+      "loss": 0.2733,
+      "step": 90285
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.551818606206739e-05,
+      "loss": 0.2635,
+      "step": 90290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.551555024453701e-05,
+      "loss": 0.2432,
+      "step": 90295
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5512914550155956e-05,
+      "loss": 0.2506,
+      "step": 90300
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.551027897895846e-05,
+      "loss": 0.2604,
+      "step": 90305
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.550764353097873e-05,
+      "loss": 0.2649,
+      "step": 90310
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5505008206250995e-05,
+      "loss": 0.2424,
+      "step": 90315
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5502373004809475e-05,
+      "loss": 0.2583,
+      "step": 90320
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5499737926688374e-05,
+      "loss": 0.2623,
+      "step": 90325
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5497102971921924e-05,
+      "loss": 0.2689,
+      "step": 90330
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5494468140544323e-05,
+      "loss": 0.2514,
+      "step": 90335
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5491833432589796e-05,
+      "loss": 0.2469,
+      "step": 90340
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5489198848092533e-05,
+      "loss": 0.24,
+      "step": 90345
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5486564387086766e-05,
+      "loss": 0.2573,
+      "step": 90350
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5483930049606686e-05,
+      "loss": 0.2465,
+      "step": 90355
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.548129583568651e-05,
+      "loss": 0.2557,
+      "step": 90360
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5478661745360423e-05,
+      "loss": 0.258,
+      "step": 90365
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.547602777866265e-05,
+      "loss": 0.2579,
+      "step": 90370
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5473393935627377e-05,
+      "loss": 0.2479,
+      "step": 90375
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5470760216288804e-05,
+      "loss": 0.2549,
+      "step": 90380
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5468126620681125e-05,
+      "loss": 0.2813,
+      "step": 90385
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5465493148838546e-05,
+      "loss": 0.2543,
+      "step": 90390
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5462859800795253e-05,
+      "loss": 0.2616,
+      "step": 90395
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5460226576585447e-05,
+      "loss": 0.2329,
+      "step": 90400
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5457593476243295e-05,
+      "loss": 0.245,
+      "step": 90405
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5454960499803016e-05,
+      "loss": 0.2666,
+      "step": 90410
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5452327647298782e-05,
+      "loss": 0.2771,
+      "step": 90415
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5449694918764785e-05,
+      "loss": 0.2595,
+      "step": 90420
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5447062314235196e-05,
+      "loss": 0.2372,
+      "step": 90425
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5444429833744213e-05,
+      "loss": 0.2799,
+      "step": 90430
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5441797477326017e-05,
+      "loss": 0.2616,
+      "step": 90435
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.543916524501477e-05,
+      "loss": 0.2696,
+      "step": 90440
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5436533136844665e-05,
+      "loss": 0.2675,
+      "step": 90445
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.543390115284988e-05,
+      "loss": 0.264,
+      "step": 90450
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5431269293064585e-05,
+      "loss": 0.2748,
+      "step": 90455
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5428637557522942e-05,
+      "loss": 0.2724,
+      "step": 90460
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.542600594625914e-05,
+      "loss": 0.2642,
+      "step": 90465
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.542337445930734e-05,
+      "loss": 0.2555,
+      "step": 90470
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5420743096701717e-05,
+      "loss": 0.2335,
+      "step": 90475
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.541811185847642e-05,
+      "loss": 0.2708,
+      "step": 90480
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5415480744665635e-05,
+      "loss": 0.2621,
+      "step": 90485
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5412849755303514e-05,
+      "loss": 0.2507,
+      "step": 90490
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5410218890424222e-05,
+      "loss": 0.2686,
+      "step": 90495
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5407588150061907e-05,
+      "loss": 0.2645,
+      "step": 90500
+    },
+    {
+      "epoch": 3.18,
+      "eval_loss": 0.2550298273563385,
+      "eval_runtime": 10.5455,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 90500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5404957534250752e-05,
+      "loss": 0.2426,
+      "step": 90505
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.54023270430249e-05,
+      "loss": 0.2579,
+      "step": 90510
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5399696676418506e-05,
+      "loss": 0.2536,
+      "step": 90515
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5397066434465712e-05,
+      "loss": 0.245,
+      "step": 90520
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5394436317200687e-05,
+      "loss": 0.2526,
+      "step": 90525
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.539180632465758e-05,
+      "loss": 0.2504,
+      "step": 90530
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.538917645687054e-05,
+      "loss": 0.266,
+      "step": 90535
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5386546713873695e-05,
+      "loss": 0.2471,
+      "step": 90540
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5383917095701218e-05,
+      "loss": 0.2568,
+      "step": 90545
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5381287602387236e-05,
+      "loss": 0.2708,
+      "step": 90550
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5378658233965893e-05,
+      "loss": 0.2876,
+      "step": 90555
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5376028990471336e-05,
+      "loss": 0.2632,
+      "step": 90560
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5373399871937706e-05,
+      "loss": 0.2386,
+      "step": 90565
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.537077087839913e-05,
+      "loss": 0.2629,
+      "step": 90570
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5368142009889737e-05,
+      "loss": 0.2412,
+      "step": 90575
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5365513266443683e-05,
+      "loss": 0.2609,
+      "step": 90580
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.536288464809509e-05,
+      "loss": 0.2865,
+      "step": 90585
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5360256154878093e-05,
+      "loss": 0.2689,
+      "step": 90590
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.53576277868268e-05,
+      "loss": 0.2651,
+      "step": 90595
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5354999543975372e-05,
+      "loss": 0.251,
+      "step": 90600
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5352371426357913e-05,
+      "loss": 0.2549,
+      "step": 90605
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5349743434008556e-05,
+      "loss": 0.245,
+      "step": 90610
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.534711556696141e-05,
+      "loss": 0.2429,
+      "step": 90615
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5344487825250624e-05,
+      "loss": 0.2646,
+      "step": 90620
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5341860208910292e-05,
+      "loss": 0.2681,
+      "step": 90625
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.533923271797454e-05,
+      "loss": 0.2423,
+      "step": 90630
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5336605352477476e-05,
+      "loss": 0.2569,
+      "step": 90635
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5333978112453235e-05,
+      "loss": 0.2557,
+      "step": 90640
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5331350997935916e-05,
+      "loss": 0.2587,
+      "step": 90645
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5328724008959637e-05,
+      "loss": 0.2696,
+      "step": 90650
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.532609714555849e-05,
+      "loss": 0.2512,
+      "step": 90655
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5323470407766602e-05,
+      "loss": 0.2534,
+      "step": 90660
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.532084379561808e-05,
+      "loss": 0.2649,
+      "step": 90665
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.531821730914702e-05,
+      "loss": 0.26,
+      "step": 90670
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5315590948387516e-05,
+      "loss": 0.2505,
+      "step": 90675
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5312964713373695e-05,
+      "loss": 0.2635,
+      "step": 90680
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5310338604139638e-05,
+      "loss": 0.2791,
+      "step": 90685
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5307712620719444e-05,
+      "loss": 0.2511,
+      "step": 90690
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.530508676314722e-05,
+      "loss": 0.2455,
+      "step": 90695
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5302461031457057e-05,
+      "loss": 0.2518,
+      "step": 90700
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.529983542568305e-05,
+      "loss": 0.2622,
+      "step": 90705
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5297209945859274e-05,
+      "loss": 0.2675,
+      "step": 90710
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5294584592019835e-05,
+      "loss": 0.2519,
+      "step": 90715
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.529195936419883e-05,
+      "loss": 0.2756,
+      "step": 90720
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5289334262430332e-05,
+      "loss": 0.2521,
+      "step": 90725
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5286709286748417e-05,
+      "loss": 0.2765,
+      "step": 90730
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5284084437187194e-05,
+      "loss": 0.2716,
+      "step": 90735
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5281459713780726e-05,
+      "loss": 0.2615,
+      "step": 90740
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.52788351165631e-05,
+      "loss": 0.2552,
+      "step": 90745
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.527621064556839e-05,
+      "loss": 0.2574,
+      "step": 90750
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5273586300830684e-05,
+      "loss": 0.2539,
+      "step": 90755
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5270962082384047e-05,
+      "loss": 0.2696,
+      "step": 90760
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5268337990262555e-05,
+      "loss": 0.2712,
+      "step": 90765
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5265714024500276e-05,
+      "loss": 0.279,
+      "step": 90770
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5263090185131287e-05,
+      "loss": 0.2613,
+      "step": 90775
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.526046647218966e-05,
+      "loss": 0.2419,
+      "step": 90780
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.525784288570946e-05,
+      "loss": 0.2895,
+      "step": 90785
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5255219425724737e-05,
+      "loss": 0.2659,
+      "step": 90790
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5252596092269575e-05,
+      "loss": 0.2426,
+      "step": 90795
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5249972885378034e-05,
+      "loss": 0.2553,
+      "step": 90800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5247349805084155e-05,
+      "loss": 0.2577,
+      "step": 90805
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5244726851422026e-05,
+      "loss": 0.2744,
+      "step": 90810
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5242104024425691e-05,
+      "loss": 0.2499,
+      "step": 90815
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5239481324129195e-05,
+      "loss": 0.2641,
+      "step": 90820
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5236858750566604e-05,
+      "loss": 0.2535,
+      "step": 90825
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5234236303771967e-05,
+      "loss": 0.2481,
+      "step": 90830
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5231613983779341e-05,
+      "loss": 0.2672,
+      "step": 90835
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.522899179062277e-05,
+      "loss": 0.2875,
+      "step": 90840
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5226369724336292e-05,
+      "loss": 0.2434,
+      "step": 90845
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5223747784953974e-05,
+      "loss": 0.2503,
+      "step": 90850
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5221125972509839e-05,
+      "loss": 0.2584,
+      "step": 90855
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5218504287037947e-05,
+      "loss": 0.2519,
+      "step": 90860
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5215882728572317e-05,
+      "loss": 0.2567,
+      "step": 90865
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5213261297147014e-05,
+      "loss": 0.2547,
+      "step": 90870
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5210639992796063e-05,
+      "loss": 0.2447,
+      "step": 90875
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5208018815553496e-05,
+      "loss": 0.2541,
+      "step": 90880
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5205397765453348e-05,
+      "loss": 0.2599,
+      "step": 90885
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5202776842529662e-05,
+      "loss": 0.2661,
+      "step": 90890
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5200156046816463e-05,
+      "loss": 0.2734,
+      "step": 90895
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5197535378347775e-05,
+      "loss": 0.2833,
+      "step": 90900
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5194914837157631e-05,
+      "loss": 0.2635,
+      "step": 90905
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5192294423280052e-05,
+      "loss": 0.2675,
+      "step": 90910
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5189674136749077e-05,
+      "loss": 0.2532,
+      "step": 90915
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5187053977598702e-05,
+      "loss": 0.2469,
+      "step": 90920
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.518443394586298e-05,
+      "loss": 0.2738,
+      "step": 90925
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5181814041575908e-05,
+      "loss": 0.2736,
+      "step": 90930
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5179194264771512e-05,
+      "loss": 0.2674,
+      "step": 90935
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5176574615483796e-05,
+      "loss": 0.2715,
+      "step": 90940
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.51739550937468e-05,
+      "loss": 0.2848,
+      "step": 90945
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5171335699594519e-05,
+      "loss": 0.2449,
+      "step": 90950
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5168716433060964e-05,
+      "loss": 0.2451,
+      "step": 90955
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.516609729418014e-05,
+      "loss": 0.2459,
+      "step": 90960
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5163478282986065e-05,
+      "loss": 0.2462,
+      "step": 90965
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.516085939951275e-05,
+      "loss": 0.2721,
+      "step": 90970
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.515824064379419e-05,
+      "loss": 0.275,
+      "step": 90975
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.515562201586438e-05,
+      "loss": 0.2739,
+      "step": 90980
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5153003515757335e-05,
+      "loss": 0.2541,
+      "step": 90985
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5150385143507052e-05,
+      "loss": 0.2435,
+      "step": 90990
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5147766899147531e-05,
+      "loss": 0.2621,
+      "step": 90995
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5145148782712748e-05,
+      "loss": 0.2644,
+      "step": 91000
+    },
+    {
+      "epoch": 3.2,
+      "eval_loss": 0.25480425357818604,
+      "eval_runtime": 10.5461,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 91000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5142530794236731e-05,
+      "loss": 0.2754,
+      "step": 91005
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5139912933753451e-05,
+      "loss": 0.2911,
+      "step": 91010
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5137295201296897e-05,
+      "loss": 0.2636,
+      "step": 91015
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5134677596901065e-05,
+      "loss": 0.2511,
+      "step": 91020
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.513206012059995e-05,
+      "loss": 0.281,
+      "step": 91025
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5129442772427531e-05,
+      "loss": 0.2354,
+      "step": 91030
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5126825552417784e-05,
+      "loss": 0.2607,
+      "step": 91035
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5124208460604708e-05,
+      "loss": 0.2657,
+      "step": 91040
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5121591497022275e-05,
+      "loss": 0.2594,
+      "step": 91045
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5118974661704465e-05,
+      "loss": 0.2513,
+      "step": 91050
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5116357954685257e-05,
+      "loss": 0.2565,
+      "step": 91055
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.511374137599863e-05,
+      "loss": 0.2841,
+      "step": 91060
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5111124925678558e-05,
+      "loss": 0.2329,
+      "step": 91065
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5108508603759008e-05,
+      "loss": 0.2446,
+      "step": 91070
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5105892410273947e-05,
+      "loss": 0.2464,
+      "step": 91075
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5103276345257368e-05,
+      "loss": 0.2742,
+      "step": 91080
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5100660408743218e-05,
+      "loss": 0.2279,
+      "step": 91085
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.509804460076547e-05,
+      "loss": 0.252,
+      "step": 91090
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5095428921358079e-05,
+      "loss": 0.2564,
+      "step": 91095
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.509281337055502e-05,
+      "loss": 0.2621,
+      "step": 91100
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5090197948390255e-05,
+      "loss": 0.2564,
+      "step": 91105
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5087582654897738e-05,
+      "loss": 0.2645,
+      "step": 91110
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5084967490111419e-05,
+      "loss": 0.265,
+      "step": 91115
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5082352454065274e-05,
+      "loss": 0.2581,
+      "step": 91120
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5079737546793237e-05,
+      "loss": 0.2452,
+      "step": 91125
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5077122768329282e-05,
+      "loss": 0.2548,
+      "step": 91130
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5074508118707334e-05,
+      "loss": 0.2566,
+      "step": 91135
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5071893597961367e-05,
+      "loss": 0.2787,
+      "step": 91140
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.506927920612532e-05,
+      "loss": 0.2527,
+      "step": 91145
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5066664943233128e-05,
+      "loss": 0.2547,
+      "step": 91150
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.506405080931875e-05,
+      "loss": 0.236,
+      "step": 91155
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5061436804416129e-05,
+      "loss": 0.2815,
+      "step": 91160
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5058822928559202e-05,
+      "loss": 0.2583,
+      "step": 91165
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5056209181781896e-05,
+      "loss": 0.2591,
+      "step": 91170
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5053595564118172e-05,
+      "loss": 0.2473,
+      "step": 91175
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5050982075601949e-05,
+      "loss": 0.2496,
+      "step": 91180
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5048368716267175e-05,
+      "loss": 0.2478,
+      "step": 91185
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5045755486147761e-05,
+      "loss": 0.2934,
+      "step": 91190
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5043142385277664e-05,
+      "loss": 0.2542,
+      "step": 91195
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5040529413690802e-05,
+      "loss": 0.2574,
+      "step": 91200
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5037916571421094e-05,
+      "loss": 0.2496,
+      "step": 91205
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5035303858502474e-05,
+      "loss": 0.2459,
+      "step": 91210
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5032691274968868e-05,
+      "loss": 0.2913,
+      "step": 91215
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5030078820854205e-05,
+      "loss": 0.28,
+      "step": 91220
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5027466496192393e-05,
+      "loss": 0.2536,
+      "step": 91225
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5024854301017349e-05,
+      "loss": 0.2663,
+      "step": 91230
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5022242235363004e-05,
+      "loss": 0.2726,
+      "step": 91235
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.501963029926327e-05,
+      "loss": 0.2587,
+      "step": 91240
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.501701849275206e-05,
+      "loss": 0.2533,
+      "step": 91245
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.501440681586328e-05,
+      "loss": 0.2719,
+      "step": 91250
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5011795268630852e-05,
+      "loss": 0.2614,
+      "step": 91255
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5009183851088677e-05,
+      "loss": 0.2336,
+      "step": 91260
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5006572563270665e-05,
+      "loss": 0.26,
+      "step": 91265
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5003961405210723e-05,
+      "loss": 0.2684,
+      "step": 91270
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5001350376942758e-05,
+      "loss": 0.2504,
+      "step": 91275
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.499873947850067e-05,
+      "loss": 0.2491,
+      "step": 91280
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.499612870991835e-05,
+      "loss": 0.2461,
+      "step": 91285
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4993518071229712e-05,
+      "loss": 0.2483,
+      "step": 91290
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4990907562468647e-05,
+      "loss": 0.2515,
+      "step": 91295
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4988297183669054e-05,
+      "loss": 0.2458,
+      "step": 91300
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4985686934864812e-05,
+      "loss": 0.2697,
+      "step": 91305
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.498307681608984e-05,
+      "loss": 0.2606,
+      "step": 91310
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4980466827378009e-05,
+      "loss": 0.2456,
+      "step": 91315
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4977856968763212e-05,
+      "loss": 0.2617,
+      "step": 91320
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4975247240279332e-05,
+      "loss": 0.2436,
+      "step": 91325
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.497263764196027e-05,
+      "loss": 0.2471,
+      "step": 91330
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4970028173839901e-05,
+      "loss": 0.2406,
+      "step": 91335
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4967418835952095e-05,
+      "loss": 0.2429,
+      "step": 91340
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.496480962833075e-05,
+      "loss": 0.2761,
+      "step": 91345
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4962200551009735e-05,
+      "loss": 0.2409,
+      "step": 91350
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4959591604022937e-05,
+      "loss": 0.2541,
+      "step": 91355
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4956982787404228e-05,
+      "loss": 0.2431,
+      "step": 91360
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4954374101187465e-05,
+      "loss": 0.241,
+      "step": 91365
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4951765545406547e-05,
+      "loss": 0.2682,
+      "step": 91370
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.4949157120095324e-05,
+      "loss": 0.2601,
+      "step": 91375
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4946548825287668e-05,
+      "loss": 0.2532,
+      "step": 91380
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4943940661017458e-05,
+      "loss": 0.2727,
+      "step": 91385
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4941332627318555e-05,
+      "loss": 0.2476,
+      "step": 91390
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4938724724224817e-05,
+      "loss": 0.256,
+      "step": 91395
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4936116951770095e-05,
+      "loss": 0.2726,
+      "step": 91400
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4933509309988271e-05,
+      "loss": 0.2599,
+      "step": 91405
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.49309017989132e-05,
+      "loss": 0.2461,
+      "step": 91410
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4928294418578732e-05,
+      "loss": 0.2535,
+      "step": 91415
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4925687169018712e-05,
+      "loss": 0.2698,
+      "step": 91420
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.492308005026702e-05,
+      "loss": 0.2688,
+      "step": 91425
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4920473062357481e-05,
+      "loss": 0.2572,
+      "step": 91430
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4917866205323966e-05,
+      "loss": 0.233,
+      "step": 91435
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.49152594792003e-05,
+      "loss": 0.2626,
+      "step": 91440
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4912652884020356e-05,
+      "loss": 0.2667,
+      "step": 91445
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4910046419817966e-05,
+      "loss": 0.2822,
+      "step": 91450
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.490744008662697e-05,
+      "loss": 0.2621,
+      "step": 91455
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4904833884481206e-05,
+      "loss": 0.2532,
+      "step": 91460
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4902227813414531e-05,
+      "loss": 0.2704,
+      "step": 91465
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4899621873460778e-05,
+      "loss": 0.2505,
+      "step": 91470
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4897016064653774e-05,
+      "loss": 0.2427,
+      "step": 91475
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.489441038702735e-05,
+      "loss": 0.2428,
+      "step": 91480
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4891804840615353e-05,
+      "loss": 0.2644,
+      "step": 91485
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.488919942545161e-05,
+      "loss": 0.2517,
+      "step": 91490
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4886594141569944e-05,
+      "loss": 0.2629,
+      "step": 91495
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4883988989004194e-05,
+      "loss": 0.2442,
+      "step": 91500
+    },
+    {
+      "epoch": 3.22,
+      "eval_loss": 0.25463610887527466,
+      "eval_runtime": 10.5511,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 91500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4881383967788181e-05,
+      "loss": 0.2399,
+      "step": 91505
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4878779077955723e-05,
+      "loss": 0.2549,
+      "step": 91510
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4876174319540647e-05,
+      "loss": 0.2428,
+      "step": 91515
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4873569692576784e-05,
+      "loss": 0.2622,
+      "step": 91520
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4870965197097944e-05,
+      "loss": 0.2583,
+      "step": 91525
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4868360833137945e-05,
+      "loss": 0.2568,
+      "step": 91530
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4865756600730597e-05,
+      "loss": 0.2574,
+      "step": 91535
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4863152499909722e-05,
+      "loss": 0.2483,
+      "step": 91540
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4861069314018015e-05,
+      "loss": 0.2347,
+      "step": 91545
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4858465450137993e-05,
+      "loss": 0.2609,
+      "step": 91550
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.485586171793912e-05,
+      "loss": 0.2563,
+      "step": 91555
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4853258117455202e-05,
+      "loss": 0.2392,
+      "step": 91560
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4850654648720031e-05,
+      "loss": 0.2914,
+      "step": 91565
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4848051311767424e-05,
+      "loss": 0.2669,
+      "step": 91570
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4845448106631182e-05,
+      "loss": 0.2672,
+      "step": 91575
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4842845033345106e-05,
+      "loss": 0.2848,
+      "step": 91580
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4840242091942986e-05,
+      "loss": 0.2635,
+      "step": 91585
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4837639282458635e-05,
+      "loss": 0.2705,
+      "step": 91590
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.483503660492584e-05,
+      "loss": 0.2306,
+      "step": 91595
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4832434059378392e-05,
+      "loss": 0.2859,
+      "step": 91600
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4829831645850084e-05,
+      "loss": 0.2609,
+      "step": 91605
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4827229364374718e-05,
+      "loss": 0.2488,
+      "step": 91610
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4824627214986072e-05,
+      "loss": 0.2563,
+      "step": 91615
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4822025197717937e-05,
+      "loss": 0.2649,
+      "step": 91620
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4819423312604091e-05,
+      "loss": 0.2611,
+      "step": 91625
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4816821559678324e-05,
+      "loss": 0.2774,
+      "step": 91630
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4814219938974428e-05,
+      "loss": 0.258,
+      "step": 91635
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.481161845052617e-05,
+      "loss": 0.2518,
+      "step": 91640
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4809017094367322e-05,
+      "loss": 0.2379,
+      "step": 91645
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4806415870531681e-05,
+      "loss": 0.2669,
+      "step": 91650
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.480381477905301e-05,
+      "loss": 0.2635,
+      "step": 91655
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.4801213819965087e-05,
+      "loss": 0.2717,
+      "step": 91660
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.479861299330167e-05,
+      "loss": 0.2847,
+      "step": 91665
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4796012299096551e-05,
+      "loss": 0.2471,
+      "step": 91670
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4793411737383488e-05,
+      "loss": 0.2323,
+      "step": 91675
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.479081130819624e-05,
+      "loss": 0.2785,
+      "step": 91680
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4788211011568581e-05,
+      "loss": 0.2688,
+      "step": 91685
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.478561084753428e-05,
+      "loss": 0.2751,
+      "step": 91690
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4783010816127088e-05,
+      "loss": 0.2484,
+      "step": 91695
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4780410917380757e-05,
+      "loss": 0.2553,
+      "step": 91700
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4777811151329068e-05,
+      "loss": 0.2591,
+      "step": 91705
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4775211518005756e-05,
+      "loss": 0.2395,
+      "step": 91710
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4772612017444592e-05,
+      "loss": 0.2454,
+      "step": 91715
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.477001264967931e-05,
+      "loss": 0.2699,
+      "step": 91720
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.476741341474368e-05,
+      "loss": 0.269,
+      "step": 91725
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4764814312671446e-05,
+      "loss": 0.2733,
+      "step": 91730
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4762215343496346e-05,
+      "loss": 0.2514,
+      "step": 91735
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4759616507252127e-05,
+      "loss": 0.2639,
+      "step": 91740
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4757017803972551e-05,
+      "loss": 0.2544,
+      "step": 91745
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.475441923369135e-05,
+      "loss": 0.2346,
+      "step": 91750
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4751820796442261e-05,
+      "loss": 0.2813,
+      "step": 91755
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4749222492259013e-05,
+      "loss": 0.2688,
+      "step": 91760
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4746624321175365e-05,
+      "loss": 0.2435,
+      "step": 91765
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4744026283225045e-05,
+      "loss": 0.265,
+      "step": 91770
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4741428378441786e-05,
+      "loss": 0.2716,
+      "step": 91775
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4738830606859309e-05,
+      "loss": 0.2371,
+      "step": 91780
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4736232968511362e-05,
+      "loss": 0.2622,
+      "step": 91785
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4733635463431661e-05,
+      "loss": 0.2683,
+      "step": 91790
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4731038091653943e-05,
+      "loss": 0.249,
+      "step": 91795
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4728440853211923e-05,
+      "loss": 0.2776,
+      "step": 91800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4725843748139335e-05,
+      "loss": 0.2587,
+      "step": 91805
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4723246776469896e-05,
+      "loss": 0.2436,
+      "step": 91810
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4720649938237313e-05,
+      "loss": 0.267,
+      "step": 91815
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.471805323347532e-05,
+      "loss": 0.2601,
+      "step": 91820
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4715456662217639e-05,
+      "loss": 0.2803,
+      "step": 91825
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4712860224497973e-05,
+      "loss": 0.2677,
+      "step": 91830
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4710263920350028e-05,
+      "loss": 0.2623,
+      "step": 91835
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4707667749807535e-05,
+      "loss": 0.2359,
+      "step": 91840
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4705071712904189e-05,
+      "loss": 0.2593,
+      "step": 91845
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4702475809673707e-05,
+      "loss": 0.2581,
+      "step": 91850
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.469988004014978e-05,
+      "loss": 0.2673,
+      "step": 91855
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4697284404366133e-05,
+      "loss": 0.2592,
+      "step": 91860
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4694688902356458e-05,
+      "loss": 0.2567,
+      "step": 91865
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.469209353415445e-05,
+      "loss": 0.2353,
+      "step": 91870
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4689498299793813e-05,
+      "loss": 0.258,
+      "step": 91875
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.468690319930825e-05,
+      "loss": 0.2777,
+      "step": 91880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4684308232731457e-05,
+      "loss": 0.2432,
+      "step": 91885
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.468171340009712e-05,
+      "loss": 0.2557,
+      "step": 91890
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4679118701438926e-05,
+      "loss": 0.2605,
+      "step": 91895
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4676524136790584e-05,
+      "loss": 0.251,
+      "step": 91900
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4673929706185769e-05,
+      "loss": 0.2479,
+      "step": 91905
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4671335409658175e-05,
+      "loss": 0.2499,
+      "step": 91910
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4668741247241475e-05,
+      "loss": 0.2418,
+      "step": 91915
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4666147218969367e-05,
+      "loss": 0.2717,
+      "step": 91920
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4663553324875529e-05,
+      "loss": 0.2779,
+      "step": 91925
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.466095956499363e-05,
+      "loss": 0.2812,
+      "step": 91930
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4658365939357366e-05,
+      "loss": 0.2558,
+      "step": 91935
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4655772448000408e-05,
+      "loss": 0.2468,
+      "step": 91940
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.4653179090956425e-05,
+      "loss": 0.2882,
+      "step": 91945
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4650585868259087e-05,
+      "loss": 0.2731,
+      "step": 91950
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4647992779942077e-05,
+      "loss": 0.2524,
+      "step": 91955
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4645399826039052e-05,
+      "loss": 0.264,
+      "step": 91960
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4642807006583698e-05,
+      "loss": 0.2514,
+      "step": 91965
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4640214321609653e-05,
+      "loss": 0.2804,
+      "step": 91970
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.463762177115061e-05,
+      "loss": 0.2562,
+      "step": 91975
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4635029355240221e-05,
+      "loss": 0.2612,
+      "step": 91980
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4632437073912136e-05,
+      "loss": 0.2569,
+      "step": 91985
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.462984492720002e-05,
+      "loss": 0.2519,
+      "step": 91990
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4627252915137546e-05,
+      "loss": 0.2588,
+      "step": 91995
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4624661037758356e-05,
+      "loss": 0.2677,
+      "step": 92000
+    },
+    {
+      "epoch": 3.24,
+      "eval_loss": 0.2548165023326874,
+      "eval_runtime": 10.5617,
+      "eval_samples_per_second": 9.468,
+      "eval_steps_per_second": 9.468,
+      "step": 92000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4622069295096096e-05,
+      "loss": 0.2453,
+      "step": 92005
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.461947768718443e-05,
+      "loss": 0.2439,
+      "step": 92010
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4616886214057007e-05,
+      "loss": 0.2609,
+      "step": 92015
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4614294875747475e-05,
+      "loss": 0.2477,
+      "step": 92020
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4611703672289484e-05,
+      "loss": 0.2486,
+      "step": 92025
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.460911260371666e-05,
+      "loss": 0.2642,
+      "step": 92030
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4606521670062673e-05,
+      "loss": 0.2674,
+      "step": 92035
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.460393087136115e-05,
+      "loss": 0.2648,
+      "step": 92040
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4601340207645726e-05,
+      "loss": 0.2583,
+      "step": 92045
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4598749678950052e-05,
+      "loss": 0.2811,
+      "step": 92050
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4596159285307758e-05,
+      "loss": 0.2561,
+      "step": 92055
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4593569026752484e-05,
+      "loss": 0.2414,
+      "step": 92060
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.459097890331785e-05,
+      "loss": 0.2629,
+      "step": 92065
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4588388915037504e-05,
+      "loss": 0.2519,
+      "step": 92070
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4585799061945065e-05,
+      "loss": 0.2487,
+      "step": 92075
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4583209344074164e-05,
+      "loss": 0.2793,
+      "step": 92080
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4580619761458414e-05,
+      "loss": 0.2651,
+      "step": 92085
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4578030314131457e-05,
+      "loss": 0.2645,
+      "step": 92090
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4575441002126913e-05,
+      "loss": 0.2459,
+      "step": 92095
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4572851825478395e-05,
+      "loss": 0.2574,
+      "step": 92100
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4570262784219513e-05,
+      "loss": 0.2627,
+      "step": 92105
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4567673878383897e-05,
+      "loss": 0.2483,
+      "step": 92110
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4565085108005172e-05,
+      "loss": 0.2788,
+      "step": 92115
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4562496473116938e-05,
+      "loss": 0.2663,
+      "step": 92120
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4559907973752798e-05,
+      "loss": 0.2607,
+      "step": 92125
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4557319609946385e-05,
+      "loss": 0.254,
+      "step": 92130
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4554731381731291e-05,
+      "loss": 0.2415,
+      "step": 92135
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4552143289141129e-05,
+      "loss": 0.2497,
+      "step": 92140
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.454955533220949e-05,
+      "loss": 0.2357,
+      "step": 92145
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.454696751097e-05,
+      "loss": 0.2616,
+      "step": 92150
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4544379825456247e-05,
+      "loss": 0.2607,
+      "step": 92155
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4541792275701821e-05,
+      "loss": 0.2507,
+      "step": 92160
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.453920486174033e-05,
+      "loss": 0.2505,
+      "step": 92165
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4536617583605381e-05,
+      "loss": 0.2799,
+      "step": 92170
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4534030441330556e-05,
+      "loss": 0.2376,
+      "step": 92175
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4531443434949439e-05,
+      "loss": 0.2609,
+      "step": 92180
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4528856564495642e-05,
+      "loss": 0.2734,
+      "step": 92185
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4526269830002737e-05,
+      "loss": 0.2529,
+      "step": 92190
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4523683231504318e-05,
+      "loss": 0.263,
+      "step": 92195
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4521096769033959e-05,
+      "loss": 0.2631,
+      "step": 92200
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.451851044262526e-05,
+      "loss": 0.2638,
+      "step": 92205
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4515924252311796e-05,
+      "loss": 0.2381,
+      "step": 92210
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4513338198127142e-05,
+      "loss": 0.2614,
+      "step": 92215
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4510752280104872e-05,
+      "loss": 0.2711,
+      "step": 92220
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4508166498278586e-05,
+      "loss": 0.2606,
+      "step": 92225
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.4505580852681843e-05,
+      "loss": 0.2653,
+      "step": 92230
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4502995343348218e-05,
+      "loss": 0.2511,
+      "step": 92235
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4500409970311268e-05,
+      "loss": 0.2755,
+      "step": 92240
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4497824733604588e-05,
+      "loss": 0.2681,
+      "step": 92245
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4495239633261728e-05,
+      "loss": 0.2792,
+      "step": 92250
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4492654669316263e-05,
+      "loss": 0.2421,
+      "step": 92255
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.449006984180174e-05,
+      "loss": 0.2473,
+      "step": 92260
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4487485150751745e-05,
+      "loss": 0.2751,
+      "step": 92265
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4484900596199816e-05,
+      "loss": 0.2476,
+      "step": 92270
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4482316178179534e-05,
+      "loss": 0.261,
+      "step": 92275
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4479731896724436e-05,
+      "loss": 0.2639,
+      "step": 92280
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4477147751868092e-05,
+      "loss": 0.2642,
+      "step": 92285
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4474563743644054e-05,
+      "loss": 0.2661,
+      "step": 92290
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4471979872085856e-05,
+      "loss": 0.2732,
+      "step": 92295
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4469396137227076e-05,
+      "loss": 0.2934,
+      "step": 92300
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4466812539101242e-05,
+      "loss": 0.2657,
+      "step": 92305
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.446422907774191e-05,
+      "loss": 0.2484,
+      "step": 92310
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4461645753182607e-05,
+      "loss": 0.2537,
+      "step": 92315
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.44590625654569e-05,
+      "loss": 0.2774,
+      "step": 92320
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.445647951459831e-05,
+      "loss": 0.2674,
+      "step": 92325
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.445389660064039e-05,
+      "loss": 0.2582,
+      "step": 92330
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.445131382361667e-05,
+      "loss": 0.2502,
+      "step": 92335
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4448731183560694e-05,
+      "loss": 0.2593,
+      "step": 92340
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4446148680505989e-05,
+      "loss": 0.2414,
+      "step": 92345
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.444356631448609e-05,
+      "loss": 0.2283,
+      "step": 92350
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4440984085534514e-05,
+      "loss": 0.272,
+      "step": 92355
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4438401993684811e-05,
+      "loss": 0.2576,
+      "step": 92360
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4435820038970498e-05,
+      "loss": 0.2794,
+      "step": 92365
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.44332382214251e-05,
+      "loss": 0.2481,
+      "step": 92370
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4430656541082133e-05,
+      "loss": 0.2631,
+      "step": 92375
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4428074997975132e-05,
+      "loss": 0.2682,
+      "step": 92380
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4425493592137601e-05,
+      "loss": 0.2555,
+      "step": 92385
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4422912323603066e-05,
+      "loss": 0.2567,
+      "step": 92390
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4420331192405053e-05,
+      "loss": 0.2441,
+      "step": 92395
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4417750198577069e-05,
+      "loss": 0.2532,
+      "step": 92400
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4415169342152624e-05,
+      "loss": 0.2683,
+      "step": 92405
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4412588623165218e-05,
+      "loss": 0.2695,
+      "step": 92410
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4410008041648388e-05,
+      "loss": 0.257,
+      "step": 92415
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4407427597635616e-05,
+      "loss": 0.2384,
+      "step": 92420
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4404847291160423e-05,
+      "loss": 0.2613,
+      "step": 92425
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4402267122256291e-05,
+      "loss": 0.2498,
+      "step": 92430
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4399687090956746e-05,
+      "loss": 0.2649,
+      "step": 92435
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4397107197295268e-05,
+      "loss": 0.2418,
+      "step": 92440
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4394527441305378e-05,
+      "loss": 0.2674,
+      "step": 92445
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.439194782302055e-05,
+      "loss": 0.2477,
+      "step": 92450
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4389368342474296e-05,
+      "loss": 0.2359,
+      "step": 92455
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4386788999700102e-05,
+      "loss": 0.2487,
+      "step": 92460
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4384209794731463e-05,
+      "loss": 0.2506,
+      "step": 92465
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4381630727601847e-05,
+      "loss": 0.255,
+      "step": 92470
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4379051798344772e-05,
+      "loss": 0.2906,
+      "step": 92475
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.437647300699371e-05,
+      "loss": 0.2423,
+      "step": 92480
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4373894353582135e-05,
+      "loss": 0.2642,
+      "step": 92485
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4371315838143545e-05,
+      "loss": 0.2657,
+      "step": 92490
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4368737460711406e-05,
+      "loss": 0.2582,
+      "step": 92495
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.436615922131922e-05,
+      "loss": 0.2564,
+      "step": 92500
+    },
+    {
+      "epoch": 3.25,
+      "eval_loss": 0.25434496998786926,
+      "eval_runtime": 10.5432,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 92500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4363581120000432e-05,
+      "loss": 0.2536,
+      "step": 92505
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.436100315678855e-05,
+      "loss": 0.2728,
+      "step": 92510
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.4358425331717023e-05,
+      "loss": 0.2573,
+      "step": 92515
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4355847644819336e-05,
+      "loss": 0.2449,
+      "step": 92520
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.435327009612894e-05,
+      "loss": 0.245,
+      "step": 92525
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4350692685679323e-05,
+      "loss": 0.2713,
+      "step": 92530
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4348115413503948e-05,
+      "loss": 0.2275,
+      "step": 92535
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4345538279636273e-05,
+      "loss": 0.2387,
+      "step": 92540
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4342961284109751e-05,
+      "loss": 0.2323,
+      "step": 92545
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4340384426957854e-05,
+      "loss": 0.254,
+      "step": 92550
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4337807708214052e-05,
+      "loss": 0.2489,
+      "step": 92555
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4335231127911786e-05,
+      "loss": 0.2568,
+      "step": 92560
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4332654686084507e-05,
+      "loss": 0.2389,
+      "step": 92565
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4330078382765688e-05,
+      "loss": 0.2694,
+      "step": 92570
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4327502217988769e-05,
+      "loss": 0.2535,
+      "step": 92575
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4324926191787202e-05,
+      "loss": 0.236,
+      "step": 92580
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.432235030419442e-05,
+      "loss": 0.2577,
+      "step": 92585
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4319774555243895e-05,
+      "loss": 0.2554,
+      "step": 92590
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4317198944969057e-05,
+      "loss": 0.2546,
+      "step": 92595
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4314623473403343e-05,
+      "loss": 0.2744,
+      "step": 92600
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.43120481405802e-05,
+      "loss": 0.2617,
+      "step": 92605
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4309472946533075e-05,
+      "loss": 0.2554,
+      "step": 92610
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4306897891295403e-05,
+      "loss": 0.2573,
+      "step": 92615
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4304322974900605e-05,
+      "loss": 0.2565,
+      "step": 92620
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4301748197382131e-05,
+      "loss": 0.2631,
+      "step": 92625
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.429917355877341e-05,
+      "loss": 0.269,
+      "step": 92630
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4296599059107865e-05,
+      "loss": 0.2527,
+      "step": 92635
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4294024698418917e-05,
+      "loss": 0.2742,
+      "step": 92640
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4291450476740018e-05,
+      "loss": 0.2426,
+      "step": 92645
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4288876394104572e-05,
+      "loss": 0.2568,
+      "step": 92650
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4286302450546003e-05,
+      "loss": 0.2631,
+      "step": 92655
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4283728646097727e-05,
+      "loss": 0.2552,
+      "step": 92660
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4281154980793187e-05,
+      "loss": 0.2428,
+      "step": 92665
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4278581454665787e-05,
+      "loss": 0.2668,
+      "step": 92670
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4276008067748941e-05,
+      "loss": 0.2527,
+      "step": 92675
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4273434820076053e-05,
+      "loss": 0.2457,
+      "step": 92680
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4270861711680553e-05,
+      "loss": 0.252,
+      "step": 92685
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4268288742595842e-05,
+      "loss": 0.2754,
+      "step": 92690
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4265715912855331e-05,
+      "loss": 0.2431,
+      "step": 92695
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4263143222492414e-05,
+      "loss": 0.2656,
+      "step": 92700
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.426057067154052e-05,
+      "loss": 0.2387,
+      "step": 92705
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4257998260033024e-05,
+      "loss": 0.2546,
+      "step": 92710
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.425542598800335e-05,
+      "loss": 0.2652,
+      "step": 92715
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4252853855484883e-05,
+      "loss": 0.2448,
+      "step": 92720
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4250281862511034e-05,
+      "loss": 0.2666,
+      "step": 92725
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4247710009115192e-05,
+      "loss": 0.2428,
+      "step": 92730
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.424513829533074e-05,
+      "loss": 0.2671,
+      "step": 92735
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4242566721191087e-05,
+      "loss": 0.2536,
+      "step": 92740
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4239995286729615e-05,
+      "loss": 0.2517,
+      "step": 92745
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4237423991979715e-05,
+      "loss": 0.2407,
+      "step": 92750
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.423485283697476e-05,
+      "loss": 0.244,
+      "step": 92755
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.423228182174816e-05,
+      "loss": 0.2608,
+      "step": 92760
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4229710946333272e-05,
+      "loss": 0.2484,
+      "step": 92765
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.42271402107635e-05,
+      "loss": 0.2848,
+      "step": 92770
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4224569615072203e-05,
+      "loss": 0.2537,
+      "step": 92775
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4221999159292776e-05,
+      "loss": 0.2552,
+      "step": 92780
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4219428843458588e-05,
+      "loss": 0.2513,
+      "step": 92785
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4216858667603015e-05,
+      "loss": 0.2473,
+      "step": 92790
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4214288631759413e-05,
+      "loss": 0.2535,
+      "step": 92795
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.4211718735961174e-05,
+      "loss": 0.2803,
+      "step": 92800
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.420914898024166e-05,
+      "loss": 0.2466,
+      "step": 92805
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4206579364634231e-05,
+      "loss": 0.2617,
+      "step": 92810
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4204009889172248e-05,
+      "loss": 0.2561,
+      "step": 92815
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4201440553889095e-05,
+      "loss": 0.2463,
+      "step": 92820
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4198871358818106e-05,
+      "loss": 0.2531,
+      "step": 92825
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4196302303992668e-05,
+      "loss": 0.2655,
+      "step": 92830
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4193733389446113e-05,
+      "loss": 0.2617,
+      "step": 92835
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4191164615211816e-05,
+      "loss": 0.2699,
+      "step": 92840
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4188595981323124e-05,
+      "loss": 0.2499,
+      "step": 92845
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.418602748781338e-05,
+      "loss": 0.2351,
+      "step": 92850
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4183459134715952e-05,
+      "loss": 0.2457,
+      "step": 92855
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4180890922064177e-05,
+      "loss": 0.2566,
+      "step": 92860
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4178322849891404e-05,
+      "loss": 0.2558,
+      "step": 92865
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4175754918230966e-05,
+      "loss": 0.256,
+      "step": 92870
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.417318712711623e-05,
+      "loss": 0.2528,
+      "step": 92875
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4170619476580513e-05,
+      "loss": 0.2599,
+      "step": 92880
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4168051966657176e-05,
+      "loss": 0.2291,
+      "step": 92885
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4165484597379534e-05,
+      "loss": 0.2616,
+      "step": 92890
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4162917368780943e-05,
+      "loss": 0.2558,
+      "step": 92895
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.416035028089473e-05,
+      "loss": 0.2478,
+      "step": 92900
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4157783333754226e-05,
+      "loss": 0.2743,
+      "step": 92905
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4155216527392747e-05,
+      "loss": 0.2864,
+      "step": 92910
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4152649861843648e-05,
+      "loss": 0.2724,
+      "step": 92915
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.415008333714024e-05,
+      "loss": 0.2614,
+      "step": 92920
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4147516953315836e-05,
+      "loss": 0.2682,
+      "step": 92925
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4144950710403788e-05,
+      "loss": 0.2617,
+      "step": 92930
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.414238460843739e-05,
+      "loss": 0.2711,
+      "step": 92935
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4139818647449981e-05,
+      "loss": 0.2508,
+      "step": 92940
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.413725282747487e-05,
+      "loss": 0.2638,
+      "step": 92945
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4134687148545362e-05,
+      "loss": 0.2476,
+      "step": 92950
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4132121610694792e-05,
+      "loss": 0.2468,
+      "step": 92955
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.412955621395646e-05,
+      "loss": 0.2461,
+      "step": 92960
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4126990958363668e-05,
+      "loss": 0.2656,
+      "step": 92965
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.412442584394974e-05,
+      "loss": 0.2755,
+      "step": 92970
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4121860870747978e-05,
+      "loss": 0.2617,
+      "step": 92975
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4119296038791683e-05,
+      "loss": 0.2502,
+      "step": 92980
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4116731348114148e-05,
+      "loss": 0.2519,
+      "step": 92985
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4114166798748684e-05,
+      "loss": 0.2423,
+      "step": 92990
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4111602390728602e-05,
+      "loss": 0.2438,
+      "step": 92995
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4109038124087181e-05,
+      "loss": 0.2675,
+      "step": 93000
+    },
+    {
+      "epoch": 3.27,
+      "eval_loss": 0.2542540431022644,
+      "eval_runtime": 10.5543,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 9.475,
+      "step": 93000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4106473998857717e-05,
+      "loss": 0.2569,
+      "step": 93005
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4103910015073518e-05,
+      "loss": 0.2748,
+      "step": 93010
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4101346172767871e-05,
+      "loss": 0.2652,
+      "step": 93015
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4098782471974057e-05,
+      "loss": 0.2259,
+      "step": 93020
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4096218912725356e-05,
+      "loss": 0.2508,
+      "step": 93025
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4093655495055081e-05,
+      "loss": 0.2728,
+      "step": 93030
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4091092218996502e-05,
+      "loss": 0.2435,
+      "step": 93035
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.408852908458289e-05,
+      "loss": 0.2542,
+      "step": 93040
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4085966091847535e-05,
+      "loss": 0.2551,
+      "step": 93045
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4083403240823732e-05,
+      "loss": 0.2636,
+      "step": 93050
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.408084053154474e-05,
+      "loss": 0.2521,
+      "step": 93055
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.407827796404384e-05,
+      "loss": 0.2496,
+      "step": 93060
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.407571553835429e-05,
+      "loss": 0.2704,
+      "step": 93065
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4073153254509386e-05,
+      "loss": 0.2535,
+      "step": 93070
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4070591112542387e-05,
+      "loss": 0.2552,
+      "step": 93075
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4068029112486547e-05,
+      "loss": 0.2795,
+      "step": 93080
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.4065467254375159e-05,
+      "loss": 0.2735,
+      "step": 93085
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.406290553824147e-05,
+      "loss": 0.2424,
+      "step": 93090
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4060343964118733e-05,
+      "loss": 0.273,
+      "step": 93095
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.405778253204022e-05,
+      "loss": 0.2706,
+      "step": 93100
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4055221242039202e-05,
+      "loss": 0.2269,
+      "step": 93105
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4052660094148922e-05,
+      "loss": 0.248,
+      "step": 93110
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.405009908840264e-05,
+      "loss": 0.254,
+      "step": 93115
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.404753822483359e-05,
+      "loss": 0.2408,
+      "step": 93120
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4044977503475051e-05,
+      "loss": 0.266,
+      "step": 93125
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4042416924360257e-05,
+      "loss": 0.2522,
+      "step": 93130
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4039856487522463e-05,
+      "loss": 0.2745,
+      "step": 93135
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4037296192994893e-05,
+      "loss": 0.2434,
+      "step": 93140
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.403473604081082e-05,
+      "loss": 0.2449,
+      "step": 93145
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4032176031003463e-05,
+      "loss": 0.2645,
+      "step": 93150
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4029616163606085e-05,
+      "loss": 0.2864,
+      "step": 93155
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4027056438651898e-05,
+      "loss": 0.2473,
+      "step": 93160
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4024496856174162e-05,
+      "loss": 0.2521,
+      "step": 93165
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4021937416206101e-05,
+      "loss": 0.2574,
+      "step": 93170
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4019378118780951e-05,
+      "loss": 0.2324,
+      "step": 93175
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4016818963931927e-05,
+      "loss": 0.2426,
+      "step": 93180
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4014259951692279e-05,
+      "loss": 0.244,
+      "step": 93185
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.401170108209523e-05,
+      "loss": 0.2634,
+      "step": 93190
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4009142355173996e-05,
+      "loss": 0.2655,
+      "step": 93195
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4006583770961798e-05,
+      "loss": 0.249,
+      "step": 93200
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4004025329491866e-05,
+      "loss": 0.2634,
+      "step": 93205
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.4001467030797428e-05,
+      "loss": 0.2556,
+      "step": 93210
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3998908874911679e-05,
+      "loss": 0.2586,
+      "step": 93215
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.399635086186786e-05,
+      "loss": 0.2485,
+      "step": 93220
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3993792991699173e-05,
+      "loss": 0.2416,
+      "step": 93225
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3991235264438833e-05,
+      "loss": 0.2645,
+      "step": 93230
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3988677680120036e-05,
+      "loss": 0.2557,
+      "step": 93235
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3986120238776012e-05,
+      "loss": 0.2617,
+      "step": 93240
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.398356294043996e-05,
+      "loss": 0.2646,
+      "step": 93245
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3981005785145085e-05,
+      "loss": 0.2721,
+      "step": 93250
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3978448772924574e-05,
+      "loss": 0.2445,
+      "step": 93255
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3975891903811653e-05,
+      "loss": 0.2669,
+      "step": 93260
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3973335177839503e-05,
+      "loss": 0.256,
+      "step": 93265
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3970778595041333e-05,
+      "loss": 0.2373,
+      "step": 93270
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.396822215545033e-05,
+      "loss": 0.2583,
+      "step": 93275
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3965665859099702e-05,
+      "loss": 0.2802,
+      "step": 93280
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.396310970602263e-05,
+      "loss": 0.2618,
+      "step": 93285
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3960553696252304e-05,
+      "loss": 0.2519,
+      "step": 93290
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3957997829821903e-05,
+      "loss": 0.2962,
+      "step": 93295
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3955442106764632e-05,
+      "loss": 0.2457,
+      "step": 93300
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.395288652711367e-05,
+      "loss": 0.2763,
+      "step": 93305
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3950331090902186e-05,
+      "loss": 0.2768,
+      "step": 93310
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3947775798163377e-05,
+      "loss": 0.291,
+      "step": 93315
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3945220648930408e-05,
+      "loss": 0.2642,
+      "step": 93320
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3942665643236475e-05,
+      "loss": 0.2497,
+      "step": 93325
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3940110781114729e-05,
+      "loss": 0.2789,
+      "step": 93330
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3937556062598367e-05,
+      "loss": 0.2427,
+      "step": 93335
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3935001487720547e-05,
+      "loss": 0.262,
+      "step": 93340
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3932447056514442e-05,
+      "loss": 0.2798,
+      "step": 93345
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3929892769013208e-05,
+      "loss": 0.2792,
+      "step": 93350
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3927338625250033e-05,
+      "loss": 0.241,
+      "step": 93355
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3924784625258069e-05,
+      "loss": 0.2577,
+      "step": 93360
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3922230769070465e-05,
+      "loss": 0.2489,
+      "step": 93365
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3919677056720405e-05,
+      "loss": 0.2431,
+      "step": 93370
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3917123488241025e-05,
+      "loss": 0.2847,
+      "step": 93375
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3914570063665505e-05,
+      "loss": 0.2718,
+      "step": 93380
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3912016783026984e-05,
+      "loss": 0.2637,
+      "step": 93385
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3909463646358611e-05,
+      "loss": 0.2626,
+      "step": 93390
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.390691065369355e-05,
+      "loss": 0.2238,
+      "step": 93395
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.390435780506495e-05,
+      "loss": 0.2608,
+      "step": 93400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3901805100505943e-05,
+      "loss": 0.2597,
+      "step": 93405
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3899252540049679e-05,
+      "loss": 0.2819,
+      "step": 93410
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3896700123729315e-05,
+      "loss": 0.2807,
+      "step": 93415
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3894147851577979e-05,
+      "loss": 0.2479,
+      "step": 93420
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.389159572362881e-05,
+      "loss": 0.2543,
+      "step": 93425
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3889043739914943e-05,
+      "loss": 0.2714,
+      "step": 93430
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3886491900469534e-05,
+      "loss": 0.258,
+      "step": 93435
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3883940205325701e-05,
+      "loss": 0.2697,
+      "step": 93440
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3881388654516573e-05,
+      "loss": 0.2575,
+      "step": 93445
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3878837248075294e-05,
+      "loss": 0.2603,
+      "step": 93450
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3876285986034981e-05,
+      "loss": 0.2569,
+      "step": 93455
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3873734868428768e-05,
+      "loss": 0.232,
+      "step": 93460
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3871183895289763e-05,
+      "loss": 0.2485,
+      "step": 93465
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3868633066651113e-05,
+      "loss": 0.2465,
+      "step": 93470
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3866082382545925e-05,
+      "loss": 0.2844,
+      "step": 93475
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.386353184300731e-05,
+      "loss": 0.2281,
+      "step": 93480
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3860981448068394e-05,
+      "loss": 0.2563,
+      "step": 93485
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3858431197762306e-05,
+      "loss": 0.2635,
+      "step": 93490
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3855881092122142e-05,
+      "loss": 0.2785,
+      "step": 93495
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3853331131181022e-05,
+      "loss": 0.2459,
+      "step": 93500
+    },
+    {
+      "epoch": 3.29,
+      "eval_loss": 0.2541661560535431,
+      "eval_runtime": 10.5324,
+      "eval_samples_per_second": 9.495,
+      "eval_steps_per_second": 9.495,
+      "step": 93500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3850781314972038e-05,
+      "loss": 0.251,
+      "step": 93505
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3848231643528325e-05,
+      "loss": 0.2656,
+      "step": 93510
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3845682116882972e-05,
+      "loss": 0.2656,
+      "step": 93515
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3843132735069087e-05,
+      "loss": 0.2524,
+      "step": 93520
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3840583498119759e-05,
+      "loss": 0.2857,
+      "step": 93525
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.383803440606811e-05,
+      "loss": 0.2584,
+      "step": 93530
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3835485458947217e-05,
+      "loss": 0.2466,
+      "step": 93535
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3832936656790202e-05,
+      "loss": 0.2683,
+      "step": 93540
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3830387999630132e-05,
+      "loss": 0.2416,
+      "step": 93545
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3827839487500121e-05,
+      "loss": 0.2635,
+      "step": 93550
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3825291120433257e-05,
+      "loss": 0.2477,
+      "step": 93555
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.382274289846261e-05,
+      "loss": 0.2627,
+      "step": 93560
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.382019482162129e-05,
+      "loss": 0.2665,
+      "step": 93565
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3817646889942373e-05,
+      "loss": 0.2546,
+      "step": 93570
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3815099103458942e-05,
+      "loss": 0.2686,
+      "step": 93575
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3812551462204071e-05,
+      "loss": 0.2588,
+      "step": 93580
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3810003966210849e-05,
+      "loss": 0.2995,
+      "step": 93585
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3807456615512348e-05,
+      "loss": 0.2539,
+      "step": 93590
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3804909410141652e-05,
+      "loss": 0.2459,
+      "step": 93595
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3802362350131826e-05,
+      "loss": 0.2339,
+      "step": 93600
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.379981543551595e-05,
+      "loss": 0.2507,
+      "step": 93605
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.379726866632709e-05,
+      "loss": 0.2696,
+      "step": 93610
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3794722042598312e-05,
+      "loss": 0.248,
+      "step": 93615
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3792175564362676e-05,
+      "loss": 0.2534,
+      "step": 93620
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3789629231653264e-05,
+      "loss": 0.2547,
+      "step": 93625
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.378708304450313e-05,
+      "loss": 0.2439,
+      "step": 93630
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.378453700294533e-05,
+      "loss": 0.2218,
+      "step": 93635
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3781991107012914e-05,
+      "loss": 0.2338,
+      "step": 93640
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3779445356738956e-05,
+      "loss": 0.2689,
+      "step": 93645
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3776899752156513e-05,
+      "loss": 0.2705,
+      "step": 93650
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.377435429329863e-05,
+      "loss": 0.2428,
+      "step": 93655
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3771808980198347e-05,
+      "loss": 0.2571,
+      "step": 93660
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3769263812888738e-05,
+      "loss": 0.2609,
+      "step": 93665
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3766718791402836e-05,
+      "loss": 0.2526,
+      "step": 93670
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3764173915773676e-05,
+      "loss": 0.2505,
+      "step": 93675
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3761629186034325e-05,
+      "loss": 0.2853,
+      "step": 93680
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3759084602217814e-05,
+      "loss": 0.2664,
+      "step": 93685
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.375654016435718e-05,
+      "loss": 0.2423,
+      "step": 93690
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3753995872485448e-05,
+      "loss": 0.2628,
+      "step": 93695
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3751451726635684e-05,
+      "loss": 0.2537,
+      "step": 93700
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3748907726840892e-05,
+      "loss": 0.2458,
+      "step": 93705
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3746363873134132e-05,
+      "loss": 0.2434,
+      "step": 93710
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.374382016554841e-05,
+      "loss": 0.2754,
+      "step": 93715
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.374127660411677e-05,
+      "loss": 0.273,
+      "step": 93720
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3738733188872238e-05,
+      "loss": 0.2431,
+      "step": 93725
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3736189919847835e-05,
+      "loss": 0.2741,
+      "step": 93730
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3733646797076571e-05,
+      "loss": 0.2543,
+      "step": 93735
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.373110382059149e-05,
+      "loss": 0.2623,
+      "step": 93740
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3728560990425598e-05,
+      "loss": 0.2502,
+      "step": 93745
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3726018306611905e-05,
+      "loss": 0.2691,
+      "step": 93750
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3723475769183442e-05,
+      "loss": 0.2653,
+      "step": 93755
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3720933378173207e-05,
+      "loss": 0.275,
+      "step": 93760
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3718391133614227e-05,
+      "loss": 0.2911,
+      "step": 93765
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3715849035539503e-05,
+      "loss": 0.252,
+      "step": 93770
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3713307083982035e-05,
+      "loss": 0.2553,
+      "step": 93775
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3710765278974847e-05,
+      "loss": 0.2501,
+      "step": 93780
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3708223620550928e-05,
+      "loss": 0.2694,
+      "step": 93785
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3705682108743276e-05,
+      "loss": 0.26,
+      "step": 93790
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3703140743584909e-05,
+      "loss": 0.2596,
+      "step": 93795
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3700599525108812e-05,
+      "loss": 0.2495,
+      "step": 93800
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3698058453347975e-05,
+      "loss": 0.2526,
+      "step": 93805
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3695517528335411e-05,
+      "loss": 0.2751,
+      "step": 93810
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3692976750104092e-05,
+      "loss": 0.2642,
+      "step": 93815
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3690436118687031e-05,
+      "loss": 0.2701,
+      "step": 93820
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3687895634117199e-05,
+      "loss": 0.2818,
+      "step": 93825
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3685355296427582e-05,
+      "loss": 0.2594,
+      "step": 93830
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3682815105651175e-05,
+      "loss": 0.2336,
+      "step": 93835
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3680275061820957e-05,
+      "loss": 0.2715,
+      "step": 93840
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3677735164969909e-05,
+      "loss": 0.264,
+      "step": 93845
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3675195415130999e-05,
+      "loss": 0.2618,
+      "step": 93850
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3672655812337221e-05,
+      "loss": 0.2503,
+      "step": 93855
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3670116356621544e-05,
+      "loss": 0.2506,
+      "step": 93860
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3667577048016927e-05,
+      "loss": 0.2564,
+      "step": 93865
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3665037886556353e-05,
+      "loss": 0.2567,
+      "step": 93870
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3662498872272806e-05,
+      "loss": 0.244,
+      "step": 93875
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3659960005199238e-05,
+      "loss": 0.2331,
+      "step": 93880
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3657421285368615e-05,
+      "loss": 0.2539,
+      "step": 93885
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3654882712813893e-05,
+      "loss": 0.2625,
+      "step": 93890
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.365234428756805e-05,
+      "loss": 0.2449,
+      "step": 93895
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.364980600966404e-05,
+      "loss": 0.2693,
+      "step": 93900
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3647267879134807e-05,
+      "loss": 0.2634,
+      "step": 93905
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3644729896013331e-05,
+      "loss": 0.2605,
+      "step": 93910
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3642192060332554e-05,
+      "loss": 0.2513,
+      "step": 93915
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3639654372125416e-05,
+      "loss": 0.2845,
+      "step": 93920
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3637116831424879e-05,
+      "loss": 0.2483,
+      "step": 93925
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3634579438263902e-05,
+      "loss": 0.2557,
+      "step": 93930
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3632042192675424e-05,
+      "loss": 0.2434,
+      "step": 93935
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3629505094692382e-05,
+      "loss": 0.2606,
+      "step": 93940
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3626968144347715e-05,
+      "loss": 0.2532,
+      "step": 93945
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3624431341674383e-05,
+      "loss": 0.2282,
+      "step": 93950
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.362189468670531e-05,
+      "loss": 0.2614,
+      "step": 93955
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3619358179473438e-05,
+      "loss": 0.2579,
+      "step": 93960
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3616821820011689e-05,
+      "loss": 0.2629,
+      "step": 93965
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3614285608353017e-05,
+      "loss": 0.2686,
+      "step": 93970
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3611749544530334e-05,
+      "loss": 0.2524,
+      "step": 93975
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.360921362857659e-05,
+      "loss": 0.2663,
+      "step": 93980
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3606677860524686e-05,
+      "loss": 0.2573,
+      "step": 93985
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3604142240407575e-05,
+      "loss": 0.2445,
+      "step": 93990
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3601606768258163e-05,
+      "loss": 0.2402,
+      "step": 93995
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3599071444109378e-05,
+      "loss": 0.2714,
+      "step": 94000
+    },
+    {
+      "epoch": 3.31,
+      "eval_loss": 0.2543759047985077,
+      "eval_runtime": 10.548,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 94000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3596536267994126e-05,
+      "loss": 0.2596,
+      "step": 94005
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3594001239945345e-05,
+      "loss": 0.2738,
+      "step": 94010
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3591466359995942e-05,
+      "loss": 0.2628,
+      "step": 94015
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.358893162817882e-05,
+      "loss": 0.2545,
+      "step": 94020
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.358639704452691e-05,
+      "loss": 0.2568,
+      "step": 94025
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.35838626090731e-05,
+      "loss": 0.2592,
+      "step": 94030
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3581328321850322e-05,
+      "loss": 0.2703,
+      "step": 94035
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3578794182891464e-05,
+      "loss": 0.2574,
+      "step": 94040
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3576260192229445e-05,
+      "loss": 0.2624,
+      "step": 94045
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3573726349897159e-05,
+      "loss": 0.2587,
+      "step": 94050
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3571192655927503e-05,
+      "loss": 0.2631,
+      "step": 94055
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.356865911035337e-05,
+      "loss": 0.2539,
+      "step": 94060
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3566125713207678e-05,
+      "loss": 0.25,
+      "step": 94065
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3563592464523305e-05,
+      "loss": 0.2691,
+      "step": 94070
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3561059364333147e-05,
+      "loss": 0.2674,
+      "step": 94075
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3558526412670086e-05,
+      "loss": 0.2457,
+      "step": 94080
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.355599360956702e-05,
+      "loss": 0.262,
+      "step": 94085
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3553460955056845e-05,
+      "loss": 0.2832,
+      "step": 94090
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3550928449172433e-05,
+      "loss": 0.2359,
+      "step": 94095
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3548396091946663e-05,
+      "loss": 0.2592,
+      "step": 94100
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3545863883412432e-05,
+      "loss": 0.2594,
+      "step": 94105
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3543331823602613e-05,
+      "loss": 0.2325,
+      "step": 94110
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3540799912550073e-05,
+      "loss": 0.2471,
+      "step": 94115
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3538268150287688e-05,
+      "loss": 0.2797,
+      "step": 94120
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3535736536848349e-05,
+      "loss": 0.2508,
+      "step": 94125
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3533205072264916e-05,
+      "loss": 0.2521,
+      "step": 94130
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3530673756570245e-05,
+      "loss": 0.2397,
+      "step": 94135
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3528142589797227e-05,
+      "loss": 0.2531,
+      "step": 94140
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3525611571978708e-05,
+      "loss": 0.2361,
+      "step": 94145
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3523080703147572e-05,
+      "loss": 0.2691,
+      "step": 94150
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3520549983336658e-05,
+      "loss": 0.2774,
+      "step": 94155
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3518019412578848e-05,
+      "loss": 0.2514,
+      "step": 94160
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3515488990906989e-05,
+      "loss": 0.266,
+      "step": 94165
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3512958718353937e-05,
+      "loss": 0.2292,
+      "step": 94170
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3510428594952537e-05,
+      "loss": 0.2441,
+      "step": 94175
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3507898620735659e-05,
+      "loss": 0.2651,
+      "step": 94180
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3505368795736143e-05,
+      "loss": 0.2452,
+      "step": 94185
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3502839119986827e-05,
+      "loss": 0.2532,
+      "step": 94190
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.350030959352058e-05,
+      "loss": 0.2476,
+      "step": 94195
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3497780216370226e-05,
+      "loss": 0.2817,
+      "step": 94200
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.349525098856862e-05,
+      "loss": 0.2676,
+      "step": 94205
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3492721910148604e-05,
+      "loss": 0.246,
+      "step": 94210
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3490192981142996e-05,
+      "loss": 0.2373,
+      "step": 94215
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3487664201584659e-05,
+      "loss": 0.2551,
+      "step": 94220
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3485135571506416e-05,
+      "loss": 0.2407,
+      "step": 94225
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3482607090941096e-05,
+      "loss": 0.2685,
+      "step": 94230
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3480078759921527e-05,
+      "loss": 0.2463,
+      "step": 94235
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3477550578480553e-05,
+      "loss": 0.2558,
+      "step": 94240
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3475022546650979e-05,
+      "loss": 0.2484,
+      "step": 94245
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3472494664465652e-05,
+      "loss": 0.2492,
+      "step": 94250
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3469966931957374e-05,
+      "loss": 0.2552,
+      "step": 94255
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.346743934915899e-05,
+      "loss": 0.2591,
+      "step": 94260
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.34649119161033e-05,
+      "loss": 0.2093,
+      "step": 94265
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3462384632823122e-05,
+      "loss": 0.2402,
+      "step": 94270
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3459857499351286e-05,
+      "loss": 0.2607,
+      "step": 94275
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3457330515720592e-05,
+      "loss": 0.2511,
+      "step": 94280
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3454803681963856e-05,
+      "loss": 0.253,
+      "step": 94285
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3452276998113878e-05,
+      "loss": 0.2711,
+      "step": 94290
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.344975046420348e-05,
+      "loss": 0.2385,
+      "step": 94295
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3447224080265462e-05,
+      "loss": 0.2699,
+      "step": 94300
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3444697846332612e-05,
+      "loss": 0.2332,
+      "step": 94305
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.344217176243775e-05,
+      "loss": 0.2502,
+      "step": 94310
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3439645828613678e-05,
+      "loss": 0.2498,
+      "step": 94315
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3437120044893187e-05,
+      "loss": 0.2674,
+      "step": 94320
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.343459441130907e-05,
+      "loss": 0.2475,
+      "step": 94325
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3432068927894118e-05,
+      "loss": 0.2423,
+      "step": 94330
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3429543594681133e-05,
+      "loss": 0.267,
+      "step": 94335
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3427018411702901e-05,
+      "loss": 0.2778,
+      "step": 94340
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.342449337899221e-05,
+      "loss": 0.2546,
+      "step": 94345
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3421968496581827e-05,
+      "loss": 0.2568,
+      "step": 94350
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3419443764504569e-05,
+      "loss": 0.2568,
+      "step": 94355
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3416919182793192e-05,
+      "loss": 0.2476,
+      "step": 94360
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3414394751480492e-05,
+      "loss": 0.2513,
+      "step": 94365
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3411870470599233e-05,
+      "loss": 0.2527,
+      "step": 94370
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.340934634018221e-05,
+      "loss": 0.2936,
+      "step": 94375
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3406822360262183e-05,
+      "loss": 0.2555,
+      "step": 94380
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3404298530871922e-05,
+      "loss": 0.2488,
+      "step": 94385
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3401774852044211e-05,
+      "loss": 0.2521,
+      "step": 94390
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3399251323811807e-05,
+      "loss": 0.2486,
+      "step": 94395
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3396727946207482e-05,
+      "loss": 0.2644,
+      "step": 94400
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3394204719263989e-05,
+      "loss": 0.2582,
+      "step": 94405
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3391681643014109e-05,
+      "loss": 0.2385,
+      "step": 94410
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.338915871749058e-05,
+      "loss": 0.2645,
+      "step": 94415
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3386635942726183e-05,
+      "loss": 0.2564,
+      "step": 94420
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.338411331875366e-05,
+      "loss": 0.255,
+      "step": 94425
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3381590845605776e-05,
+      "loss": 0.2679,
+      "step": 94430
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.337906852331528e-05,
+      "loss": 0.2662,
+      "step": 94435
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.337654635191492e-05,
+      "loss": 0.2493,
+      "step": 94440
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3374024331437435e-05,
+      "loss": 0.2521,
+      "step": 94445
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3371502461915592e-05,
+      "loss": 0.2722,
+      "step": 94450
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3368980743382126e-05,
+      "loss": 0.2694,
+      "step": 94455
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.336645917586978e-05,
+      "loss": 0.2432,
+      "step": 94460
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3363937759411287e-05,
+      "loss": 0.2721,
+      "step": 94465
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.336141649403939e-05,
+      "loss": 0.2625,
+      "step": 94470
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3358895379786842e-05,
+      "loss": 0.2666,
+      "step": 94475
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3356374416686368e-05,
+      "loss": 0.2839,
+      "step": 94480
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3353853604770683e-05,
+      "loss": 0.2386,
+      "step": 94485
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3351332944072548e-05,
+      "loss": 0.2651,
+      "step": 94490
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3348812434624677e-05,
+      "loss": 0.2861,
+      "step": 94495
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3346292076459788e-05,
+      "loss": 0.261,
+      "step": 94500
+    },
+    {
+      "epoch": 3.32,
+      "eval_loss": 0.25450006127357483,
+      "eval_runtime": 10.5436,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 94500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3343771869610627e-05,
+      "loss": 0.2545,
+      "step": 94505
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3341251814109906e-05,
+      "loss": 0.2607,
+      "step": 94510
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3338731909990348e-05,
+      "loss": 0.2558,
+      "step": 94515
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3336212157284658e-05,
+      "loss": 0.2538,
+      "step": 94520
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3333692556025568e-05,
+      "loss": 0.2324,
+      "step": 94525
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3331173106245804e-05,
+      "loss": 0.2737,
+      "step": 94530
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3328653807978064e-05,
+      "loss": 0.2455,
+      "step": 94535
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3326134661255055e-05,
+      "loss": 0.2488,
+      "step": 94540
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3323615666109502e-05,
+      "loss": 0.2556,
+      "step": 94545
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3321096822574106e-05,
+      "loss": 0.2651,
+      "step": 94550
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.331857813068157e-05,
+      "loss": 0.2757,
+      "step": 94555
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3316059590464591e-05,
+      "loss": 0.2661,
+      "step": 94560
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3313541201955885e-05,
+      "loss": 0.2849,
+      "step": 94565
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3311022965188147e-05,
+      "loss": 0.2361,
+      "step": 94570
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.330850488019406e-05,
+      "loss": 0.2496,
+      "step": 94575
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3305986947006343e-05,
+      "loss": 0.2545,
+      "step": 94580
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3303469165657673e-05,
+      "loss": 0.2494,
+      "step": 94585
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.330095153618075e-05,
+      "loss": 0.2668,
+      "step": 94590
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3298434058608267e-05,
+      "loss": 0.2508,
+      "step": 94595
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3295916732972896e-05,
+      "loss": 0.2632,
+      "step": 94600
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.329339955930734e-05,
+      "loss": 0.2781,
+      "step": 94605
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3290882537644278e-05,
+      "loss": 0.2595,
+      "step": 94610
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3288365668016378e-05,
+      "loss": 0.264,
+      "step": 94615
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3285848950456341e-05,
+      "loss": 0.2602,
+      "step": 94620
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3283332384996838e-05,
+      "loss": 0.2481,
+      "step": 94625
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3280815971670535e-05,
+      "loss": 0.2513,
+      "step": 94630
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3278299710510117e-05,
+      "loss": 0.2562,
+      "step": 94635
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3275783601548248e-05,
+      "loss": 0.2786,
+      "step": 94640
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3273267644817607e-05,
+      "loss": 0.2617,
+      "step": 94645
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3270751840350861e-05,
+      "loss": 0.2527,
+      "step": 94650
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3268236188180661e-05,
+      "loss": 0.2464,
+      "step": 94655
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3265720688339695e-05,
+      "loss": 0.2613,
+      "step": 94660
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.326320534086061e-05,
+      "loss": 0.2414,
+      "step": 94665
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.326069014577607e-05,
+      "loss": 0.2516,
+      "step": 94670
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3258175103118719e-05,
+      "loss": 0.2528,
+      "step": 94675
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.325566021292124e-05,
+      "loss": 0.2653,
+      "step": 94680
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3253145475216261e-05,
+      "loss": 0.2526,
+      "step": 94685
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3250630890036458e-05,
+      "loss": 0.2568,
+      "step": 94690
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3248116457414461e-05,
+      "loss": 0.289,
+      "step": 94695
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3245602177382936e-05,
+      "loss": 0.2483,
+      "step": 94700
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3243088049974516e-05,
+      "loss": 0.2605,
+      "step": 94705
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3240574075221856e-05,
+      "loss": 0.2474,
+      "step": 94710
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.323806025315758e-05,
+      "loss": 0.2783,
+      "step": 94715
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3235546583814348e-05,
+      "loss": 0.2654,
+      "step": 94720
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3233033067224792e-05,
+      "loss": 0.2342,
+      "step": 94725
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3230519703421537e-05,
+      "loss": 0.2719,
+      "step": 94730
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3228006492437236e-05,
+      "loss": 0.2502,
+      "step": 94735
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3225493434304514e-05,
+      "loss": 0.2529,
+      "step": 94740
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.322298052905599e-05,
+      "loss": 0.2508,
+      "step": 94745
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3220467776724301e-05,
+      "loss": 0.252,
+      "step": 94750
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3217955177342087e-05,
+      "loss": 0.2732,
+      "step": 94755
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.321544273094196e-05,
+      "loss": 0.264,
+      "step": 94760
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3212930437556542e-05,
+      "loss": 0.2485,
+      "step": 94765
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3210418297218446e-05,
+      "loss": 0.2582,
+      "step": 94770
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3207906309960305e-05,
+      "loss": 0.2761,
+      "step": 94775
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.3205394475814732e-05,
+      "loss": 0.2472,
+      "step": 94780
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.320288279481434e-05,
+      "loss": 0.2437,
+      "step": 94785
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.320037126699173e-05,
+      "loss": 0.2663,
+      "step": 94790
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.319785989237953e-05,
+      "loss": 0.2613,
+      "step": 94795
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3195348671010333e-05,
+      "loss": 0.2638,
+      "step": 94800
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3192837602916764e-05,
+      "loss": 0.2416,
+      "step": 94805
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3190326688131407e-05,
+      "loss": 0.2669,
+      "step": 94810
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3187815926686886e-05,
+      "loss": 0.2434,
+      "step": 94815
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.318530531861579e-05,
+      "loss": 0.2341,
+      "step": 94820
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.318279486395072e-05,
+      "loss": 0.2696,
+      "step": 94825
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.318028456272426e-05,
+      "loss": 0.2584,
+      "step": 94830
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3177774414969023e-05,
+      "loss": 0.2407,
+      "step": 94835
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3175264420717594e-05,
+      "loss": 0.2605,
+      "step": 94840
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3172754580002556e-05,
+      "loss": 0.2672,
+      "step": 94845
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3170746817998638e-05,
+      "loss": 0.2632,
+      "step": 94850
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.316823725373125e-05,
+      "loss": 0.2434,
+      "step": 94855
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3165727843091493e-05,
+      "loss": 0.2421,
+      "step": 94860
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3163218586111972e-05,
+      "loss": 0.2645,
+      "step": 94865
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3160709482825257e-05,
+      "loss": 0.2445,
+      "step": 94870
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3158200533263926e-05,
+      "loss": 0.2647,
+      "step": 94875
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3155691737460545e-05,
+      "loss": 0.2784,
+      "step": 94880
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3153183095447713e-05,
+      "loss": 0.25,
+      "step": 94885
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3150674607257983e-05,
+      "loss": 0.2598,
+      "step": 94890
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.314816627292394e-05,
+      "loss": 0.2393,
+      "step": 94895
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3145658092478141e-05,
+      "loss": 0.2418,
+      "step": 94900
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.314315006595317e-05,
+      "loss": 0.2527,
+      "step": 94905
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3140642193381578e-05,
+      "loss": 0.2348,
+      "step": 94910
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3138134474795921e-05,
+      "loss": 0.2663,
+      "step": 94915
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3135626910228784e-05,
+      "loss": 0.2632,
+      "step": 94920
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3133119499712712e-05,
+      "loss": 0.2536,
+      "step": 94925
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.313061224328026e-05,
+      "loss": 0.2499,
+      "step": 94930
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3128105140963981e-05,
+      "loss": 0.2533,
+      "step": 94935
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.312559819279644e-05,
+      "loss": 0.2737,
+      "step": 94940
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.312309139881017e-05,
+      "loss": 0.2586,
+      "step": 94945
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3120584759037741e-05,
+      "loss": 0.2556,
+      "step": 94950
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.311807827351168e-05,
+      "loss": 0.2735,
+      "step": 94955
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3115571942264554e-05,
+      "loss": 0.2577,
+      "step": 94960
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3113065765328892e-05,
+      "loss": 0.2492,
+      "step": 94965
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3110559742737239e-05,
+      "loss": 0.2564,
+      "step": 94970
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.310805387452212e-05,
+      "loss": 0.2664,
+      "step": 94975
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3105548160716097e-05,
+      "loss": 0.2677,
+      "step": 94980
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.310304260135169e-05,
+      "loss": 0.2575,
+      "step": 94985
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3100537196461427e-05,
+      "loss": 0.2438,
+      "step": 94990
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3098031946077854e-05,
+      "loss": 0.2433,
+      "step": 94995
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3095526850233486e-05,
+      "loss": 0.2506,
+      "step": 95000
+    },
+    {
+      "epoch": 3.34,
+      "eval_loss": 0.25423306226730347,
+      "eval_runtime": 10.5407,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 95000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3093021908960862e-05,
+      "loss": 0.2546,
+      "step": 95005
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3090517122292505e-05,
+      "loss": 0.234,
+      "step": 95010
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3088012490260924e-05,
+      "loss": 0.2825,
+      "step": 95015
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3085508012898659e-05,
+      "loss": 0.2709,
+      "step": 95020
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.308300369023822e-05,
+      "loss": 0.2723,
+      "step": 95025
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3080499522312118e-05,
+      "loss": 0.2856,
+      "step": 95030
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3077995509152882e-05,
+      "loss": 0.2701,
+      "step": 95035
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3075491650793017e-05,
+      "loss": 0.2622,
+      "step": 95040
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3072987947265037e-05,
+      "loss": 0.2584,
+      "step": 95045
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3070484398601434e-05,
+      "loss": 0.2455,
+      "step": 95050
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3067981004834734e-05,
+      "loss": 0.2491,
+      "step": 95055
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3065477765997442e-05,
+      "loss": 0.2469,
+      "step": 95060
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3062974682122059e-05,
+      "loss": 0.2695,
+      "step": 95065
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3060471753241074e-05,
+      "loss": 0.2539,
+      "step": 95070
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3057968979387004e-05,
+      "loss": 0.2544,
+      "step": 95075
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3055466360592337e-05,
+      "loss": 0.2343,
+      "step": 95080
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3052963896889565e-05,
+      "loss": 0.2827,
+      "step": 95085
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3050461588311175e-05,
+      "loss": 0.2629,
+      "step": 95090
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3047959434889679e-05,
+      "loss": 0.2746,
+      "step": 95095
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.304545743665755e-05,
+      "loss": 0.2742,
+      "step": 95100
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3042955593647267e-05,
+      "loss": 0.2586,
+      "step": 95105
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3040453905891328e-05,
+      "loss": 0.2966,
+      "step": 95110
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3037952373422225e-05,
+      "loss": 0.2524,
+      "step": 95115
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3035450996272422e-05,
+      "loss": 0.2349,
+      "step": 95120
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3032949774474407e-05,
+      "loss": 0.2827,
+      "step": 95125
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3030448708060639e-05,
+      "loss": 0.2449,
+      "step": 95130
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3027947797063619e-05,
+      "loss": 0.2624,
+      "step": 95135
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3025447041515806e-05,
+      "loss": 0.2681,
+      "step": 95140
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3022946441449663e-05,
+      "loss": 0.2653,
+      "step": 95145
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3020445996897673e-05,
+      "loss": 0.2469,
+      "step": 95150
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3017945707892305e-05,
+      "loss": 0.2505,
+      "step": 95155
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3015445574465999e-05,
+      "loss": 0.249,
+      "step": 95160
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3012945596651243e-05,
+      "loss": 0.2543,
+      "step": 95165
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3010445774480484e-05,
+      "loss": 0.2588,
+      "step": 95170
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3007946107986191e-05,
+      "loss": 0.2457,
+      "step": 95175
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3005446597200815e-05,
+      "loss": 0.2692,
+      "step": 95180
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3002947242156805e-05,
+      "loss": 0.2647,
+      "step": 95185
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3000448042886626e-05,
+      "loss": 0.2713,
+      "step": 95190
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2997948999422719e-05,
+      "loss": 0.2444,
+      "step": 95195
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2995450111797535e-05,
+      "loss": 0.2554,
+      "step": 95200
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.299295138004351e-05,
+      "loss": 0.2439,
+      "step": 95205
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.299045280419311e-05,
+      "loss": 0.2501,
+      "step": 95210
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2987954384278756e-05,
+      "loss": 0.2727,
+      "step": 95215
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2985456120332906e-05,
+      "loss": 0.2381,
+      "step": 95220
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2982958012387985e-05,
+      "loss": 0.2882,
+      "step": 95225
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.298046006047644e-05,
+      "loss": 0.247,
+      "step": 95230
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.29779622646307e-05,
+      "loss": 0.2704,
+      "step": 95235
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.29754646248832e-05,
+      "loss": 0.2382,
+      "step": 95240
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2972967141266357e-05,
+      "loss": 0.2845,
+      "step": 95245
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2970469813812619e-05,
+      "loss": 0.2614,
+      "step": 95250
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.29679726425544e-05,
+      "loss": 0.2622,
+      "step": 95255
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2965475627524122e-05,
+      "loss": 0.2555,
+      "step": 95260
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2962978768754217e-05,
+      "loss": 0.2572,
+      "step": 95265
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.296048206627709e-05,
+      "loss": 0.2654,
+      "step": 95270
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2957985520125182e-05,
+      "loss": 0.2568,
+      "step": 95275
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2955489130330884e-05,
+      "loss": 0.2593,
+      "step": 95280
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2952992896926636e-05,
+      "loss": 0.2639,
+      "step": 95285
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.295049681994483e-05,
+      "loss": 0.2456,
+      "step": 95290
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2948000899417887e-05,
+      "loss": 0.2499,
+      "step": 95295
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2945505135378196e-05,
+      "loss": 0.2455,
+      "step": 95300
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2943009527858187e-05,
+      "loss": 0.2623,
+      "step": 95305
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2940514076890257e-05,
+      "loss": 0.2515,
+      "step": 95310
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.29380187825068e-05,
+      "loss": 0.2423,
+      "step": 95315
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2935523644740213e-05,
+      "loss": 0.256,
+      "step": 95320
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2933028663622909e-05,
+      "loss": 0.2667,
+      "step": 95325
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2930533839187265e-05,
+      "loss": 0.271,
+      "step": 95330
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2928039171465697e-05,
+      "loss": 0.2523,
+      "step": 95335
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2925544660490573e-05,
+      "loss": 0.2573,
+      "step": 95340
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2923050306294305e-05,
+      "loss": 0.262,
+      "step": 95345
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2920556108909271e-05,
+      "loss": 0.2591,
+      "step": 95350
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2918062068367853e-05,
+      "loss": 0.2597,
+      "step": 95355
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2915568184702425e-05,
+      "loss": 0.2418,
+      "step": 95360
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2913074457945396e-05,
+      "loss": 0.2486,
+      "step": 95365
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2910580888129126e-05,
+      "loss": 0.2508,
+      "step": 95370
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2908087475285984e-05,
+      "loss": 0.2783,
+      "step": 95375
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2905594219448368e-05,
+      "loss": 0.2565,
+      "step": 95380
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2903101120648637e-05,
+      "loss": 0.2734,
+      "step": 95385
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.290060817891917e-05,
+      "loss": 0.2653,
+      "step": 95390
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2898115394292321e-05,
+      "loss": 0.2656,
+      "step": 95395
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.289562276680048e-05,
+      "loss": 0.2528,
+      "step": 95400
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2893130296476e-05,
+      "loss": 0.2752,
+      "step": 95405
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2890637983351244e-05,
+      "loss": 0.2442,
+      "step": 95410
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2888145827458564e-05,
+      "loss": 0.2665,
+      "step": 95415
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.288565382883034e-05,
+      "loss": 0.2613,
+      "step": 95420
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2883161987498915e-05,
+      "loss": 0.2647,
+      "step": 95425
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.288067030349664e-05,
+      "loss": 0.2593,
+      "step": 95430
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.287817877685588e-05,
+      "loss": 0.2676,
+      "step": 95435
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2875687407608972e-05,
+      "loss": 0.2488,
+      "step": 95440
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2873196195788279e-05,
+      "loss": 0.263,
+      "step": 95445
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2870705141426146e-05,
+      "loss": 0.2592,
+      "step": 95450
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2868214244554902e-05,
+      "loss": 0.2432,
+      "step": 95455
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2865723505206912e-05,
+      "loss": 0.2637,
+      "step": 95460
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2863232923414504e-05,
+      "loss": 0.2573,
+      "step": 95465
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2860742499210016e-05,
+      "loss": 0.2492,
+      "step": 95470
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2858252232625778e-05,
+      "loss": 0.2399,
+      "step": 95475
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2855762123694145e-05,
+      "loss": 0.2297,
+      "step": 95480
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2853272172447437e-05,
+      "loss": 0.2794,
+      "step": 95485
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2850782378917975e-05,
+      "loss": 0.2753,
+      "step": 95490
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2848292743138097e-05,
+      "loss": 0.2583,
+      "step": 95495
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2845803265140138e-05,
+      "loss": 0.2693,
+      "step": 95500
+    },
+    {
+      "epoch": 3.36,
+      "eval_loss": 0.2542358934879303,
+      "eval_runtime": 10.5386,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 95500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2843313944956414e-05,
+      "loss": 0.2593,
+      "step": 95505
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2840824782619237e-05,
+      "loss": 0.2635,
+      "step": 95510
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2838335778160948e-05,
+      "loss": 0.2435,
+      "step": 95515
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2835846931613853e-05,
+      "loss": 0.2347,
+      "step": 95520
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2833358243010268e-05,
+      "loss": 0.2579,
+      "step": 95525
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2830869712382504e-05,
+      "loss": 0.2742,
+      "step": 95530
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.282838133976288e-05,
+      "loss": 0.261,
+      "step": 95535
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2825893125183709e-05,
+      "loss": 0.2383,
+      "step": 95540
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2823405068677277e-05,
+      "loss": 0.2771,
+      "step": 95545
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2820917170275909e-05,
+      "loss": 0.2735,
+      "step": 95550
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2818429430011914e-05,
+      "loss": 0.2574,
+      "step": 95555
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2815941847917584e-05,
+      "loss": 0.2508,
+      "step": 95560
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.281345442402522e-05,
+      "loss": 0.2654,
+      "step": 95565
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.281096715836711e-05,
+      "loss": 0.254,
+      "step": 95570
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2808480050975569e-05,
+      "loss": 0.262,
+      "step": 95575
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.280599310188288e-05,
+      "loss": 0.2479,
+      "step": 95580
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2803506311121332e-05,
+      "loss": 0.2974,
+      "step": 95585
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2801019678723208e-05,
+      "loss": 0.2636,
+      "step": 95590
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2798533204720814e-05,
+      "loss": 0.2263,
+      "step": 95595
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2796046889146417e-05,
+      "loss": 0.2639,
+      "step": 95600
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2793560732032317e-05,
+      "loss": 0.254,
+      "step": 95605
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2791074733410776e-05,
+      "loss": 0.2232,
+      "step": 95610
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2788588893314096e-05,
+      "loss": 0.2799,
+      "step": 95615
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2786103211774538e-05,
+      "loss": 0.2552,
+      "step": 95620
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2783617688824373e-05,
+      "loss": 0.2556,
+      "step": 95625
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2781132324495892e-05,
+      "loss": 0.2572,
+      "step": 95630
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2778647118821352e-05,
+      "loss": 0.2541,
+      "step": 95635
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2776162071833028e-05,
+      "loss": 0.2767,
+      "step": 95640
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2773677183563171e-05,
+      "loss": 0.2583,
+      "step": 95645
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2771192454044073e-05,
+      "loss": 0.2749,
+      "step": 95650
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2768707883307968e-05,
+      "loss": 0.26,
+      "step": 95655
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.276622347138714e-05,
+      "loss": 0.2622,
+      "step": 95660
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.276373921831383e-05,
+      "loss": 0.2501,
+      "step": 95665
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2761255124120313e-05,
+      "loss": 0.2671,
+      "step": 95670
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2758771188838831e-05,
+      "loss": 0.2439,
+      "step": 95675
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2756287412501638e-05,
+      "loss": 0.2651,
+      "step": 95680
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2753803795140973e-05,
+      "loss": 0.2558,
+      "step": 95685
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2751320336789108e-05,
+      "loss": 0.2531,
+      "step": 95690
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2748837037478272e-05,
+      "loss": 0.2478,
+      "step": 95695
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2746353897240715e-05,
+      "loss": 0.2585,
+      "step": 95700
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2743870916108668e-05,
+      "loss": 0.2654,
+      "step": 95705
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2741388094114382e-05,
+      "loss": 0.2564,
+      "step": 95710
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.27389054312901e-05,
+      "loss": 0.2517,
+      "step": 95715
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2736422927668052e-05,
+      "loss": 0.2956,
+      "step": 95720
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2733940583280457e-05,
+      "loss": 0.2528,
+      "step": 95725
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2731458398159573e-05,
+      "loss": 0.2563,
+      "step": 95730
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2728976372337617e-05,
+      "loss": 0.2618,
+      "step": 95735
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2726494505846803e-05,
+      "loss": 0.2516,
+      "step": 95740
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.272401279871938e-05,
+      "loss": 0.2465,
+      "step": 95745
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2721531250987565e-05,
+      "loss": 0.2483,
+      "step": 95750
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2719049862683572e-05,
+      "loss": 0.2505,
+      "step": 95755
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2716568633839612e-05,
+      "loss": 0.2483,
+      "step": 95760
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2714087564487925e-05,
+      "loss": 0.2572,
+      "step": 95765
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2711606654660701e-05,
+      "loss": 0.2597,
+      "step": 95770
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.270912590439018e-05,
+      "loss": 0.2515,
+      "step": 95775
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2706645313708548e-05,
+      "loss": 0.2443,
+      "step": 95780
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2704164882648034e-05,
+      "loss": 0.2494,
+      "step": 95785
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2701684611240835e-05,
+      "loss": 0.2548,
+      "step": 95790
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2699204499519157e-05,
+      "loss": 0.2632,
+      "step": 95795
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2696724547515193e-05,
+      "loss": 0.2684,
+      "step": 95800
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2694244755261164e-05,
+      "loss": 0.2437,
+      "step": 95805
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2691765122789256e-05,
+      "loss": 0.2587,
+      "step": 95810
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2689285650131654e-05,
+      "loss": 0.2623,
+      "step": 95815
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2686806337320578e-05,
+      "loss": 0.2859,
+      "step": 95820
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2684327184388195e-05,
+      "loss": 0.2551,
+      "step": 95825
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2681848191366719e-05,
+      "loss": 0.2557,
+      "step": 95830
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2679369358288323e-05,
+      "loss": 0.2455,
+      "step": 95835
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2676890685185188e-05,
+      "loss": 0.2614,
+      "step": 95840
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2674412172089517e-05,
+      "loss": 0.2673,
+      "step": 95845
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2671933819033482e-05,
+      "loss": 0.255,
+      "step": 95850
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2669455626049251e-05,
+      "loss": 0.2618,
+      "step": 95855
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.266697759316902e-05,
+      "loss": 0.247,
+      "step": 95860
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.266449972042496e-05,
+      "loss": 0.2691,
+      "step": 95865
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2662022007849234e-05,
+      "loss": 0.2587,
+      "step": 95870
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2659544455474026e-05,
+      "loss": 0.2492,
+      "step": 95875
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2657067063331492e-05,
+      "loss": 0.2823,
+      "step": 95880
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.265458983145382e-05,
+      "loss": 0.2437,
+      "step": 95885
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.265211275987316e-05,
+      "loss": 0.2672,
+      "step": 95890
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2649635848621672e-05,
+      "loss": 0.2517,
+      "step": 95895
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2647159097731532e-05,
+      "loss": 0.2606,
+      "step": 95900
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.264468250723489e-05,
+      "loss": 0.2482,
+      "step": 95905
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.26422060771639e-05,
+      "loss": 0.2635,
+      "step": 95910
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2639729807550715e-05,
+      "loss": 0.2477,
+      "step": 95915
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2637253698427498e-05,
+      "loss": 0.2564,
+      "step": 95920
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.26347777498264e-05,
+      "loss": 0.2578,
+      "step": 95925
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2632301961779547e-05,
+      "loss": 0.2635,
+      "step": 95930
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2629826334319108e-05,
+      "loss": 0.2843,
+      "step": 95935
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2627350867477231e-05,
+      "loss": 0.2645,
+      "step": 95940
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2624875561286049e-05,
+      "loss": 0.262,
+      "step": 95945
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2622400415777699e-05,
+      "loss": 0.2535,
+      "step": 95950
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2619925430984313e-05,
+      "loss": 0.2517,
+      "step": 95955
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2617450606938052e-05,
+      "loss": 0.2612,
+      "step": 95960
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2614975943671031e-05,
+      "loss": 0.2554,
+      "step": 95965
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2612501441215374e-05,
+      "loss": 0.2715,
+      "step": 95970
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2610027099603233e-05,
+      "loss": 0.2533,
+      "step": 95975
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2607552918866725e-05,
+      "loss": 0.2504,
+      "step": 95980
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2605078899037964e-05,
+      "loss": 0.2624,
+      "step": 95985
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.260260504014909e-05,
+      "loss": 0.2599,
+      "step": 95990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2600131342232224e-05,
+      "loss": 0.2644,
+      "step": 95995
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2597657805319485e-05,
+      "loss": 0.2574,
+      "step": 96000
+    },
+    {
+      "epoch": 3.38,
+      "eval_loss": 0.2541818916797638,
+      "eval_runtime": 10.5477,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 96000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.259518442944298e-05,
+      "loss": 0.2428,
+      "step": 96005
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2592711214634826e-05,
+      "loss": 0.251,
+      "step": 96010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2590238160927148e-05,
+      "loss": 0.2491,
+      "step": 96015
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2587765268352048e-05,
+      "loss": 0.2583,
+      "step": 96020
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2585292536941637e-05,
+      "loss": 0.2616,
+      "step": 96025
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.258281996672801e-05,
+      "loss": 0.2655,
+      "step": 96030
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2580347557743294e-05,
+      "loss": 0.2491,
+      "step": 96035
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2577875310019571e-05,
+      "loss": 0.2575,
+      "step": 96040
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2575403223588956e-05,
+      "loss": 0.2572,
+      "step": 96045
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2572931298483537e-05,
+      "loss": 0.2645,
+      "step": 96050
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2570459534735424e-05,
+      "loss": 0.2598,
+      "step": 96055
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.25679879323767e-05,
+      "loss": 0.2502,
+      "step": 96060
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.256551649143946e-05,
+      "loss": 0.2473,
+      "step": 96065
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2563045211955787e-05,
+      "loss": 0.2557,
+      "step": 96070
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2560574093957784e-05,
+      "loss": 0.2662,
+      "step": 96075
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2558103137477527e-05,
+      "loss": 0.2786,
+      "step": 96080
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2555632342547096e-05,
+      "loss": 0.2703,
+      "step": 96085
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2553161709198583e-05,
+      "loss": 0.2625,
+      "step": 96090
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2550691237464054e-05,
+      "loss": 0.2605,
+      "step": 96095
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2548220927375604e-05,
+      "loss": 0.228,
+      "step": 96100
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2545750778965292e-05,
+      "loss": 0.2524,
+      "step": 96105
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2543280792265206e-05,
+      "loss": 0.2701,
+      "step": 96110
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2540810967307409e-05,
+      "loss": 0.2554,
+      "step": 96115
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2538341304123973e-05,
+      "loss": 0.2725,
+      "step": 96120
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2535871802746951e-05,
+      "loss": 0.2588,
+      "step": 96125
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2533402463208432e-05,
+      "loss": 0.2784,
+      "step": 96130
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2530933285540464e-05,
+      "loss": 0.2701,
+      "step": 96135
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2528464269775108e-05,
+      "loss": 0.2697,
+      "step": 96140
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2525995415944419e-05,
+      "loss": 0.2534,
+      "step": 96145
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2523526724080456e-05,
+      "loss": 0.272,
+      "step": 96150
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2521058194215284e-05,
+      "loss": 0.2502,
+      "step": 96155
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.251858982638095e-05,
+      "loss": 0.2615,
+      "step": 96160
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2516121620609495e-05,
+      "loss": 0.2699,
+      "step": 96165
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2513653576932977e-05,
+      "loss": 0.2711,
+      "step": 96170
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2511185695383443e-05,
+      "loss": 0.2523,
+      "step": 96175
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2508717975992932e-05,
+      "loss": 0.2591,
+      "step": 96180
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2506250418793475e-05,
+      "loss": 0.2689,
+      "step": 96185
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.250378302381713e-05,
+      "loss": 0.2366,
+      "step": 96190
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.250131579109593e-05,
+      "loss": 0.2754,
+      "step": 96195
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2498848720661899e-05,
+      "loss": 0.2485,
+      "step": 96200
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2496381812547086e-05,
+      "loss": 0.2825,
+      "step": 96205
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2493915066783506e-05,
+      "loss": 0.2758,
+      "step": 96210
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2491448483403206e-05,
+      "loss": 0.2512,
+      "step": 96215
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2488982062438196e-05,
+      "loss": 0.2478,
+      "step": 96220
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2486515803920518e-05,
+      "loss": 0.2593,
+      "step": 96225
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2484049707882187e-05,
+      "loss": 0.2603,
+      "step": 96230
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2481583774355218e-05,
+      "loss": 0.2691,
+      "step": 96235
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2479118003371628e-05,
+      "loss": 0.2614,
+      "step": 96240
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2476652394963448e-05,
+      "loss": 0.2582,
+      "step": 96245
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2474186949162681e-05,
+      "loss": 0.2583,
+      "step": 96250
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2471721666001334e-05,
+      "loss": 0.2465,
+      "step": 96255
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2469256545511435e-05,
+      "loss": 0.2863,
+      "step": 96260
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2466791587724972e-05,
+      "loss": 0.2601,
+      "step": 96265
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2464326792673969e-05,
+      "loss": 0.2444,
+      "step": 96270
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2461862160390423e-05,
+      "loss": 0.2784,
+      "step": 96275
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2459397690906321e-05,
+      "loss": 0.247,
+      "step": 96280
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2456933384253689e-05,
+      "loss": 0.2483,
+      "step": 96285
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2454469240464506e-05,
+      "loss": 0.255,
+      "step": 96290
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2452005259570776e-05,
+      "loss": 0.2449,
+      "step": 96295
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2449541441604478e-05,
+      "loss": 0.2525,
+      "step": 96300
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2447077786597621e-05,
+      "loss": 0.264,
+      "step": 96305
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.244461429458218e-05,
+      "loss": 0.2626,
+      "step": 96310
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2442150965590154e-05,
+      "loss": 0.2405,
+      "step": 96315
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2439687799653516e-05,
+      "loss": 0.2791,
+      "step": 96320
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2437224796804262e-05,
+      "loss": 0.2476,
+      "step": 96325
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2434761957074365e-05,
+      "loss": 0.2756,
+      "step": 96330
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2432299280495794e-05,
+      "loss": 0.2364,
+      "step": 96335
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2429836767100541e-05,
+      "loss": 0.2747,
+      "step": 96340
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2427374416920575e-05,
+      "loss": 0.2762,
+      "step": 96345
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.242491222998787e-05,
+      "loss": 0.2775,
+      "step": 96350
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.242245020633438e-05,
+      "loss": 0.241,
+      "step": 96355
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2419988345992095e-05,
+      "loss": 0.2515,
+      "step": 96360
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2417526648992972e-05,
+      "loss": 0.2556,
+      "step": 96365
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2415065115368962e-05,
+      "loss": 0.2698,
+      "step": 96370
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2412603745152038e-05,
+      "loss": 0.2618,
+      "step": 96375
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2410142538374167e-05,
+      "loss": 0.2633,
+      "step": 96380
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2407681495067303e-05,
+      "loss": 0.2549,
+      "step": 96385
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.240522061526339e-05,
+      "loss": 0.2463,
+      "step": 96390
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2402759898994382e-05,
+      "loss": 0.2453,
+      "step": 96395
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2400299346292244e-05,
+      "loss": 0.2395,
+      "step": 96400
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2397838957188912e-05,
+      "loss": 0.2588,
+      "step": 96405
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.239537873171634e-05,
+      "loss": 0.2691,
+      "step": 96410
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2392918669906456e-05,
+      "loss": 0.2514,
+      "step": 96415
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2390458771791227e-05,
+      "loss": 0.2651,
+      "step": 96420
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.238799903740257e-05,
+      "loss": 0.2565,
+      "step": 96425
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2385539466772434e-05,
+      "loss": 0.2823,
+      "step": 96430
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2383080059932764e-05,
+      "loss": 0.2646,
+      "step": 96435
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2380620816915484e-05,
+      "loss": 0.2515,
+      "step": 96440
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2378161737752525e-05,
+      "loss": 0.2457,
+      "step": 96445
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.237570282247581e-05,
+      "loss": 0.2659,
+      "step": 96450
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2373244071117285e-05,
+      "loss": 0.2197,
+      "step": 96455
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2370785483708863e-05,
+      "loss": 0.2739,
+      "step": 96460
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2368327060282466e-05,
+      "loss": 0.2702,
+      "step": 96465
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2365868800870011e-05,
+      "loss": 0.2533,
+      "step": 96470
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2363410705503433e-05,
+      "loss": 0.2643,
+      "step": 96475
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2360952774214629e-05,
+      "loss": 0.2736,
+      "step": 96480
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2358495007035534e-05,
+      "loss": 0.2449,
+      "step": 96485
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2356037403998042e-05,
+      "loss": 0.264,
+      "step": 96490
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.235357996513408e-05,
+      "loss": 0.246,
+      "step": 96495
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2351122690475544e-05,
+      "loss": 0.2577,
+      "step": 96500
+    },
+    {
+      "epoch": 3.4,
+      "eval_loss": 0.25399091839790344,
+      "eval_runtime": 10.5362,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 96500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2348665580054347e-05,
+      "loss": 0.2472,
+      "step": 96505
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.234620863390238e-05,
+      "loss": 0.2501,
+      "step": 96510
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2343751852051566e-05,
+      "loss": 0.2679,
+      "step": 96515
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2341295234533792e-05,
+      "loss": 0.2335,
+      "step": 96520
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2338838781380957e-05,
+      "loss": 0.2598,
+      "step": 96525
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2336382492624948e-05,
+      "loss": 0.2633,
+      "step": 96530
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2333926368297666e-05,
+      "loss": 0.2346,
+      "step": 96535
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2331470408431012e-05,
+      "loss": 0.2565,
+      "step": 96540
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2329014613056855e-05,
+      "loss": 0.277,
+      "step": 96545
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2326558982207103e-05,
+      "loss": 0.2606,
+      "step": 96550
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2324103515913632e-05,
+      "loss": 0.2622,
+      "step": 96555
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2321648214208322e-05,
+      "loss": 0.2664,
+      "step": 96560
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2319193077123048e-05,
+      "loss": 0.2556,
+      "step": 96565
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2316738104689702e-05,
+      "loss": 0.2709,
+      "step": 96570
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2314283296940151e-05,
+      "loss": 0.2433,
+      "step": 96575
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2311828653906274e-05,
+      "loss": 0.2474,
+      "step": 96580
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2309374175619932e-05,
+      "loss": 0.2536,
+      "step": 96585
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2306919862112998e-05,
+      "loss": 0.2532,
+      "step": 96590
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2304465713417357e-05,
+      "loss": 0.2661,
+      "step": 96595
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2302011729564861e-05,
+      "loss": 0.2787,
+      "step": 96600
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2299557910587366e-05,
+      "loss": 0.2458,
+      "step": 96605
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.229710425651675e-05,
+      "loss": 0.2451,
+      "step": 96610
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2294650767384866e-05,
+      "loss": 0.2549,
+      "step": 96615
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2292197443223567e-05,
+      "loss": 0.2621,
+      "step": 96620
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.22897442840647e-05,
+      "loss": 0.2517,
+      "step": 96625
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2287291289940137e-05,
+      "loss": 0.2579,
+      "step": 96630
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2284838460881718e-05,
+      "loss": 0.2619,
+      "step": 96635
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2282385796921284e-05,
+      "loss": 0.2806,
+      "step": 96640
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2279933298090695e-05,
+      "loss": 0.2447,
+      "step": 96645
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2277480964421784e-05,
+      "loss": 0.2726,
+      "step": 96650
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2275028795946405e-05,
+      "loss": 0.264,
+      "step": 96655
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2272576792696383e-05,
+      "loss": 0.2581,
+      "step": 96660
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2270124954703575e-05,
+      "loss": 0.247,
+      "step": 96665
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.22676732819998e-05,
+      "loss": 0.2573,
+      "step": 96670
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.22652217746169e-05,
+      "loss": 0.2721,
+      "step": 96675
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2262770432586692e-05,
+      "loss": 0.2413,
+      "step": 96680
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2260319255941028e-05,
+      "loss": 0.2581,
+      "step": 96685
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2257868244711717e-05,
+      "loss": 0.2703,
+      "step": 96690
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2255417398930586e-05,
+      "loss": 0.2916,
+      "step": 96695
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.225296671862947e-05,
+      "loss": 0.2592,
+      "step": 96700
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2250516203840167e-05,
+      "loss": 0.2426,
+      "step": 96705
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.224806585459452e-05,
+      "loss": 0.2486,
+      "step": 96710
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2245615670924334e-05,
+      "loss": 0.2327,
+      "step": 96715
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2243165652861416e-05,
+      "loss": 0.2592,
+      "step": 96720
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2240715800437591e-05,
+      "loss": 0.2651,
+      "step": 96725
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2238266113684666e-05,
+      "loss": 0.2399,
+      "step": 96730
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2235816592634441e-05,
+      "loss": 0.2465,
+      "step": 96735
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2233367237318718e-05,
+      "loss": 0.251,
+      "step": 96740
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2230918047769319e-05,
+      "loss": 0.2387,
+      "step": 96745
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2228469024018024e-05,
+      "loss": 0.24,
+      "step": 96750
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.222602016609665e-05,
+      "loss": 0.2775,
+      "step": 96755
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2223571474036976e-05,
+      "loss": 0.2399,
+      "step": 96760
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.222112294787082e-05,
+      "loss": 0.2629,
+      "step": 96765
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2218674587629959e-05,
+      "loss": 0.2504,
+      "step": 96770
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2216226393346173e-05,
+      "loss": 0.2398,
+      "step": 96775
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2213778365051273e-05,
+      "loss": 0.2523,
+      "step": 96780
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2211330502777036e-05,
+      "loss": 0.2766,
+      "step": 96785
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2208882806555243e-05,
+      "loss": 0.2539,
+      "step": 96790
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2206435276417669e-05,
+      "loss": 0.2499,
+      "step": 96795
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2203987912396108e-05,
+      "loss": 0.2583,
+      "step": 96800
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2201540714522336e-05,
+      "loss": 0.2588,
+      "step": 96805
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2199093682828111e-05,
+      "loss": 0.2623,
+      "step": 96810
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2196646817345217e-05,
+      "loss": 0.2326,
+      "step": 96815
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.219420011810544e-05,
+      "loss": 0.2569,
+      "step": 96820
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.219175358514053e-05,
+      "loss": 0.2758,
+      "step": 96825
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2189307218482263e-05,
+      "loss": 0.2601,
+      "step": 96830
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2186861018162387e-05,
+      "loss": 0.2643,
+      "step": 96835
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2184414984212689e-05,
+      "loss": 0.2625,
+      "step": 96840
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2181969116664912e-05,
+      "loss": 0.2553,
+      "step": 96845
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2179523415550822e-05,
+      "loss": 0.2407,
+      "step": 96850
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2177077880902162e-05,
+      "loss": 0.2398,
+      "step": 96855
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2174632512750706e-05,
+      "loss": 0.2668,
+      "step": 96860
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2172187311128181e-05,
+      "loss": 0.2671,
+      "step": 96865
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2169742276066362e-05,
+      "loss": 0.2285,
+      "step": 96870
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2167297407596972e-05,
+      "loss": 0.2851,
+      "step": 96875
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2164852705751782e-05,
+      "loss": 0.2614,
+      "step": 96880
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2162408170562518e-05,
+      "loss": 0.257,
+      "step": 96885
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2159963802060925e-05,
+      "loss": 0.2346,
+      "step": 96890
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.215751960027873e-05,
+      "loss": 0.249,
+      "step": 96895
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2155075565247686e-05,
+      "loss": 0.2442,
+      "step": 96900
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2152631696999525e-05,
+      "loss": 0.2477,
+      "step": 96905
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2150187995565965e-05,
+      "loss": 0.262,
+      "step": 96910
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2147744460978753e-05,
+      "loss": 0.2629,
+      "step": 96915
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2145301093269598e-05,
+      "loss": 0.2606,
+      "step": 96920
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.214285789247025e-05,
+      "loss": 0.2496,
+      "step": 96925
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.214041485861241e-05,
+      "loss": 0.2718,
+      "step": 96930
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2137971991727815e-05,
+      "loss": 0.245,
+      "step": 96935
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2135529291848177e-05,
+      "loss": 0.2643,
+      "step": 96940
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2133086759005217e-05,
+      "loss": 0.2774,
+      "step": 96945
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.213064439323063e-05,
+      "loss": 0.2857,
+      "step": 96950
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.212820219455616e-05,
+      "loss": 0.2753,
+      "step": 96955
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2125760163013497e-05,
+      "loss": 0.241,
+      "step": 96960
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2123318298634356e-05,
+      "loss": 0.2469,
+      "step": 96965
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2120876601450432e-05,
+      "loss": 0.2252,
+      "step": 96970
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.211843507149344e-05,
+      "loss": 0.2413,
+      "step": 96975
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2115993708795085e-05,
+      "loss": 0.2777,
+      "step": 96980
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2113552513387061e-05,
+      "loss": 0.2447,
+      "step": 96985
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2111111485301054e-05,
+      "loss": 0.259,
+      "step": 96990
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2108670624568786e-05,
+      "loss": 0.2624,
+      "step": 96995
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2106229931221929e-05,
+      "loss": 0.2697,
+      "step": 97000
+    },
+    {
+      "epoch": 3.41,
+      "eval_loss": 0.2534578740596771,
+      "eval_runtime": 10.5485,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 97000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.210378940529218e-05,
+      "loss": 0.2617,
+      "step": 97005
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2101349046811216e-05,
+      "loss": 0.2551,
+      "step": 97010
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2098908855810745e-05,
+      "loss": 0.2594,
+      "step": 97015
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.209646883232244e-05,
+      "loss": 0.2491,
+      "step": 97020
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2094028976377978e-05,
+      "loss": 0.2569,
+      "step": 97025
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.209158928800904e-05,
+      "loss": 0.2469,
+      "step": 97030
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2089149767247317e-05,
+      "loss": 0.2686,
+      "step": 97035
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2086710414124477e-05,
+      "loss": 0.2714,
+      "step": 97040
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2084271228672183e-05,
+      "loss": 0.261,
+      "step": 97045
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2081832210922125e-05,
+      "loss": 0.2628,
+      "step": 97050
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.207939336090596e-05,
+      "loss": 0.2624,
+      "step": 97055
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2076954678655361e-05,
+      "loss": 0.2594,
+      "step": 97060
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2074516164201977e-05,
+      "loss": 0.2724,
+      "step": 97065
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2072077817577493e-05,
+      "loss": 0.2579,
+      "step": 97070
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2069639638813557e-05,
+      "loss": 0.23,
+      "step": 97075
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.206720162794182e-05,
+      "loss": 0.2357,
+      "step": 97080
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2064763784993958e-05,
+      "loss": 0.248,
+      "step": 97085
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2062326110001603e-05,
+      "loss": 0.2556,
+      "step": 97090
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2059888602996428e-05,
+      "loss": 0.2812,
+      "step": 97095
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.205745126401007e-05,
+      "loss": 0.2516,
+      "step": 97100
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2055014093074168e-05,
+      "loss": 0.2561,
+      "step": 97105
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2052577090220387e-05,
+      "loss": 0.2539,
+      "step": 97110
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2050140255480361e-05,
+      "loss": 0.2494,
+      "step": 97115
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2047703588885728e-05,
+      "loss": 0.2374,
+      "step": 97120
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2045267090468117e-05,
+      "loss": 0.245,
+      "step": 97125
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2042830760259189e-05,
+      "loss": 0.2559,
+      "step": 97130
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2040394598290553e-05,
+      "loss": 0.2556,
+      "step": 97135
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.203795860459386e-05,
+      "loss": 0.2499,
+      "step": 97140
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2035522779200725e-05,
+      "loss": 0.2616,
+      "step": 97145
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.203308712214279e-05,
+      "loss": 0.2649,
+      "step": 97150
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2030651633451676e-05,
+      "loss": 0.2547,
+      "step": 97155
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2028216313158994e-05,
+      "loss": 0.2481,
+      "step": 97160
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2025781161296382e-05,
+      "loss": 0.2559,
+      "step": 97165
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2023346177895453e-05,
+      "loss": 0.2568,
+      "step": 97170
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2020911362987822e-05,
+      "loss": 0.2789,
+      "step": 97175
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2018476716605092e-05,
+      "loss": 0.2556,
+      "step": 97180
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2016042238778896e-05,
+      "loss": 0.2463,
+      "step": 97185
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2013607929540826e-05,
+      "loss": 0.2639,
+      "step": 97190
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2011173788922508e-05,
+      "loss": 0.259,
+      "step": 97195
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.200873981695553e-05,
+      "loss": 0.2368,
+      "step": 97200
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2006306013671515e-05,
+      "loss": 0.2639,
+      "step": 97205
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2003872379102049e-05,
+      "loss": 0.2753,
+      "step": 97210
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2001438913278739e-05,
+      "loss": 0.2755,
+      "step": 97215
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1999005616233166e-05,
+      "loss": 0.2661,
+      "step": 97220
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.199657248799695e-05,
+      "loss": 0.2821,
+      "step": 97225
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1994139528601667e-05,
+      "loss": 0.2348,
+      "step": 97230
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1991706738078912e-05,
+      "loss": 0.2759,
+      "step": 97235
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1989274116460264e-05,
+      "loss": 0.2274,
+      "step": 97240
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1986841663777326e-05,
+      "loss": 0.2486,
+      "step": 97245
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1984409380061665e-05,
+      "loss": 0.2528,
+      "step": 97250
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1981977265344868e-05,
+      "loss": 0.2428,
+      "step": 97255
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1979545319658527e-05,
+      "loss": 0.2572,
+      "step": 97260
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.197711354303421e-05,
+      "loss": 0.2478,
+      "step": 97265
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1974681935503493e-05,
+      "loss": 0.2638,
+      "step": 97270
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1972250497097933e-05,
+      "loss": 0.2489,
+      "step": 97275
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.196981922784913e-05,
+      "loss": 0.2545,
+      "step": 97280
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1967388127788634e-05,
+      "loss": 0.2686,
+      "step": 97285
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1964957196948012e-05,
+      "loss": 0.2514,
+      "step": 97290
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1962526435358825e-05,
+      "loss": 0.2623,
+      "step": 97295
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1960095843052647e-05,
+      "loss": 0.2366,
+      "step": 97300
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1957665420061023e-05,
+      "loss": 0.2658,
+      "step": 97305
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1955235166415526e-05,
+      "loss": 0.2777,
+      "step": 97310
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1952805082147697e-05,
+      "loss": 0.2421,
+      "step": 97315
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1950375167289102e-05,
+      "loss": 0.2495,
+      "step": 97320
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1947945421871288e-05,
+      "loss": 0.237,
+      "step": 97325
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1945515845925798e-05,
+      "loss": 0.2598,
+      "step": 97330
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1943086439484174e-05,
+      "loss": 0.2453,
+      "step": 97335
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1940657202577977e-05,
+      "loss": 0.271,
+      "step": 97340
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1938228135238741e-05,
+      "loss": 0.2707,
+      "step": 97345
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1935799237498002e-05,
+      "loss": 0.2652,
+      "step": 97350
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1933370509387293e-05,
+      "loss": 0.2505,
+      "step": 97355
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.193142764905355e-05,
+      "loss": 0.2565,
+      "step": 97360
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1928999226356375e-05,
+      "loss": 0.2557,
+      "step": 97365
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.192657097337754e-05,
+      "loss": 0.2775,
+      "step": 97370
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1924142890148566e-05,
+      "loss": 0.2629,
+      "step": 97375
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1921714976700979e-05,
+      "loss": 0.2723,
+      "step": 97380
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1919287233066295e-05,
+      "loss": 0.2582,
+      "step": 97385
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1916859659276055e-05,
+      "loss": 0.2901,
+      "step": 97390
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.191443225536176e-05,
+      "loss": 0.2493,
+      "step": 97395
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.191200502135495e-05,
+      "loss": 0.24,
+      "step": 97400
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1909577957287119e-05,
+      "loss": 0.2765,
+      "step": 97405
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1907151063189797e-05,
+      "loss": 0.2789,
+      "step": 97410
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1904724339094494e-05,
+      "loss": 0.2526,
+      "step": 97415
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1902297785032715e-05,
+      "loss": 0.2481,
+      "step": 97420
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1899871401035958e-05,
+      "loss": 0.2529,
+      "step": 97425
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1897445187135747e-05,
+      "loss": 0.2799,
+      "step": 97430
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1895019143363576e-05,
+      "loss": 0.2723,
+      "step": 97435
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1892593269750935e-05,
+      "loss": 0.2673,
+      "step": 97440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.189016756632934e-05,
+      "loss": 0.2484,
+      "step": 97445
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.188774203313027e-05,
+      "loss": 0.2577,
+      "step": 97450
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.188531667018524e-05,
+      "loss": 0.2783,
+      "step": 97455
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.188289147752572e-05,
+      "loss": 0.2529,
+      "step": 97460
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.188046645518322e-05,
+      "loss": 0.2641,
+      "step": 97465
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1878041603189213e-05,
+      "loss": 0.2413,
+      "step": 97470
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1875616921575191e-05,
+      "loss": 0.2463,
+      "step": 97475
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1873192410372628e-05,
+      "loss": 0.2614,
+      "step": 97480
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1870768069613014e-05,
+      "loss": 0.2712,
+      "step": 97485
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1868343899327827e-05,
+      "loss": 0.2498,
+      "step": 97490
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1865919899548542e-05,
+      "loss": 0.2691,
+      "step": 97495
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1863496070306618e-05,
+      "loss": 0.2492,
+      "step": 97500
+    },
+    {
+      "epoch": 3.43,
+      "eval_loss": 0.25291621685028076,
+      "eval_runtime": 10.5567,
+      "eval_samples_per_second": 9.473,
+      "eval_steps_per_second": 9.473,
+      "step": 97500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1861072411633542e-05,
+      "loss": 0.2492,
+      "step": 97505
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1858648923560792e-05,
+      "loss": 0.2829,
+      "step": 97510
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1856225606119825e-05,
+      "loss": 0.2566,
+      "step": 97515
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1853802459342098e-05,
+      "loss": 0.2599,
+      "step": 97520
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.185137948325909e-05,
+      "loss": 0.259,
+      "step": 97525
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1848956677902254e-05,
+      "loss": 0.2482,
+      "step": 97530
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.184653404330305e-05,
+      "loss": 0.2563,
+      "step": 97535
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1844111579492923e-05,
+      "loss": 0.2484,
+      "step": 97540
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1841689286503347e-05,
+      "loss": 0.2535,
+      "step": 97545
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1839267164365764e-05,
+      "loss": 0.2615,
+      "step": 97550
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1836845213111614e-05,
+      "loss": 0.26,
+      "step": 97555
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1834423432772354e-05,
+      "loss": 0.2566,
+      "step": 97560
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1832001823379438e-05,
+      "loss": 0.2528,
+      "step": 97565
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.18295803849643e-05,
+      "loss": 0.264,
+      "step": 97570
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1827159117558373e-05,
+      "loss": 0.2678,
+      "step": 97575
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1824738021193113e-05,
+      "loss": 0.2706,
+      "step": 97580
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1822317095899946e-05,
+      "loss": 0.2569,
+      "step": 97585
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1819896341710307e-05,
+      "loss": 0.2325,
+      "step": 97590
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1817475758655618e-05,
+      "loss": 0.261,
+      "step": 97595
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1815055346767332e-05,
+      "loss": 0.2611,
+      "step": 97600
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.181263510607686e-05,
+      "loss": 0.2512,
+      "step": 97605
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1810215036615622e-05,
+      "loss": 0.2693,
+      "step": 97610
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1807795138415048e-05,
+      "loss": 0.2683,
+      "step": 97615
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.180537541150657e-05,
+      "loss": 0.2503,
+      "step": 97620
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1802955855921596e-05,
+      "loss": 0.2489,
+      "step": 97625
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.1800536471691545e-05,
+      "loss": 0.2223,
+      "step": 97630
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1798117258847818e-05,
+      "loss": 0.2399,
+      "step": 97635
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1795698217421849e-05,
+      "loss": 0.2515,
+      "step": 97640
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1793279347445036e-05,
+      "loss": 0.2619,
+      "step": 97645
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1790860648948787e-05,
+      "loss": 0.269,
+      "step": 97650
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1788442121964496e-05,
+      "loss": 0.2618,
+      "step": 97655
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.178602376652359e-05,
+      "loss": 0.2519,
+      "step": 97660
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1783605582657444e-05,
+      "loss": 0.2638,
+      "step": 97665
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1781187570397483e-05,
+      "loss": 0.2562,
+      "step": 97670
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.177876972977508e-05,
+      "loss": 0.2606,
+      "step": 97675
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1776352060821646e-05,
+      "loss": 0.254,
+      "step": 97680
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1773934563568568e-05,
+      "loss": 0.2534,
+      "step": 97685
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1771517238047223e-05,
+      "loss": 0.2525,
+      "step": 97690
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1769100084289017e-05,
+      "loss": 0.2712,
+      "step": 97695
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1766683102325326e-05,
+      "loss": 0.2546,
+      "step": 97700
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1764266292187538e-05,
+      "loss": 0.2382,
+      "step": 97705
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1761849653907017e-05,
+      "loss": 0.2577,
+      "step": 97710
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1759433187515163e-05,
+      "loss": 0.2438,
+      "step": 97715
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1757016893043335e-05,
+      "loss": 0.2657,
+      "step": 97720
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1754600770522925e-05,
+      "loss": 0.2439,
+      "step": 97725
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1752184819985285e-05,
+      "loss": 0.2766,
+      "step": 97730
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1749769041461803e-05,
+      "loss": 0.2521,
+      "step": 97735
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1747353434983841e-05,
+      "loss": 0.2627,
+      "step": 97740
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1744938000582756e-05,
+      "loss": 0.252,
+      "step": 97745
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.174252273828991e-05,
+      "loss": 0.2612,
+      "step": 97750
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1740107648136677e-05,
+      "loss": 0.2575,
+      "step": 97755
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1737692730154409e-05,
+      "loss": 0.2804,
+      "step": 97760
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.173527798437446e-05,
+      "loss": 0.2669,
+      "step": 97765
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1732863410828174e-05,
+      "loss": 0.2583,
+      "step": 97770
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1730449009546915e-05,
+      "loss": 0.2429,
+      "step": 97775
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1728034780562038e-05,
+      "loss": 0.2754,
+      "step": 97780
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1725620723904877e-05,
+      "loss": 0.2643,
+      "step": 97785
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1723206839606787e-05,
+      "loss": 0.2565,
+      "step": 97790
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.172079312769911e-05,
+      "loss": 0.2546,
+      "step": 97795
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1718379588213182e-05,
+      "loss": 0.2491,
+      "step": 97800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1715966221180333e-05,
+      "loss": 0.2714,
+      "step": 97805
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1713553026631918e-05,
+      "loss": 0.2475,
+      "step": 97810
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1711140004599262e-05,
+      "loss": 0.2586,
+      "step": 97815
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1708727155113692e-05,
+      "loss": 0.2523,
+      "step": 97820
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1706314478206531e-05,
+      "loss": 0.2639,
+      "step": 97825
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1703901973909129e-05,
+      "loss": 0.2606,
+      "step": 97830
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1701489642252786e-05,
+      "loss": 0.2492,
+      "step": 97835
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1699077483268844e-05,
+      "loss": 0.2723,
+      "step": 97840
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1696665496988606e-05,
+      "loss": 0.2448,
+      "step": 97845
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1694253683443413e-05,
+      "loss": 0.2654,
+      "step": 97850
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.169184204266456e-05,
+      "loss": 0.2614,
+      "step": 97855
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1689430574683375e-05,
+      "loss": 0.2569,
+      "step": 97860
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1687019279531147e-05,
+      "loss": 0.2485,
+      "step": 97865
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1684608157239214e-05,
+      "loss": 0.2755,
+      "step": 97870
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1682197207838866e-05,
+      "loss": 0.2449,
+      "step": 97875
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1679786431361401e-05,
+      "loss": 0.2429,
+      "step": 97880
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1677375827838142e-05,
+      "loss": 0.2594,
+      "step": 97885
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1674965397300367e-05,
+      "loss": 0.2398,
+      "step": 97890
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1672555139779392e-05,
+      "loss": 0.2532,
+      "step": 97895
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.16701450553065e-05,
+      "loss": 0.2591,
+      "step": 97900
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1667735143912998e-05,
+      "loss": 0.2373,
+      "step": 97905
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.1665325405630167e-05,
+      "loss": 0.2712,
+      "step": 97910
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.16629158404893e-05,
+      "loss": 0.2522,
+      "step": 97915
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1660506448521672e-05,
+      "loss": 0.2578,
+      "step": 97920
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1658097229758586e-05,
+      "loss": 0.2519,
+      "step": 97925
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1655688184231314e-05,
+      "loss": 0.2566,
+      "step": 97930
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1653279311971135e-05,
+      "loss": 0.2403,
+      "step": 97935
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1650870613009321e-05,
+      "loss": 0.2754,
+      "step": 97940
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1648462087377154e-05,
+      "loss": 0.2559,
+      "step": 97945
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1646053735105919e-05,
+      "loss": 0.2337,
+      "step": 97950
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1643645556226876e-05,
+      "loss": 0.2396,
+      "step": 97955
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.164123755077128e-05,
+      "loss": 0.2341,
+      "step": 97960
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1638829718770421e-05,
+      "loss": 0.241,
+      "step": 97965
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1636422060255555e-05,
+      "loss": 0.2725,
+      "step": 97970
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1634014575257943e-05,
+      "loss": 0.2507,
+      "step": 97975
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.163160726380883e-05,
+      "loss": 0.2498,
+      "step": 97980
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1629200125939498e-05,
+      "loss": 0.2688,
+      "step": 97985
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1626793161681191e-05,
+      "loss": 0.2578,
+      "step": 97990
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.162438637106515e-05,
+      "loss": 0.2772,
+      "step": 97995
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.162197975412264e-05,
+      "loss": 0.2662,
+      "step": 98000
+    },
+    {
+      "epoch": 3.45,
+      "eval_loss": 0.25310197472572327,
+      "eval_runtime": 10.5476,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 98000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1619573310884913e-05,
+      "loss": 0.2549,
+      "step": 98005
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1617167041383208e-05,
+      "loss": 0.2417,
+      "step": 98010
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1614760945648765e-05,
+      "loss": 0.2693,
+      "step": 98015
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1612355023712834e-05,
+      "loss": 0.2729,
+      "step": 98020
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1609949275606651e-05,
+      "loss": 0.2596,
+      "step": 98025
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1607543701361455e-05,
+      "loss": 0.2607,
+      "step": 98030
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1605138301008466e-05,
+      "loss": 0.2725,
+      "step": 98035
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1602733074578939e-05,
+      "loss": 0.263,
+      "step": 98040
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1600328022104095e-05,
+      "loss": 0.2431,
+      "step": 98045
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1597923143615152e-05,
+      "loss": 0.2503,
+      "step": 98050
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1595518439143344e-05,
+      "loss": 0.2678,
+      "step": 98055
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1593113908719902e-05,
+      "loss": 0.2826,
+      "step": 98060
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1590709552376042e-05,
+      "loss": 0.232,
+      "step": 98065
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1588305370142982e-05,
+      "loss": 0.2463,
+      "step": 98070
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1585901362051929e-05,
+      "loss": 0.3027,
+      "step": 98075
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1583497528134116e-05,
+      "loss": 0.2829,
+      "step": 98080
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1581093868420745e-05,
+      "loss": 0.2723,
+      "step": 98085
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1578690382943027e-05,
+      "loss": 0.2583,
+      "step": 98090
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1576287071732158e-05,
+      "loss": 0.2432,
+      "step": 98095
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1573883934819369e-05,
+      "loss": 0.2546,
+      "step": 98100
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1571480972235834e-05,
+      "loss": 0.2613,
+      "step": 98105
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1569078184012782e-05,
+      "loss": 0.2659,
+      "step": 98110
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.156667557018139e-05,
+      "loss": 0.2676,
+      "step": 98115
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1564273130772868e-05,
+      "loss": 0.2521,
+      "step": 98120
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1561870865818408e-05,
+      "loss": 0.2378,
+      "step": 98125
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1559468775349197e-05,
+      "loss": 0.2688,
+      "step": 98130
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.155706685939642e-05,
+      "loss": 0.2599,
+      "step": 98135
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1554665117991275e-05,
+      "loss": 0.2635,
+      "step": 98140
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1552263551164947e-05,
+      "loss": 0.245,
+      "step": 98145
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1549862158948604e-05,
+      "loss": 0.2681,
+      "step": 98150
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1547460941373445e-05,
+      "loss": 0.2921,
+      "step": 98155
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1545059898470634e-05,
+      "loss": 0.2582,
+      "step": 98160
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1542659030271358e-05,
+      "loss": 0.2489,
+      "step": 98165
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1540258336806778e-05,
+      "loss": 0.2556,
+      "step": 98170
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1537857818108082e-05,
+      "loss": 0.2798,
+      "step": 98175
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1535457474206433e-05,
+      "loss": 0.2701,
+      "step": 98180
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1533057305132995e-05,
+      "loss": 0.2527,
+      "step": 98185
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1530657310918922e-05,
+      "loss": 0.2337,
+      "step": 98190
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1528257491595395e-05,
+      "loss": 0.24,
+      "step": 98195
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1525857847193568e-05,
+      "loss": 0.2533,
+      "step": 98200
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1523458377744598e-05,
+      "loss": 0.2501,
+      "step": 98205
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1521059083279629e-05,
+      "loss": 0.2454,
+      "step": 98210
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1518659963829826e-05,
+      "loss": 0.2358,
+      "step": 98215
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1516261019426345e-05,
+      "loss": 0.2607,
+      "step": 98220
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1513862250100333e-05,
+      "loss": 0.2938,
+      "step": 98225
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1511463655882917e-05,
+      "loss": 0.2692,
+      "step": 98230
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.150906523680527e-05,
+      "loss": 0.2624,
+      "step": 98235
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1506666992898519e-05,
+      "loss": 0.2545,
+      "step": 98240
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1504268924193803e-05,
+      "loss": 0.2491,
+      "step": 98245
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1501871030722256e-05,
+      "loss": 0.2624,
+      "step": 98250
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1499473312515025e-05,
+      "loss": 0.2537,
+      "step": 98255
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1497075769603233e-05,
+      "loss": 0.2647,
+      "step": 98260
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1494678402018008e-05,
+      "loss": 0.2603,
+      "step": 98265
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.149228120979049e-05,
+      "loss": 0.2296,
+      "step": 98270
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.148988419295179e-05,
+      "loss": 0.274,
+      "step": 98275
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1487487351533051e-05,
+      "loss": 0.2443,
+      "step": 98280
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1485090685565375e-05,
+      "loss": 0.2331,
+      "step": 98285
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1482694195079897e-05,
+      "loss": 0.2856,
+      "step": 98290
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1480297880107726e-05,
+      "loss": 0.2525,
+      "step": 98295
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.147790174067998e-05,
+      "loss": 0.2415,
+      "step": 98300
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1475505776827757e-05,
+      "loss": 0.2257,
+      "step": 98305
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.147310998858219e-05,
+      "loss": 0.2586,
+      "step": 98310
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1470714375974373e-05,
+      "loss": 0.2403,
+      "step": 98315
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1468318939035405e-05,
+      "loss": 0.2549,
+      "step": 98320
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.146592367779641e-05,
+      "loss": 0.2518,
+      "step": 98325
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1463528592288464e-05,
+      "loss": 0.2237,
+      "step": 98330
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1461133682542687e-05,
+      "loss": 0.2486,
+      "step": 98335
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1458738948590166e-05,
+      "loss": 0.2736,
+      "step": 98340
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.145634439046199e-05,
+      "loss": 0.2724,
+      "step": 98345
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1453950008189266e-05,
+      "loss": 0.2465,
+      "step": 98350
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1451555801803071e-05,
+      "loss": 0.2525,
+      "step": 98355
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1449161771334496e-05,
+      "loss": 0.225,
+      "step": 98360
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1446767916814618e-05,
+      "loss": 0.2672,
+      "step": 98365
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1444374238274533e-05,
+      "loss": 0.2658,
+      "step": 98370
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1441980735745317e-05,
+      "loss": 0.2542,
+      "step": 98375
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1439587409258035e-05,
+      "loss": 0.2588,
+      "step": 98380
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1437194258843775e-05,
+      "loss": 0.28,
+      "step": 98385
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1434801284533619e-05,
+      "loss": 0.2688,
+      "step": 98390
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1432408486358628e-05,
+      "loss": 0.2388,
+      "step": 98395
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1430015864349863e-05,
+      "loss": 0.2906,
+      "step": 98400
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1427623418538409e-05,
+      "loss": 0.2705,
+      "step": 98405
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1425231148955318e-05,
+      "loss": 0.255,
+      "step": 98410
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1422839055631654e-05,
+      "loss": 0.2498,
+      "step": 98415
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.142044713859847e-05,
+      "loss": 0.2692,
+      "step": 98420
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1418055397886837e-05,
+      "loss": 0.239,
+      "step": 98425
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1415663833527807e-05,
+      "loss": 0.2592,
+      "step": 98430
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.141327244555242e-05,
+      "loss": 0.2535,
+      "step": 98435
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1410881233991736e-05,
+      "loss": 0.2289,
+      "step": 98440
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1408490198876812e-05,
+      "loss": 0.2741,
+      "step": 98445
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1406099340238686e-05,
+      "loss": 0.2558,
+      "step": 98450
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1403708658108404e-05,
+      "loss": 0.2714,
+      "step": 98455
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1401318152516991e-05,
+      "loss": 0.2539,
+      "step": 98460
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1398927823495512e-05,
+      "loss": 0.2633,
+      "step": 98465
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.139653767107499e-05,
+      "loss": 0.2585,
+      "step": 98470
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1394147695286461e-05,
+      "loss": 0.2629,
+      "step": 98475
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.139175789616095e-05,
+      "loss": 0.2507,
+      "step": 98480
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1389368273729504e-05,
+      "loss": 0.265,
+      "step": 98485
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1386978828023132e-05,
+      "loss": 0.249,
+      "step": 98490
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1384589559072866e-05,
+      "loss": 0.2654,
+      "step": 98495
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1382200466909745e-05,
+      "loss": 0.2584,
+      "step": 98500
+    },
+    {
+      "epoch": 3.47,
+      "eval_loss": 0.2523847818374634,
+      "eval_runtime": 10.5582,
+      "eval_samples_per_second": 9.471,
+      "eval_steps_per_second": 9.471,
+      "step": 98500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1379811551564773e-05,
+      "loss": 0.2517,
+      "step": 98505
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1377422813068975e-05,
+      "loss": 0.2734,
+      "step": 98510
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1375034251453353e-05,
+      "loss": 0.2441,
+      "step": 98515
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.137264586674894e-05,
+      "loss": 0.2562,
+      "step": 98520
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1370257658986743e-05,
+      "loss": 0.2511,
+      "step": 98525
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1367869628197766e-05,
+      "loss": 0.2513,
+      "step": 98530
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1365481774413009e-05,
+      "loss": 0.2531,
+      "step": 98535
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1363094097663495e-05,
+      "loss": 0.2581,
+      "step": 98540
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1360706597980208e-05,
+      "loss": 0.2849,
+      "step": 98545
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1358319275394167e-05,
+      "loss": 0.2515,
+      "step": 98550
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1355932129936348e-05,
+      "loss": 0.2516,
+      "step": 98555
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1353545161637771e-05,
+      "loss": 0.2606,
+      "step": 98560
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1351158370529414e-05,
+      "loss": 0.2511,
+      "step": 98565
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1348771756642273e-05,
+      "loss": 0.2594,
+      "step": 98570
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1346385320007324e-05,
+      "loss": 0.2801,
+      "step": 98575
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1343999060655571e-05,
+      "loss": 0.2821,
+      "step": 98580
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1341612978617994e-05,
+      "loss": 0.2317,
+      "step": 98585
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.133922707392557e-05,
+      "loss": 0.2723,
+      "step": 98590
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.133684134660927e-05,
+      "loss": 0.2516,
+      "step": 98595
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1334455796700081e-05,
+      "loss": 0.2523,
+      "step": 98600
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1332070424228985e-05,
+      "loss": 0.2486,
+      "step": 98605
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1329685229226941e-05,
+      "loss": 0.2606,
+      "step": 98610
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.132730021172493e-05,
+      "loss": 0.256,
+      "step": 98615
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.132491537175392e-05,
+      "loss": 0.2728,
+      "step": 98620
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.132253070934487e-05,
+      "loss": 0.2528,
+      "step": 98625
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1320146224528735e-05,
+      "loss": 0.2654,
+      "step": 98630
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1317761917336494e-05,
+      "loss": 0.2539,
+      "step": 98635
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.13153777877991e-05,
+      "loss": 0.2663,
+      "step": 98640
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1312993835947506e-05,
+      "loss": 0.2685,
+      "step": 98645
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1310610061812657e-05,
+      "loss": 0.2615,
+      "step": 98650
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1308226465425516e-05,
+      "loss": 0.2569,
+      "step": 98655
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1305843046817041e-05,
+      "loss": 0.2481,
+      "step": 98660
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.130345980601817e-05,
+      "loss": 0.2643,
+      "step": 98665
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1301076743059833e-05,
+      "loss": 0.2595,
+      "step": 98670
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1298693857973e-05,
+      "loss": 0.2522,
+      "step": 98675
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1296311150788596e-05,
+      "loss": 0.2526,
+      "step": 98680
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1293928621537561e-05,
+      "loss": 0.2947,
+      "step": 98685
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.129154627025082e-05,
+      "loss": 0.2512,
+      "step": 98690
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1289164096959326e-05,
+      "loss": 0.2723,
+      "step": 98695
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1286782101694002e-05,
+      "loss": 0.2568,
+      "step": 98700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1284400284485766e-05,
+      "loss": 0.2901,
+      "step": 98705
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.128201864536556e-05,
+      "loss": 0.2345,
+      "step": 98710
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1279637184364294e-05,
+      "loss": 0.2344,
+      "step": 98715
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1277255901512907e-05,
+      "loss": 0.2376,
+      "step": 98720
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1274874796842296e-05,
+      "loss": 0.2682,
+      "step": 98725
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1272493870383404e-05,
+      "loss": 0.2536,
+      "step": 98730
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1270113122167131e-05,
+      "loss": 0.2728,
+      "step": 98735
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1267732552224388e-05,
+      "loss": 0.2463,
+      "step": 98740
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1265352160586082e-05,
+      "loss": 0.2672,
+      "step": 98745
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1262971947283135e-05,
+      "loss": 0.247,
+      "step": 98750
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1260591912346444e-05,
+      "loss": 0.2392,
+      "step": 98755
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1258212055806904e-05,
+      "loss": 0.2306,
+      "step": 98760
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1255832377695433e-05,
+      "loss": 0.2488,
+      "step": 98765
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1253452878042912e-05,
+      "loss": 0.2726,
+      "step": 98770
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1251073556880254e-05,
+      "loss": 0.2537,
+      "step": 98775
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1248694414238346e-05,
+      "loss": 0.2447,
+      "step": 98780
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1246315450148067e-05,
+      "loss": 0.2559,
+      "step": 98785
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.124393666464033e-05,
+      "loss": 0.2508,
+      "step": 98790
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.124155805774601e-05,
+      "loss": 0.2415,
+      "step": 98795
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.123917962949599e-05,
+      "loss": 0.2632,
+      "step": 98800
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1236801379921145e-05,
+      "loss": 0.2436,
+      "step": 98805
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1234423309052371e-05,
+      "loss": 0.2469,
+      "step": 98810
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1232045416920542e-05,
+      "loss": 0.2319,
+      "step": 98815
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1229667703556521e-05,
+      "loss": 0.2872,
+      "step": 98820
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1227290168991187e-05,
+      "loss": 0.2447,
+      "step": 98825
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1224912813255422e-05,
+      "loss": 0.2335,
+      "step": 98830
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1222535636380089e-05,
+      "loss": 0.2465,
+      "step": 98835
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1220158638396042e-05,
+      "loss": 0.2574,
+      "step": 98840
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1217781819334161e-05,
+      "loss": 0.2807,
+      "step": 98845
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.12154051792253e-05,
+      "loss": 0.2698,
+      "step": 98850
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1213028718100319e-05,
+      "loss": 0.2538,
+      "step": 98855
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1210652435990065e-05,
+      "loss": 0.2862,
+      "step": 98860
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.120827633292541e-05,
+      "loss": 0.2738,
+      "step": 98865
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.12059004089372e-05,
+      "loss": 0.2577,
+      "step": 98870
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1203524664056273e-05,
+      "loss": 0.2518,
+      "step": 98875
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1201149098313485e-05,
+      "loss": 0.2623,
+      "step": 98880
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1198773711739691e-05,
+      "loss": 0.2556,
+      "step": 98885
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1196398504365727e-05,
+      "loss": 0.2792,
+      "step": 98890
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1194023476222429e-05,
+      "loss": 0.2372,
+      "step": 98895
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1191648627340628e-05,
+      "loss": 0.2521,
+      "step": 98900
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1189273957751181e-05,
+      "loss": 0.2586,
+      "step": 98905
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.118689946748491e-05,
+      "loss": 0.2539,
+      "step": 98910
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1184525156572646e-05,
+      "loss": 0.2706,
+      "step": 98915
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1182151025045206e-05,
+      "loss": 0.2535,
+      "step": 98920
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1179777072933443e-05,
+      "loss": 0.2416,
+      "step": 98925
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1177403300268152e-05,
+      "loss": 0.2593,
+      "step": 98930
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1175029707080184e-05,
+      "loss": 0.259,
+      "step": 98935
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1172656293400333e-05,
+      "loss": 0.2734,
+      "step": 98940
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1170283059259434e-05,
+      "loss": 0.2653,
+      "step": 98945
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1167910004688298e-05,
+      "loss": 0.2646,
+      "step": 98950
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1165537129717727e-05,
+      "loss": 0.2462,
+      "step": 98955
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1163164434378545e-05,
+      "loss": 0.2366,
+      "step": 98960
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1160791918701557e-05,
+      "loss": 0.2477,
+      "step": 98965
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1158419582717563e-05,
+      "loss": 0.2542,
+      "step": 98970
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1156047426457363e-05,
+      "loss": 0.2579,
+      "step": 98975
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.115367544995177e-05,
+      "loss": 0.2721,
+      "step": 98980
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1151303653231568e-05,
+      "loss": 0.2704,
+      "step": 98985
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1148932036327573e-05,
+      "loss": 0.2792,
+      "step": 98990
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1146560599270553e-05,
+      "loss": 0.252,
+      "step": 98995
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1144189342091328e-05,
+      "loss": 0.2534,
+      "step": 99000
+    },
+    {
+      "epoch": 3.48,
+      "eval_loss": 0.25207310914993286,
+      "eval_runtime": 10.551,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 99000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.114181826482067e-05,
+      "loss": 0.2452,
+      "step": 99005
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.113944736748937e-05,
+      "loss": 0.2669,
+      "step": 99010
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.11370766501282e-05,
+      "loss": 0.2615,
+      "step": 99015
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1134706112767965e-05,
+      "loss": 0.2689,
+      "step": 99020
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1132335755439435e-05,
+      "loss": 0.2541,
+      "step": 99025
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1129965578173385e-05,
+      "loss": 0.27,
+      "step": 99030
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.112759558100058e-05,
+      "loss": 0.2776,
+      "step": 99035
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1125225763951805e-05,
+      "loss": 0.2625,
+      "step": 99040
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.112285612705784e-05,
+      "loss": 0.2843,
+      "step": 99045
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1120486670349442e-05,
+      "loss": 0.2572,
+      "step": 99050
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1118117393857369e-05,
+      "loss": 0.2688,
+      "step": 99055
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1115748297612403e-05,
+      "loss": 0.2671,
+      "step": 99060
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1113379381645294e-05,
+      "loss": 0.2572,
+      "step": 99065
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1111010645986794e-05,
+      "loss": 0.2505,
+      "step": 99070
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1108642090667676e-05,
+      "loss": 0.2631,
+      "step": 99075
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1106273715718685e-05,
+      "loss": 0.2634,
+      "step": 99080
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1103905521170574e-05,
+      "loss": 0.2694,
+      "step": 99085
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1101537507054083e-05,
+      "loss": 0.2526,
+      "step": 99090
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1099169673399967e-05,
+      "loss": 0.2336,
+      "step": 99095
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1096802020238984e-05,
+      "loss": 0.2396,
+      "step": 99100
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1094434547601862e-05,
+      "loss": 0.2503,
+      "step": 99105
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1092067255519333e-05,
+      "loss": 0.2572,
+      "step": 99110
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1089700144022155e-05,
+      "loss": 0.2532,
+      "step": 99115
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1087333213141055e-05,
+      "loss": 0.2654,
+      "step": 99120
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1084966462906765e-05,
+      "loss": 0.2429,
+      "step": 99125
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1082599893350005e-05,
+      "loss": 0.248,
+      "step": 99130
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1080233504501519e-05,
+      "loss": 0.2411,
+      "step": 99135
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.107786729639203e-05,
+      "loss": 0.2428,
+      "step": 99140
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1075501269052252e-05,
+      "loss": 0.2668,
+      "step": 99145
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.107313542251292e-05,
+      "loss": 0.2729,
+      "step": 99150
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1070769756804739e-05,
+      "loss": 0.2708,
+      "step": 99155
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.106840427195844e-05,
+      "loss": 0.2568,
+      "step": 99160
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1066038968004734e-05,
+      "loss": 0.2346,
+      "step": 99165
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1063673844974318e-05,
+      "loss": 0.2624,
+      "step": 99170
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1061308902897922e-05,
+      "loss": 0.2456,
+      "step": 99175
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1058944141806243e-05,
+      "loss": 0.2458,
+      "step": 99180
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1056579561729976e-05,
+      "loss": 0.27,
+      "step": 99185
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1054215162699846e-05,
+      "loss": 0.2613,
+      "step": 99190
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1051850944746543e-05,
+      "loss": 0.2632,
+      "step": 99195
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1049486907900752e-05,
+      "loss": 0.2627,
+      "step": 99200
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.104712305219319e-05,
+      "loss": 0.2485,
+      "step": 99205
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.104475937765453e-05,
+      "loss": 0.2587,
+      "step": 99210
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1042395884315481e-05,
+      "loss": 0.2514,
+      "step": 99215
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1040032572206724e-05,
+      "loss": 0.2589,
+      "step": 99220
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1037669441358933e-05,
+      "loss": 0.2746,
+      "step": 99225
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1035306491802815e-05,
+      "loss": 0.2571,
+      "step": 99230
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1032943723569039e-05,
+      "loss": 0.2562,
+      "step": 99235
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.103058113668828e-05,
+      "loss": 0.2581,
+      "step": 99240
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1028218731191215e-05,
+      "loss": 0.2801,
+      "step": 99245
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1025856507108529e-05,
+      "loss": 0.2654,
+      "step": 99250
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1023494464470885e-05,
+      "loss": 0.2478,
+      "step": 99255
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1021132603308949e-05,
+      "loss": 0.2558,
+      "step": 99260
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1018770923653391e-05,
+      "loss": 0.269,
+      "step": 99265
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1016409425534888e-05,
+      "loss": 0.2642,
+      "step": 99270
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1014048108984093e-05,
+      "loss": 0.2572,
+      "step": 99275
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1011686974031668e-05,
+      "loss": 0.2507,
+      "step": 99280
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1009326020708257e-05,
+      "loss": 0.2368,
+      "step": 99285
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.100696524904454e-05,
+      "loss": 0.2572,
+      "step": 99290
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1004604659071155e-05,
+      "loss": 0.2535,
+      "step": 99295
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1002244250818746e-05,
+      "loss": 0.2556,
+      "step": 99300
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.099988402431798e-05,
+      "loss": 0.2535,
+      "step": 99305
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0997523979599494e-05,
+      "loss": 0.2563,
+      "step": 99310
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0995164116693924e-05,
+      "loss": 0.261,
+      "step": 99315
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0992804435631915e-05,
+      "loss": 0.2546,
+      "step": 99320
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.099044493644412e-05,
+      "loss": 0.2593,
+      "step": 99325
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0988085619161165e-05,
+      "loss": 0.2948,
+      "step": 99330
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.0985726483813683e-05,
+      "loss": 0.2426,
+      "step": 99335
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0983367530432298e-05,
+      "loss": 0.2673,
+      "step": 99340
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.098100875904766e-05,
+      "loss": 0.2426,
+      "step": 99345
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.097865016969038e-05,
+      "loss": 0.259,
+      "step": 99350
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0976291762391092e-05,
+      "loss": 0.2576,
+      "step": 99355
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.09739335371804e-05,
+      "loss": 0.2639,
+      "step": 99360
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0971575494088948e-05,
+      "loss": 0.2561,
+      "step": 99365
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0969217633147336e-05,
+      "loss": 0.2426,
+      "step": 99370
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0966859954386194e-05,
+      "loss": 0.2757,
+      "step": 99375
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0964502457836118e-05,
+      "loss": 0.2638,
+      "step": 99380
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0962145143527735e-05,
+      "loss": 0.2468,
+      "step": 99385
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0959788011491649e-05,
+      "loss": 0.2661,
+      "step": 99390
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.095743106175846e-05,
+      "loss": 0.2553,
+      "step": 99395
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0955074294358769e-05,
+      "loss": 0.2562,
+      "step": 99400
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.095271770932319e-05,
+      "loss": 0.256,
+      "step": 99405
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0950361306682313e-05,
+      "loss": 0.2602,
+      "step": 99410
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0948005086466737e-05,
+      "loss": 0.2556,
+      "step": 99415
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0945649048707044e-05,
+      "loss": 0.2603,
+      "step": 99420
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0943293193433834e-05,
+      "loss": 0.2637,
+      "step": 99425
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.094093752067771e-05,
+      "loss": 0.2701,
+      "step": 99430
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0938582030469238e-05,
+      "loss": 0.2635,
+      "step": 99435
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0936226722839018e-05,
+      "loss": 0.2507,
+      "step": 99440
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.093387159781763e-05,
+      "loss": 0.2497,
+      "step": 99445
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0931516655435645e-05,
+      "loss": 0.253,
+      "step": 99450
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0929161895723639e-05,
+      "loss": 0.2477,
+      "step": 99455
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0926807318712199e-05,
+      "loss": 0.2548,
+      "step": 99460
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0924452924431895e-05,
+      "loss": 0.2588,
+      "step": 99465
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0922098712913289e-05,
+      "loss": 0.2448,
+      "step": 99470
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0919744684186947e-05,
+      "loss": 0.2646,
+      "step": 99475
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0917390838283442e-05,
+      "loss": 0.2531,
+      "step": 99480
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0915037175233345e-05,
+      "loss": 0.2822,
+      "step": 99485
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0912683695067205e-05,
+      "loss": 0.2679,
+      "step": 99490
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0910330397815576e-05,
+      "loss": 0.2748,
+      "step": 99495
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0907977283509032e-05,
+      "loss": 0.2566,
+      "step": 99500
+    },
+    {
+      "epoch": 3.5,
+      "eval_loss": 0.25208380818367004,
+      "eval_runtime": 10.545,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 99500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0905624352178115e-05,
+      "loss": 0.2539,
+      "step": 99505
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0903271603853376e-05,
+      "loss": 0.2521,
+      "step": 99510
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0900919038565357e-05,
+      "loss": 0.2631,
+      "step": 99515
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0898566656344622e-05,
+      "loss": 0.269,
+      "step": 99520
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0896214457221706e-05,
+      "loss": 0.2594,
+      "step": 99525
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.089386244122714e-05,
+      "loss": 0.2525,
+      "step": 99530
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0891510608391475e-05,
+      "loss": 0.2582,
+      "step": 99535
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0889158958745257e-05,
+      "loss": 0.2852,
+      "step": 99540
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.088680749231901e-05,
+      "loss": 0.2508,
+      "step": 99545
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0884456209143257e-05,
+      "loss": 0.263,
+      "step": 99550
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0882105109248547e-05,
+      "loss": 0.2546,
+      "step": 99555
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0879754192665397e-05,
+      "loss": 0.2612,
+      "step": 99560
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0877403459424335e-05,
+      "loss": 0.2559,
+      "step": 99565
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0875052909555872e-05,
+      "loss": 0.2799,
+      "step": 99570
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0872702543090547e-05,
+      "loss": 0.2395,
+      "step": 99575
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.087035236005887e-05,
+      "loss": 0.251,
+      "step": 99580
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0868002360491344e-05,
+      "loss": 0.2693,
+      "step": 99585
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0865652544418506e-05,
+      "loss": 0.2612,
+      "step": 99590
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0863302911870846e-05,
+      "loss": 0.2632,
+      "step": 99595
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0860953462878889e-05,
+      "loss": 0.2443,
+      "step": 99600
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0858604197473135e-05,
+      "loss": 0.2425,
+      "step": 99605
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0856255115684075e-05,
+      "loss": 0.2392,
+      "step": 99610
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.0853906217542231e-05,
+      "loss": 0.2635,
+      "step": 99615
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.085155750307809e-05,
+      "loss": 0.2571,
+      "step": 99620
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0849208972322156e-05,
+      "loss": 0.2595,
+      "step": 99625
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0846860625304906e-05,
+      "loss": 0.2321,
+      "step": 99630
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.084451246205685e-05,
+      "loss": 0.2596,
+      "step": 99635
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0842164482608463e-05,
+      "loss": 0.2488,
+      "step": 99640
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.083981668699025e-05,
+      "loss": 0.2511,
+      "step": 99645
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0837469075232678e-05,
+      "loss": 0.2557,
+      "step": 99650
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0835121647366243e-05,
+      "loss": 0.2735,
+      "step": 99655
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.083277440342142e-05,
+      "loss": 0.2476,
+      "step": 99660
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0830427343428673e-05,
+      "loss": 0.2678,
+      "step": 99665
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.08280804674185e-05,
+      "loss": 0.2673,
+      "step": 99670
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0825733775421362e-05,
+      "loss": 0.2607,
+      "step": 99675
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0823387267467733e-05,
+      "loss": 0.2653,
+      "step": 99680
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0821040943588066e-05,
+      "loss": 0.2476,
+      "step": 99685
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0818694803812846e-05,
+      "loss": 0.2518,
+      "step": 99690
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0816348848172528e-05,
+      "loss": 0.2531,
+      "step": 99695
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0814003076697565e-05,
+      "loss": 0.2656,
+      "step": 99700
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0811657489418423e-05,
+      "loss": 0.2415,
+      "step": 99705
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0809312086365566e-05,
+      "loss": 0.2528,
+      "step": 99710
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0806966867569441e-05,
+      "loss": 0.2445,
+      "step": 99715
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0804621833060496e-05,
+      "loss": 0.2759,
+      "step": 99720
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0802276982869176e-05,
+      "loss": 0.2405,
+      "step": 99725
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0799932317025937e-05,
+      "loss": 0.262,
+      "step": 99730
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.079758783556122e-05,
+      "loss": 0.2699,
+      "step": 99735
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0795243538505468e-05,
+      "loss": 0.2695,
+      "step": 99740
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0792899425889105e-05,
+      "loss": 0.2483,
+      "step": 99745
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.079055549774259e-05,
+      "loss": 0.2657,
+      "step": 99750
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.078821175409634e-05,
+      "loss": 0.2653,
+      "step": 99755
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0785868194980803e-05,
+      "loss": 0.2405,
+      "step": 99760
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.078352482042639e-05,
+      "loss": 0.268,
+      "step": 99765
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0781181630463547e-05,
+      "loss": 0.2504,
+      "step": 99770
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0778838625122692e-05,
+      "loss": 0.2692,
+      "step": 99775
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0776495804434233e-05,
+      "loss": 0.2375,
+      "step": 99780
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0774153168428616e-05,
+      "loss": 0.2549,
+      "step": 99785
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0771810717136242e-05,
+      "loss": 0.2362,
+      "step": 99790
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0769468450587528e-05,
+      "loss": 0.2604,
+      "step": 99795
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.076712636881288e-05,
+      "loss": 0.2494,
+      "step": 99800
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0764784471842726e-05,
+      "loss": 0.2472,
+      "step": 99805
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0762442759707456e-05,
+      "loss": 0.233,
+      "step": 99810
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0760101232437492e-05,
+      "loss": 0.2798,
+      "step": 99815
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0757759890063219e-05,
+      "loss": 0.253,
+      "step": 99820
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0755418732615059e-05,
+      "loss": 0.243,
+      "step": 99825
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0753077760123399e-05,
+      "loss": 0.2646,
+      "step": 99830
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0750736972618633e-05,
+      "loss": 0.2589,
+      "step": 99835
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0748396370131147e-05,
+      "loss": 0.2586,
+      "step": 99840
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0746055952691353e-05,
+      "loss": 0.2589,
+      "step": 99845
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0743715720329628e-05,
+      "loss": 0.2689,
+      "step": 99850
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.074137567307636e-05,
+      "loss": 0.257,
+      "step": 99855
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0739035810961923e-05,
+      "loss": 0.2485,
+      "step": 99860
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0736696134016708e-05,
+      "loss": 0.2514,
+      "step": 99865
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.07343566422711e-05,
+      "loss": 0.2608,
+      "step": 99870
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0732017335755473e-05,
+      "loss": 0.2521,
+      "step": 99875
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0729678214500189e-05,
+      "loss": 0.2462,
+      "step": 99880
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0727339278535636e-05,
+      "loss": 0.2633,
+      "step": 99885
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.072500052789218e-05,
+      "loss": 0.2736,
+      "step": 99890
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0722661962600169e-05,
+      "loss": 0.2583,
+      "step": 99895
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0720323582689998e-05,
+      "loss": 0.2506,
+      "step": 99900
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.0717985388192012e-05,
+      "loss": 0.2657,
+      "step": 99905
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0715647379136575e-05,
+      "loss": 0.2707,
+      "step": 99910
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0713309555554032e-05,
+      "loss": 0.2525,
+      "step": 99915
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0710971917474752e-05,
+      "loss": 0.2493,
+      "step": 99920
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0708634464929091e-05,
+      "loss": 0.2662,
+      "step": 99925
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0706297197947393e-05,
+      "loss": 0.2626,
+      "step": 99930
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0703960116559999e-05,
+      "loss": 0.2644,
+      "step": 99935
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0701623220797272e-05,
+      "loss": 0.256,
+      "step": 99940
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0699286510689541e-05,
+      "loss": 0.2534,
+      "step": 99945
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0696949986267152e-05,
+      "loss": 0.2299,
+      "step": 99950
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.069461364756043e-05,
+      "loss": 0.2468,
+      "step": 99955
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0692277494599734e-05,
+      "loss": 0.2435,
+      "step": 99960
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0689941527415384e-05,
+      "loss": 0.2582,
+      "step": 99965
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0687605746037705e-05,
+      "loss": 0.2606,
+      "step": 99970
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0685270150497032e-05,
+      "loss": 0.2559,
+      "step": 99975
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.06829347408237e-05,
+      "loss": 0.2676,
+      "step": 99980
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0681066546929894e-05,
+      "loss": 0.2738,
+      "step": 99985
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0678731471894177e-05,
+      "loss": 0.2751,
+      "step": 99990
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0676396582810693e-05,
+      "loss": 0.268,
+      "step": 99995
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0674061879709762e-05,
+      "loss": 0.2599,
+      "step": 100000
+    },
+    {
+      "epoch": 3.52,
+      "eval_loss": 0.2520838975906372,
+      "eval_runtime": 10.5365,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 100000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0671727362621686e-05,
+      "loss": 0.2531,
+      "step": 100005
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0669393031576788e-05,
+      "loss": 0.2713,
+      "step": 100010
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0667058886605388e-05,
+      "loss": 0.2459,
+      "step": 100015
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0664724927737783e-05,
+      "loss": 0.2487,
+      "step": 100020
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0662391155004265e-05,
+      "loss": 0.2499,
+      "step": 100025
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0660057568435158e-05,
+      "loss": 0.2693,
+      "step": 100030
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0657724168060756e-05,
+      "loss": 0.282,
+      "step": 100035
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0655390953911352e-05,
+      "loss": 0.2716,
+      "step": 100040
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0653057926017235e-05,
+      "loss": 0.2725,
+      "step": 100045
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0650725084408713e-05,
+      "loss": 0.2495,
+      "step": 100050
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0648392429116072e-05,
+      "loss": 0.232,
+      "step": 100055
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0646059960169585e-05,
+      "loss": 0.2373,
+      "step": 100060
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0643727677599553e-05,
+      "loss": 0.2393,
+      "step": 100065
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0641395581436264e-05,
+      "loss": 0.2558,
+      "step": 100070
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.063906367170999e-05,
+      "loss": 0.2498,
+      "step": 100075
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0636731948451001e-05,
+      "loss": 0.2549,
+      "step": 100080
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.063440041168959e-05,
+      "loss": 0.2481,
+      "step": 100085
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0632069061456024e-05,
+      "loss": 0.2654,
+      "step": 100090
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0629737897780574e-05,
+      "loss": 0.2545,
+      "step": 100095
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0627406920693494e-05,
+      "loss": 0.2416,
+      "step": 100100
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0625076130225075e-05,
+      "loss": 0.2567,
+      "step": 100105
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0622745526405566e-05,
+      "loss": 0.2655,
+      "step": 100110
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0620415109265228e-05,
+      "loss": 0.2459,
+      "step": 100115
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.061808487883432e-05,
+      "loss": 0.2647,
+      "step": 100120
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.061575483514311e-05,
+      "loss": 0.2402,
+      "step": 100125
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0613424978221845e-05,
+      "loss": 0.2637,
+      "step": 100130
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0611095308100774e-05,
+      "loss": 0.2624,
+      "step": 100135
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.060876582481014e-05,
+      "loss": 0.2645,
+      "step": 100140
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0606436528380207e-05,
+      "loss": 0.2463,
+      "step": 100145
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0604107418841209e-05,
+      "loss": 0.2536,
+      "step": 100150
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0601778496223389e-05,
+      "loss": 0.2507,
+      "step": 100155
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0599449760556978e-05,
+      "loss": 0.2748,
+      "step": 100160
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0597121211872232e-05,
+      "loss": 0.253,
+      "step": 100165
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0594792850199363e-05,
+      "loss": 0.2503,
+      "step": 100170
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0592464675568627e-05,
+      "loss": 0.2762,
+      "step": 100175
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0590136688010232e-05,
+      "loss": 0.2657,
+      "step": 100180
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0587808887554427e-05,
+      "loss": 0.2391,
+      "step": 100185
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.0585481274231421e-05,
+      "loss": 0.2671,
+      "step": 100190
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0583153848071437e-05,
+      "loss": 0.2434,
+      "step": 100195
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0580826609104704e-05,
+      "loss": 0.2563,
+      "step": 100200
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0578499557361438e-05,
+      "loss": 0.2772,
+      "step": 100205
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.057617269287185e-05,
+      "loss": 0.2472,
+      "step": 100210
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0573846015666145e-05,
+      "loss": 0.2396,
+      "step": 100215
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.057151952577455e-05,
+      "loss": 0.2741,
+      "step": 100220
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0569193223227259e-05,
+      "loss": 0.2355,
+      "step": 100225
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.056686710805449e-05,
+      "loss": 0.2578,
+      "step": 100230
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0564541180286434e-05,
+      "loss": 0.2587,
+      "step": 100235
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0562215439953302e-05,
+      "loss": 0.2684,
+      "step": 100240
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0559889887085288e-05,
+      "loss": 0.2392,
+      "step": 100245
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0557564521712593e-05,
+      "loss": 0.2883,
+      "step": 100250
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0555239343865392e-05,
+      "loss": 0.2483,
+      "step": 100255
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0552914353573897e-05,
+      "loss": 0.2482,
+      "step": 100260
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0550589550868289e-05,
+      "loss": 0.293,
+      "step": 100265
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0548264935778754e-05,
+      "loss": 0.2732,
+      "step": 100270
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0545940508335464e-05,
+      "loss": 0.2652,
+      "step": 100275
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0544081101506425e-05,
+      "loss": 0.2416,
+      "step": 100280
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0541757011902451e-05,
+      "loss": 0.2487,
+      "step": 100285
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0539433110029242e-05,
+      "loss": 0.2596,
+      "step": 100290
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0537109395916966e-05,
+      "loss": 0.2557,
+      "step": 100295
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0534785869595793e-05,
+      "loss": 0.2529,
+      "step": 100300
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0532462531095883e-05,
+      "loss": 0.2673,
+      "step": 100305
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0530139380447421e-05,
+      "loss": 0.2484,
+      "step": 100310
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0527816417680555e-05,
+      "loss": 0.2821,
+      "step": 100315
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0525493642825462e-05,
+      "loss": 0.251,
+      "step": 100320
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.052317105591229e-05,
+      "loss": 0.266,
+      "step": 100325
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0520848656971208e-05,
+      "loss": 0.3008,
+      "step": 100330
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0518526446032365e-05,
+      "loss": 0.2552,
+      "step": 100335
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.051620442312591e-05,
+      "loss": 0.2618,
+      "step": 100340
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.051388258828199e-05,
+      "loss": 0.2678,
+      "step": 100345
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0511560941530765e-05,
+      "loss": 0.2644,
+      "step": 100350
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0509239482902373e-05,
+      "loss": 0.2646,
+      "step": 100355
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0506918212426956e-05,
+      "loss": 0.3034,
+      "step": 100360
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0504597130134647e-05,
+      "loss": 0.2515,
+      "step": 100365
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0502276236055589e-05,
+      "loss": 0.2508,
+      "step": 100370
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.049995553021993e-05,
+      "loss": 0.2688,
+      "step": 100375
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0497635012657783e-05,
+      "loss": 0.2463,
+      "step": 100380
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0495314683399298e-05,
+      "loss": 0.2498,
+      "step": 100385
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0492994542474593e-05,
+      "loss": 0.2574,
+      "step": 100390
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0490674589913794e-05,
+      "loss": 0.2614,
+      "step": 100395
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0488354825747013e-05,
+      "loss": 0.2473,
+      "step": 100400
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0486035250004392e-05,
+      "loss": 0.2613,
+      "step": 100405
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0483715862716037e-05,
+      "loss": 0.2537,
+      "step": 100410
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0481396663912063e-05,
+      "loss": 0.2646,
+      "step": 100415
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0479077653622579e-05,
+      "loss": 0.2528,
+      "step": 100420
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.047675883187771e-05,
+      "loss": 0.2698,
+      "step": 100425
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0474440198707546e-05,
+      "loss": 0.2347,
+      "step": 100430
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0472121754142213e-05,
+      "loss": 0.2728,
+      "step": 100435
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0469803498211797e-05,
+      "loss": 0.2621,
+      "step": 100440
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0467485430946416e-05,
+      "loss": 0.2744,
+      "step": 100445
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0465167552376159e-05,
+      "loss": 0.2605,
+      "step": 100450
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.046284986253112e-05,
+      "loss": 0.2494,
+      "step": 100455
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0460532361441391e-05,
+      "loss": 0.25,
+      "step": 100460
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0458215049137074e-05,
+      "loss": 0.2382,
+      "step": 100465
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.045589792564825e-05,
+      "loss": 0.2719,
+      "step": 100470
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.0453580991004999e-05,
+      "loss": 0.2378,
+      "step": 100475
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0451264245237422e-05,
+      "loss": 0.2533,
+      "step": 100480
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.044894768837558e-05,
+      "loss": 0.2508,
+      "step": 100485
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0446631320449571e-05,
+      "loss": 0.2745,
+      "step": 100490
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0444315141489458e-05,
+      "loss": 0.248,
+      "step": 100495
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0441999151525326e-05,
+      "loss": 0.2534,
+      "step": 100500
+    },
+    {
+      "epoch": 3.54,
+      "eval_loss": 0.25199830532073975,
+      "eval_runtime": 10.5358,
+      "eval_samples_per_second": 9.491,
+      "eval_steps_per_second": 9.491,
+      "step": 100500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0439683350587243e-05,
+      "loss": 0.2719,
+      "step": 100505
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0437367738705273e-05,
+      "loss": 0.2638,
+      "step": 100510
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.043505231590948e-05,
+      "loss": 0.2566,
+      "step": 100515
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.043273708222994e-05,
+      "loss": 0.256,
+      "step": 100520
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0430422037696713e-05,
+      "loss": 0.2569,
+      "step": 100525
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.042810718233984e-05,
+      "loss": 0.2712,
+      "step": 100530
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0425792516189406e-05,
+      "loss": 0.2612,
+      "step": 100535
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0423478039275438e-05,
+      "loss": 0.2688,
+      "step": 100540
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0421163751628016e-05,
+      "loss": 0.2562,
+      "step": 100545
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0418849653277171e-05,
+      "loss": 0.2504,
+      "step": 100550
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0416535744252945e-05,
+      "loss": 0.2534,
+      "step": 100555
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0414222024585402e-05,
+      "loss": 0.2655,
+      "step": 100560
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.041190849430457e-05,
+      "loss": 0.2611,
+      "step": 100565
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0409595153440497e-05,
+      "loss": 0.2421,
+      "step": 100570
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0407282002023203e-05,
+      "loss": 0.2549,
+      "step": 100575
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0404969040082746e-05,
+      "loss": 0.2332,
+      "step": 100580
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0402656267649148e-05,
+      "loss": 0.23,
+      "step": 100585
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0400343684752428e-05,
+      "loss": 0.2538,
+      "step": 100590
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0398031291422621e-05,
+      "loss": 0.2485,
+      "step": 100595
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0395719087689768e-05,
+      "loss": 0.2665,
+      "step": 100600
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0393407073583877e-05,
+      "loss": 0.2479,
+      "step": 100605
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0391095249134969e-05,
+      "loss": 0.2536,
+      "step": 100610
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0388783614373052e-05,
+      "loss": 0.2444,
+      "step": 100615
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0386472169328159e-05,
+      "loss": 0.2464,
+      "step": 100620
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0384160914030297e-05,
+      "loss": 0.2735,
+      "step": 100625
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0381849848509461e-05,
+      "loss": 0.2655,
+      "step": 100630
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0379538972795683e-05,
+      "loss": 0.259,
+      "step": 100635
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0377228286918956e-05,
+      "loss": 0.2766,
+      "step": 100640
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0374917790909272e-05,
+      "loss": 0.2667,
+      "step": 100645
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0372607484796643e-05,
+      "loss": 0.2517,
+      "step": 100650
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0370297368611073e-05,
+      "loss": 0.28,
+      "step": 100655
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0367987442382554e-05,
+      "loss": 0.2532,
+      "step": 100660
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0365677706141072e-05,
+      "loss": 0.2579,
+      "step": 100665
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0363368159916614e-05,
+      "loss": 0.2502,
+      "step": 100670
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0361058803739182e-05,
+      "loss": 0.2588,
+      "step": 100675
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0358749637638754e-05,
+      "loss": 0.2443,
+      "step": 100680
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0356440661645313e-05,
+      "loss": 0.2402,
+      "step": 100685
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0354131875788833e-05,
+      "loss": 0.2557,
+      "step": 100690
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0351823280099304e-05,
+      "loss": 0.2556,
+      "step": 100695
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0349514874606692e-05,
+      "loss": 0.2534,
+      "step": 100700
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.034720665934098e-05,
+      "loss": 0.255,
+      "step": 100705
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0344898634332123e-05,
+      "loss": 0.258,
+      "step": 100710
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0342590799610111e-05,
+      "loss": 0.2445,
+      "step": 100715
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0340283155204895e-05,
+      "loss": 0.2602,
+      "step": 100720
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.033797570114644e-05,
+      "loss": 0.2574,
+      "step": 100725
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0335668437464705e-05,
+      "loss": 0.2721,
+      "step": 100730
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0333361364189652e-05,
+      "loss": 0.2454,
+      "step": 100735
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0331054481351244e-05,
+      "loss": 0.2615,
+      "step": 100740
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0328747788979412e-05,
+      "loss": 0.2755,
+      "step": 100745
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0326441287104132e-05,
+      "loss": 0.244,
+      "step": 100750
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.0324134975755334e-05,
+      "loss": 0.2653,
+      "step": 100755
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0321828854962978e-05,
+      "loss": 0.2529,
+      "step": 100760
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0319522924756994e-05,
+      "loss": 0.2557,
+      "step": 100765
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0317217185167338e-05,
+      "loss": 0.2609,
+      "step": 100770
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0314911636223942e-05,
+      "loss": 0.2805,
+      "step": 100775
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0312606277956738e-05,
+      "loss": 0.2377,
+      "step": 100780
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0310301110395657e-05,
+      "loss": 0.2782,
+      "step": 100785
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.030799613357064e-05,
+      "loss": 0.2543,
+      "step": 100790
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0305691347511612e-05,
+      "loss": 0.2685,
+      "step": 100795
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0303386752248497e-05,
+      "loss": 0.2632,
+      "step": 100800
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0301082347811214e-05,
+      "loss": 0.2473,
+      "step": 100805
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0298778134229689e-05,
+      "loss": 0.2704,
+      "step": 100810
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0296474111533849e-05,
+      "loss": 0.2618,
+      "step": 100815
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0294170279753602e-05,
+      "loss": 0.2552,
+      "step": 100820
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0291866638918854e-05,
+      "loss": 0.2588,
+      "step": 100825
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0289563189059534e-05,
+      "loss": 0.2607,
+      "step": 100830
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0287259930205542e-05,
+      "loss": 0.2637,
+      "step": 100835
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0284956862386783e-05,
+      "loss": 0.2253,
+      "step": 100840
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0282653985633154e-05,
+      "loss": 0.2524,
+      "step": 100845
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0280351299974572e-05,
+      "loss": 0.243,
+      "step": 100850
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0278048805440926e-05,
+      "loss": 0.2449,
+      "step": 100855
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0275746502062108e-05,
+      "loss": 0.2679,
+      "step": 100860
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0273444389868025e-05,
+      "loss": 0.2617,
+      "step": 100865
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0271142468888553e-05,
+      "loss": 0.2594,
+      "step": 100870
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0268840739153598e-05,
+      "loss": 0.2612,
+      "step": 100875
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.026653920069303e-05,
+      "loss": 0.2597,
+      "step": 100880
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0264237853536748e-05,
+      "loss": 0.2657,
+      "step": 100885
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0261936697714628e-05,
+      "loss": 0.259,
+      "step": 100890
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0259635733256548e-05,
+      "loss": 0.254,
+      "step": 100895
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0257334960192372e-05,
+      "loss": 0.2554,
+      "step": 100900
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0255034378551994e-05,
+      "loss": 0.2457,
+      "step": 100905
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0252733988365281e-05,
+      "loss": 0.2942,
+      "step": 100910
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0250433789662087e-05,
+      "loss": 0.2503,
+      "step": 100915
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0248133782472299e-05,
+      "loss": 0.2548,
+      "step": 100920
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0245833966825766e-05,
+      "loss": 0.2613,
+      "step": 100925
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0243534342752362e-05,
+      "loss": 0.2564,
+      "step": 100930
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0241234910281938e-05,
+      "loss": 0.2618,
+      "step": 100935
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0238935669444347e-05,
+      "loss": 0.2408,
+      "step": 100940
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0236636620269455e-05,
+      "loss": 0.2501,
+      "step": 100945
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0234337762787107e-05,
+      "loss": 0.2572,
+      "step": 100950
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0232039097027151e-05,
+      "loss": 0.2402,
+      "step": 100955
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0229740623019429e-05,
+      "loss": 0.2628,
+      "step": 100960
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0227442340793799e-05,
+      "loss": 0.2499,
+      "step": 100965
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0225144250380086e-05,
+      "loss": 0.2416,
+      "step": 100970
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0222846351808146e-05,
+      "loss": 0.2722,
+      "step": 100975
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.02205486451078e-05,
+      "loss": 0.251,
+      "step": 100980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0218251130308898e-05,
+      "loss": 0.2553,
+      "step": 100985
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0215953807441262e-05,
+      "loss": 0.2722,
+      "step": 100990
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0213656676534717e-05,
+      "loss": 0.2825,
+      "step": 100995
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.02113597376191e-05,
+      "loss": 0.2753,
+      "step": 101000
+    },
+    {
+      "epoch": 3.55,
+      "eval_loss": 0.2520409822463989,
+      "eval_runtime": 10.5341,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 101000
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0209062990724233e-05,
+      "loss": 0.2884,
+      "step": 101005
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0206766435879937e-05,
+      "loss": 0.2577,
+      "step": 101010
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0204470073116021e-05,
+      "loss": 0.2315,
+      "step": 101015
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0202173902462317e-05,
+      "loss": 0.2592,
+      "step": 101020
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0199877923948634e-05,
+      "loss": 0.2662,
+      "step": 101025
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0197582137604773e-05,
+      "loss": 0.2575,
+      "step": 101030
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0195286543460553e-05,
+      "loss": 0.2617,
+      "step": 101035
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.0192991141545788e-05,
+      "loss": 0.2393,
+      "step": 101040
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0190695931890277e-05,
+      "loss": 0.2726,
+      "step": 101045
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0188400914523818e-05,
+      "loss": 0.2583,
+      "step": 101050
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0186106089476205e-05,
+      "loss": 0.2676,
+      "step": 101055
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.018381145677725e-05,
+      "loss": 0.2549,
+      "step": 101060
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0181517016456738e-05,
+      "loss": 0.2461,
+      "step": 101065
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0179222768544463e-05,
+      "loss": 0.2421,
+      "step": 101070
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0176928713070205e-05,
+      "loss": 0.2284,
+      "step": 101075
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0174634850063765e-05,
+      "loss": 0.2318,
+      "step": 101080
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0172341179554915e-05,
+      "loss": 0.2498,
+      "step": 101085
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.017004770157344e-05,
+      "loss": 0.2441,
+      "step": 101090
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0167754416149134e-05,
+      "loss": 0.2806,
+      "step": 101095
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0165461323311762e-05,
+      "loss": 0.2589,
+      "step": 101100
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0163168423091096e-05,
+      "loss": 0.2643,
+      "step": 101105
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0160875715516902e-05,
+      "loss": 0.2619,
+      "step": 101110
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0158583200618967e-05,
+      "loss": 0.2684,
+      "step": 101115
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.015629087842705e-05,
+      "loss": 0.2496,
+      "step": 101120
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.015399874897091e-05,
+      "loss": 0.2581,
+      "step": 101125
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0151706812280306e-05,
+      "loss": 0.2518,
+      "step": 101130
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0149415068385013e-05,
+      "loss": 0.2583,
+      "step": 101135
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0147123517314765e-05,
+      "loss": 0.2328,
+      "step": 101140
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0144832159099343e-05,
+      "loss": 0.2369,
+      "step": 101145
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0142540993768476e-05,
+      "loss": 0.2563,
+      "step": 101150
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0140250021351932e-05,
+      "loss": 0.2836,
+      "step": 101155
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.013795924187945e-05,
+      "loss": 0.2418,
+      "step": 101160
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.013566865538077e-05,
+      "loss": 0.2433,
+      "step": 101165
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0133378261885626e-05,
+      "loss": 0.2598,
+      "step": 101170
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0131088061423779e-05,
+      "loss": 0.2743,
+      "step": 101175
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0128798054024954e-05,
+      "loss": 0.2507,
+      "step": 101180
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0126508239718888e-05,
+      "loss": 0.2571,
+      "step": 101185
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.01242186185353e-05,
+      "loss": 0.2496,
+      "step": 101190
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0121929190503929e-05,
+      "loss": 0.2665,
+      "step": 101195
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0119639955654512e-05,
+      "loss": 0.2489,
+      "step": 101200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0117350914016754e-05,
+      "loss": 0.2636,
+      "step": 101205
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0115062065620398e-05,
+      "loss": 0.2412,
+      "step": 101210
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.011277341049515e-05,
+      "loss": 0.2654,
+      "step": 101215
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0110484948670726e-05,
+      "loss": 0.2726,
+      "step": 101220
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0108196680176835e-05,
+      "loss": 0.251,
+      "step": 101225
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0105908605043204e-05,
+      "loss": 0.2586,
+      "step": 101230
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0103620723299537e-05,
+      "loss": 0.2657,
+      "step": 101235
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0101333034975535e-05,
+      "loss": 0.2426,
+      "step": 101240
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0099045540100898e-05,
+      "loss": 0.261,
+      "step": 101245
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0096758238705334e-05,
+      "loss": 0.2671,
+      "step": 101250
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.009447113081855e-05,
+      "loss": 0.2714,
+      "step": 101255
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0092184216470238e-05,
+      "loss": 0.254,
+      "step": 101260
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0089897495690079e-05,
+      "loss": 0.2622,
+      "step": 101265
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0087610968507786e-05,
+      "loss": 0.2505,
+      "step": 101270
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0085324634953033e-05,
+      "loss": 0.2544,
+      "step": 101275
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0083038495055511e-05,
+      "loss": 0.2556,
+      "step": 101280
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0080752548844895e-05,
+      "loss": 0.241,
+      "step": 101285
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0078466796350885e-05,
+      "loss": 0.2487,
+      "step": 101290
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0076181237603152e-05,
+      "loss": 0.2585,
+      "step": 101295
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0073895872631359e-05,
+      "loss": 0.265,
+      "step": 101300
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.00716107014652e-05,
+      "loss": 0.2519,
+      "step": 101305
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0069325724134328e-05,
+      "loss": 0.2672,
+      "step": 101310
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0067040940668435e-05,
+      "loss": 0.241,
+      "step": 101315
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0064756351097163e-05,
+      "loss": 0.2439,
+      "step": 101320
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.0062471955450194e-05,
+      "loss": 0.2713,
+      "step": 101325
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0060187753757183e-05,
+      "loss": 0.25,
+      "step": 101330
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.005790374604779e-05,
+      "loss": 0.2872,
+      "step": 101335
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.005561993235166e-05,
+      "loss": 0.2552,
+      "step": 101340
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0053336312698466e-05,
+      "loss": 0.2638,
+      "step": 101345
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0051052887117848e-05,
+      "loss": 0.2532,
+      "step": 101350
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.004876965563945e-05,
+      "loss": 0.2728,
+      "step": 101355
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0046486618292933e-05,
+      "loss": 0.2583,
+      "step": 101360
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0044203775107921e-05,
+      "loss": 0.2902,
+      "step": 101365
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0041921126114079e-05,
+      "loss": 0.2379,
+      "step": 101370
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0039638671341031e-05,
+      "loss": 0.2592,
+      "step": 101375
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.003735641081841e-05,
+      "loss": 0.2424,
+      "step": 101380
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0035074344575863e-05,
+      "loss": 0.2278,
+      "step": 101385
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.003279247264301e-05,
+      "loss": 0.2786,
+      "step": 101390
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0030510795049486e-05,
+      "loss": 0.2683,
+      "step": 101395
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0028229311824904e-05,
+      "loss": 0.2549,
+      "step": 101400
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0025948022998908e-05,
+      "loss": 0.2537,
+      "step": 101405
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.00236669286011e-05,
+      "loss": 0.2422,
+      "step": 101410
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0021386028661115e-05,
+      "loss": 0.2395,
+      "step": 101415
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0019105323208549e-05,
+      "loss": 0.2659,
+      "step": 101420
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0016824812273042e-05,
+      "loss": 0.2536,
+      "step": 101425
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0014544495884188e-05,
+      "loss": 0.2819,
+      "step": 101430
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.001226437407159e-05,
+      "loss": 0.2659,
+      "step": 101435
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.000998444686487e-05,
+      "loss": 0.2414,
+      "step": 101440
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0007704714293623e-05,
+      "loss": 0.2716,
+      "step": 101445
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0005425176387448e-05,
+      "loss": 0.2525,
+      "step": 101450
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.000314583317594e-05,
+      "loss": 0.2505,
+      "step": 101455
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.0000866684688706e-05,
+      "loss": 0.2487,
+      "step": 101460
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.998587730955336e-06,
+      "loss": 0.2531,
+      "step": 101465
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.996308972005408e-06,
+      "loss": 0.2609,
+      "step": 101470
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.99403040786852e-06,
+      "loss": 0.26,
+      "step": 101475
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.991752038574267e-06,
+      "loss": 0.2382,
+      "step": 101480
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.989473864152221e-06,
+      "loss": 0.2853,
+      "step": 101485
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.987195884631967e-06,
+      "loss": 0.2495,
+      "step": 101490
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.984918100043072e-06,
+      "loss": 0.2519,
+      "step": 101495
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.982640510415128e-06,
+      "loss": 0.2421,
+      "step": 101500
+    },
+    {
+      "epoch": 3.57,
+      "eval_loss": 0.25163590908050537,
+      "eval_runtime": 10.5398,
+      "eval_samples_per_second": 9.488,
+      "eval_steps_per_second": 9.488,
+      "step": 101500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.9803631157777e-06,
+      "loss": 0.2317,
+      "step": 101505
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.978085916160359e-06,
+      "loss": 0.265,
+      "step": 101510
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.975808911592662e-06,
+      "loss": 0.2652,
+      "step": 101515
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.973532102104197e-06,
+      "loss": 0.2534,
+      "step": 101520
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.971255487724504e-06,
+      "loss": 0.2545,
+      "step": 101525
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.968979068483164e-06,
+      "loss": 0.2848,
+      "step": 101530
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.966702844409714e-06,
+      "loss": 0.2729,
+      "step": 101535
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.964426815533732e-06,
+      "loss": 0.2398,
+      "step": 101540
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.962150981884757e-06,
+      "loss": 0.2565,
+      "step": 101545
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.959875343492334e-06,
+      "loss": 0.2747,
+      "step": 101550
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.957599900386024e-06,
+      "loss": 0.2581,
+      "step": 101555
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.955324652595369e-06,
+      "loss": 0.2666,
+      "step": 101560
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.953049600149908e-06,
+      "loss": 0.2615,
+      "step": 101565
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.950774743079172e-06,
+      "loss": 0.2477,
+      "step": 101570
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.948500081412717e-06,
+      "loss": 0.2734,
+      "step": 101575
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.94622561518006e-06,
+      "loss": 0.2722,
+      "step": 101580
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.943951344410751e-06,
+      "loss": 0.248,
+      "step": 101585
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.941677269134303e-06,
+      "loss": 0.259,
+      "step": 101590
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.939403389380261e-06,
+      "loss": 0.2713,
+      "step": 101595
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.937129705178142e-06,
+      "loss": 0.2572,
+      "step": 101600
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.934856216557465e-06,
+      "loss": 0.2642,
+      "step": 101605
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.932582923547745e-06,
+      "loss": 0.2454,
+      "step": 101610
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.930309826178513e-06,
+      "loss": 0.2612,
+      "step": 101615
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.928036924479276e-06,
+      "loss": 0.2666,
+      "step": 101620
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.925764218479549e-06,
+      "loss": 0.2315,
+      "step": 101625
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.923491708208829e-06,
+      "loss": 0.2454,
+      "step": 101630
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.921219393696634e-06,
+      "loss": 0.2555,
+      "step": 101635
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.918947274972476e-06,
+      "loss": 0.2696,
+      "step": 101640
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.916675352065849e-06,
+      "loss": 0.2541,
+      "step": 101645
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.914403625006245e-06,
+      "loss": 0.244,
+      "step": 101650
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.912132093823173e-06,
+      "loss": 0.2263,
+      "step": 101655
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.909860758546128e-06,
+      "loss": 0.2481,
+      "step": 101660
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.907589619204583e-06,
+      "loss": 0.2358,
+      "step": 101665
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.905318675828052e-06,
+      "loss": 0.2522,
+      "step": 101670
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.903047928446007e-06,
+      "loss": 0.2481,
+      "step": 101675
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.900777377087939e-06,
+      "loss": 0.2524,
+      "step": 101680
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.898507021783315e-06,
+      "loss": 0.2466,
+      "step": 101685
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.896236862561625e-06,
+      "loss": 0.2591,
+      "step": 101690
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.893966899452356e-06,
+      "loss": 0.2576,
+      "step": 101695
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.89169713248497e-06,
+      "loss": 0.2258,
+      "step": 101700
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.889427561688933e-06,
+      "loss": 0.2534,
+      "step": 101705
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.88715818709373e-06,
+      "loss": 0.2556,
+      "step": 101710
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.884889008728814e-06,
+      "loss": 0.239,
+      "step": 101715
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.882620026623657e-06,
+      "loss": 0.2524,
+      "step": 101720
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.880351240807707e-06,
+      "loss": 0.2553,
+      "step": 101725
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.878082651310439e-06,
+      "loss": 0.2408,
+      "step": 101730
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.875814258161303e-06,
+      "loss": 0.2744,
+      "step": 101735
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.873546061389743e-06,
+      "loss": 0.2682,
+      "step": 101740
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.871278061025227e-06,
+      "loss": 0.2772,
+      "step": 101745
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.869010257097185e-06,
+      "loss": 0.2773,
+      "step": 101750
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.866742649635085e-06,
+      "loss": 0.2547,
+      "step": 101755
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.864475238668358e-06,
+      "loss": 0.2462,
+      "step": 101760
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.862208024226435e-06,
+      "loss": 0.2654,
+      "step": 101765
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.859941006338773e-06,
+      "loss": 0.2604,
+      "step": 101770
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.8576741850348e-06,
+      "loss": 0.258,
+      "step": 101775
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.855407560343941e-06,
+      "loss": 0.2535,
+      "step": 101780
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.853141132295643e-06,
+      "loss": 0.2421,
+      "step": 101785
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.850874900919325e-06,
+      "loss": 0.2586,
+      "step": 101790
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.848608866244403e-06,
+      "loss": 0.2664,
+      "step": 101795
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.846343028300318e-06,
+      "loss": 0.2558,
+      "step": 101800
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.844077387116474e-06,
+      "loss": 0.2521,
+      "step": 101805
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.841811942722306e-06,
+      "loss": 0.2885,
+      "step": 101810
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.83954669514722e-06,
+      "loss": 0.2562,
+      "step": 101815
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.837281644420621e-06,
+      "loss": 0.2398,
+      "step": 101820
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.835016790571935e-06,
+      "loss": 0.2481,
+      "step": 101825
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.832752133630563e-06,
+      "loss": 0.279,
+      "step": 101830
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.830487673625907e-06,
+      "loss": 0.2565,
+      "step": 101835
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.828223410587364e-06,
+      "loss": 0.2632,
+      "step": 101840
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.825959344544353e-06,
+      "loss": 0.2412,
+      "step": 101845
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.823695475526245e-06,
+      "loss": 0.2662,
+      "step": 101850
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.821431803562465e-06,
+      "loss": 0.2596,
+      "step": 101855
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.819168328682377e-06,
+      "loss": 0.2805,
+      "step": 101860
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.816905050915392e-06,
+      "loss": 0.2598,
+      "step": 101865
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.814641970290891e-06,
+      "loss": 0.2627,
+      "step": 101870
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.812379086838255e-06,
+      "loss": 0.2407,
+      "step": 101875
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.810116400586861e-06,
+      "loss": 0.2531,
+      "step": 101880
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.807853911566103e-06,
+      "loss": 0.2484,
+      "step": 101885
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.80559161980535e-06,
+      "loss": 0.2519,
+      "step": 101890
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.803329525333977e-06,
+      "loss": 0.2567,
+      "step": 101895
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.801067628181343e-06,
+      "loss": 0.2459,
+      "step": 101900
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.798805928376839e-06,
+      "loss": 0.2837,
+      "step": 101905
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.796544425949817e-06,
+      "loss": 0.2564,
+      "step": 101910
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.794283120929643e-06,
+      "loss": 0.2478,
+      "step": 101915
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.792022013345692e-06,
+      "loss": 0.2534,
+      "step": 101920
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.78976110322731e-06,
+      "loss": 0.2457,
+      "step": 101925
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.78750039060386e-06,
+      "loss": 0.2703,
+      "step": 101930
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.785239875504682e-06,
+      "loss": 0.2379,
+      "step": 101935
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.782979557959142e-06,
+      "loss": 0.2682,
+      "step": 101940
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.780719437996588e-06,
+      "loss": 0.2586,
+      "step": 101945
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.778459515646363e-06,
+      "loss": 0.2426,
+      "step": 101950
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.776199790937798e-06,
+      "loss": 0.2492,
+      "step": 101955
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.773940263900258e-06,
+      "loss": 0.2677,
+      "step": 101960
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.77168093456306e-06,
+      "loss": 0.2639,
+      "step": 101965
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.769421802955556e-06,
+      "loss": 0.2521,
+      "step": 101970
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.767162869107063e-06,
+      "loss": 0.2546,
+      "step": 101975
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.764904133046934e-06,
+      "loss": 0.2585,
+      "step": 101980
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.762645594804481e-06,
+      "loss": 0.2574,
+      "step": 101985
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.760387254409034e-06,
+      "loss": 0.2536,
+      "step": 101990
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.758129111889908e-06,
+      "loss": 0.2695,
+      "step": 101995
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.75587116727644e-06,
+      "loss": 0.2769,
+      "step": 102000
+    },
+    {
+      "epoch": 3.59,
+      "eval_loss": 0.25186842679977417,
+      "eval_runtime": 10.5576,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 9.472,
+      "step": 102000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.753613420597935e-06,
+      "loss": 0.2926,
+      "step": 102005
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.751355871883713e-06,
+      "loss": 0.2503,
+      "step": 102010
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.74909852116308e-06,
+      "loss": 0.2665,
+      "step": 102015
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.746841368465351e-06,
+      "loss": 0.2326,
+      "step": 102020
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.74458441381984e-06,
+      "loss": 0.2563,
+      "step": 102025
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.742327657255843e-06,
+      "loss": 0.248,
+      "step": 102030
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.740071098802672e-06,
+      "loss": 0.2442,
+      "step": 102035
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.73781473848962e-06,
+      "loss": 0.2555,
+      "step": 102040
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.735558576345988e-06,
+      "loss": 0.2531,
+      "step": 102045
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.733302612401058e-06,
+      "loss": 0.2493,
+      "step": 102050
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.731046846684142e-06,
+      "loss": 0.2575,
+      "step": 102055
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.728791279224519e-06,
+      "loss": 0.2549,
+      "step": 102060
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.726535910051477e-06,
+      "loss": 0.2429,
+      "step": 102065
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.724280739194294e-06,
+      "loss": 0.2302,
+      "step": 102070
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.722025766682254e-06,
+      "loss": 0.2686,
+      "step": 102075
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.719770992544655e-06,
+      "loss": 0.2265,
+      "step": 102080
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.717516416810756e-06,
+      "loss": 0.2452,
+      "step": 102085
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.715262039509829e-06,
+      "loss": 0.2526,
+      "step": 102090
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.713007860671158e-06,
+      "loss": 0.2746,
+      "step": 102095
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.710753880324005e-06,
+      "loss": 0.2545,
+      "step": 102100
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.708500098497638e-06,
+      "loss": 0.2662,
+      "step": 102105
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.706246515221312e-06,
+      "loss": 0.2388,
+      "step": 102110
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.703993130524302e-06,
+      "loss": 0.2357,
+      "step": 102115
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.701739944435864e-06,
+      "loss": 0.2314,
+      "step": 102120
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.69948695698524e-06,
+      "loss": 0.249,
+      "step": 102125
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.697234168201696e-06,
+      "loss": 0.262,
+      "step": 102130
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.69498157811449e-06,
+      "loss": 0.2401,
+      "step": 102135
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.69272918675286e-06,
+      "loss": 0.2485,
+      "step": 102140
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.690476994146044e-06,
+      "loss": 0.2577,
+      "step": 102145
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.688225000323306e-06,
+      "loss": 0.2478,
+      "step": 102150
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.685973205313876e-06,
+      "loss": 0.2574,
+      "step": 102155
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.683721609146989e-06,
+      "loss": 0.255,
+      "step": 102160
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.681470211851878e-06,
+      "loss": 0.2443,
+      "step": 102165
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.679219013457789e-06,
+      "loss": 0.2414,
+      "step": 102170
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.676968013993943e-06,
+      "loss": 0.2342,
+      "step": 102175
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.67471721348956e-06,
+      "loss": 0.2552,
+      "step": 102180
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.672466611973882e-06,
+      "loss": 0.2323,
+      "step": 102185
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.670216209476118e-06,
+      "loss": 0.2672,
+      "step": 102190
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.6679660060255e-06,
+      "loss": 0.259,
+      "step": 102195
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.66571600165124e-06,
+      "loss": 0.2511,
+      "step": 102200
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.663466196382544e-06,
+      "loss": 0.2588,
+      "step": 102205
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.66121659024864e-06,
+      "loss": 0.2726,
+      "step": 102210
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.658967183278731e-06,
+      "loss": 0.2671,
+      "step": 102215
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.656717975502023e-06,
+      "loss": 0.2524,
+      "step": 102220
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.654468966947711e-06,
+      "loss": 0.2584,
+      "step": 102225
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.652220157645014e-06,
+      "loss": 0.2484,
+      "step": 102230
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.649971547623119e-06,
+      "loss": 0.2767,
+      "step": 102235
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.647723136911233e-06,
+      "loss": 0.2603,
+      "step": 102240
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.645474925538536e-06,
+      "loss": 0.2558,
+      "step": 102245
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.643226913534236e-06,
+      "loss": 0.2626,
+      "step": 102250
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.640979100927517e-06,
+      "loss": 0.267,
+      "step": 102255
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.63873148774755e-06,
+      "loss": 0.2714,
+      "step": 102260
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.636484074023542e-06,
+      "loss": 0.2546,
+      "step": 102265
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.634236859784663e-06,
+      "loss": 0.2572,
+      "step": 102270
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.631989845060094e-06,
+      "loss": 0.2706,
+      "step": 102275
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.629743029878995e-06,
+      "loss": 0.2668,
+      "step": 102280
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.627496414270568e-06,
+      "loss": 0.2564,
+      "step": 102285
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.625249998263958e-06,
+      "loss": 0.245,
+      "step": 102290
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.62300378188835e-06,
+      "loss": 0.2856,
+      "step": 102295
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.6207577651729e-06,
+      "loss": 0.2775,
+      "step": 102300
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.618511948146783e-06,
+      "loss": 0.281,
+      "step": 102305
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.61626633083915e-06,
+      "loss": 0.2514,
+      "step": 102310
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.614020913279161e-06,
+      "loss": 0.2598,
+      "step": 102315
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.611775695495961e-06,
+      "loss": 0.2906,
+      "step": 102320
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.609530677518724e-06,
+      "loss": 0.2555,
+      "step": 102325
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.607285859376587e-06,
+      "loss": 0.2633,
+      "step": 102330
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.605041241098697e-06,
+      "loss": 0.2663,
+      "step": 102335
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.602796822714194e-06,
+      "loss": 0.2656,
+      "step": 102340
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.600552604252237e-06,
+      "loss": 0.257,
+      "step": 102345
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.598308585741944e-06,
+      "loss": 0.2543,
+      "step": 102350
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.596064767212477e-06,
+      "loss": 0.2677,
+      "step": 102355
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.593821148692945e-06,
+      "loss": 0.2726,
+      "step": 102360
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.591577730212503e-06,
+      "loss": 0.2637,
+      "step": 102365
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.58933451180027e-06,
+      "loss": 0.2391,
+      "step": 102370
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.58709149348536e-06,
+      "loss": 0.2445,
+      "step": 102375
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.58484867529692e-06,
+      "loss": 0.2422,
+      "step": 102380
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.582606057264062e-06,
+      "loss": 0.2746,
+      "step": 102385
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.580363639415905e-06,
+      "loss": 0.2527,
+      "step": 102390
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.578121421781552e-06,
+      "loss": 0.2454,
+      "step": 102395
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.57587940439014e-06,
+      "loss": 0.2562,
+      "step": 102400
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.573637587270759e-06,
+      "loss": 0.2745,
+      "step": 102405
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.571395970452537e-06,
+      "loss": 0.2321,
+      "step": 102410
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.569154553964563e-06,
+      "loss": 0.2629,
+      "step": 102415
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.566913337835956e-06,
+      "loss": 0.2616,
+      "step": 102420
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.564672322095808e-06,
+      "loss": 0.2612,
+      "step": 102425
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.562431506773217e-06,
+      "loss": 0.2632,
+      "step": 102430
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.560190891897272e-06,
+      "loss": 0.2614,
+      "step": 102435
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.55795047749708e-06,
+      "loss": 0.244,
+      "step": 102440
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.555710263601728e-06,
+      "loss": 0.2503,
+      "step": 102445
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.553470250240296e-06,
+      "loss": 0.2715,
+      "step": 102450
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.551230437441866e-06,
+      "loss": 0.2537,
+      "step": 102455
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.548990825235527e-06,
+      "loss": 0.2516,
+      "step": 102460
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.546751413650368e-06,
+      "loss": 0.25,
+      "step": 102465
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.544512202715458e-06,
+      "loss": 0.2575,
+      "step": 102470
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.542273192459863e-06,
+      "loss": 0.2565,
+      "step": 102475
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.540034382912673e-06,
+      "loss": 0.2556,
+      "step": 102480
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.537795774102948e-06,
+      "loss": 0.2792,
+      "step": 102485
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.535557366059747e-06,
+      "loss": 0.236,
+      "step": 102490
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.533319158812148e-06,
+      "loss": 0.2493,
+      "step": 102495
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.53108115238921e-06,
+      "loss": 0.2487,
+      "step": 102500
+    },
+    {
+      "epoch": 3.61,
+      "eval_loss": 0.2519163489341736,
+      "eval_runtime": 10.5511,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 102500
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.528843346819989e-06,
+      "loss": 0.2756,
+      "step": 102505
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.526605742133529e-06,
+      "loss": 0.2386,
+      "step": 102510
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.524368338358897e-06,
+      "loss": 0.2588,
+      "step": 102515
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.522131135525153e-06,
+      "loss": 0.2416,
+      "step": 102520
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.519894133661337e-06,
+      "loss": 0.2571,
+      "step": 102525
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.517657332796485e-06,
+      "loss": 0.2645,
+      "step": 102530
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.515420732959654e-06,
+      "loss": 0.2454,
+      "step": 102535
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.513184334179883e-06,
+      "loss": 0.2603,
+      "step": 102540
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.510948136486209e-06,
+      "loss": 0.2355,
+      "step": 102545
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.508712139907654e-06,
+      "loss": 0.2529,
+      "step": 102550
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.506476344473272e-06,
+      "loss": 0.2456,
+      "step": 102555
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.504240750212088e-06,
+      "loss": 0.2642,
+      "step": 102560
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.502005357153113e-06,
+      "loss": 0.2516,
+      "step": 102565
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.499770165325386e-06,
+      "loss": 0.2802,
+      "step": 102570
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.497535174757934e-06,
+      "loss": 0.2455,
+      "step": 102575
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.495300385479777e-06,
+      "loss": 0.2484,
+      "step": 102580
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.493065797519923e-06,
+      "loss": 0.2567,
+      "step": 102585
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.490831410907382e-06,
+      "loss": 0.2646,
+      "step": 102590
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.488597225671184e-06,
+      "loss": 0.2687,
+      "step": 102595
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.486363241840326e-06,
+      "loss": 0.2501,
+      "step": 102600
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.48412945944381e-06,
+      "loss": 0.2685,
+      "step": 102605
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.481895878510658e-06,
+      "loss": 0.2516,
+      "step": 102610
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.47966249906986e-06,
+      "loss": 0.264,
+      "step": 102615
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.477429321150409e-06,
+      "loss": 0.248,
+      "step": 102620
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.475196344781315e-06,
+      "loss": 0.2655,
+      "step": 102625
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.472963569991555e-06,
+      "loss": 0.2697,
+      "step": 102630
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.470730996810142e-06,
+      "loss": 0.2444,
+      "step": 102635
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.468498625266051e-06,
+      "loss": 0.2492,
+      "step": 102640
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.466266455388262e-06,
+      "loss": 0.2685,
+      "step": 102645
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.464034487205775e-06,
+      "loss": 0.2759,
+      "step": 102650
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.461802720747562e-06,
+      "loss": 0.2395,
+      "step": 102655
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.459571156042599e-06,
+      "loss": 0.2341,
+      "step": 102660
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.457339793119858e-06,
+      "loss": 0.2506,
+      "step": 102665
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.45510863200832e-06,
+      "loss": 0.2364,
+      "step": 102670
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.452877672736946e-06,
+      "loss": 0.2614,
+      "step": 102675
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.450646915334718e-06,
+      "loss": 0.2472,
+      "step": 102680
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.448416359830583e-06,
+      "loss": 0.2616,
+      "step": 102685
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.446186006253523e-06,
+      "loss": 0.258,
+      "step": 102690
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.443955854632488e-06,
+      "loss": 0.2586,
+      "step": 102695
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.441725904996431e-06,
+      "loss": 0.2646,
+      "step": 102700
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.439496157374303e-06,
+      "loss": 0.2476,
+      "step": 102705
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.437266611795073e-06,
+      "loss": 0.2542,
+      "step": 102710
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.435037268287677e-06,
+      "loss": 0.2702,
+      "step": 102715
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.432808126881057e-06,
+      "loss": 0.2632,
+      "step": 102720
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.430579187604174e-06,
+      "loss": 0.2578,
+      "step": 102725
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.428350450485951e-06,
+      "loss": 0.251,
+      "step": 102730
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.426121915555341e-06,
+      "loss": 0.2416,
+      "step": 102735
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.423893582841268e-06,
+      "loss": 0.2561,
+      "step": 102740
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.42166545237268e-06,
+      "loss": 0.2564,
+      "step": 102745
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.419437524178499e-06,
+      "loss": 0.2659,
+      "step": 102750
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.417209798287654e-06,
+      "loss": 0.2413,
+      "step": 102755
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.414982274729061e-06,
+      "loss": 0.2587,
+      "step": 102760
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.412754953531663e-06,
+      "loss": 0.2574,
+      "step": 102765
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.410527834724369e-06,
+      "loss": 0.2834,
+      "step": 102770
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.408300918336097e-06,
+      "loss": 0.248,
+      "step": 102775
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.406074204395754e-06,
+      "loss": 0.2823,
+      "step": 102780
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.403847692932268e-06,
+      "loss": 0.2686,
+      "step": 102785
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.401621383974532e-06,
+      "loss": 0.2514,
+      "step": 102790
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.399395277551473e-06,
+      "loss": 0.2579,
+      "step": 102795
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.397169373691977e-06,
+      "loss": 0.2614,
+      "step": 102800
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.394943672424964e-06,
+      "loss": 0.2541,
+      "step": 102805
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.392718173779319e-06,
+      "loss": 0.2648,
+      "step": 102810
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.390492877783946e-06,
+      "loss": 0.2603,
+      "step": 102815
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.388267784467725e-06,
+      "loss": 0.2555,
+      "step": 102820
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.38604289385957e-06,
+      "loss": 0.2853,
+      "step": 102825
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.383818205988356e-06,
+      "loss": 0.2637,
+      "step": 102830
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.381593720882961e-06,
+      "loss": 0.2786,
+      "step": 102835
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.379369438572292e-06,
+      "loss": 0.2633,
+      "step": 102840
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.377145359085205e-06,
+      "loss": 0.2493,
+      "step": 102845
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.374921482450597e-06,
+      "loss": 0.2476,
+      "step": 102850
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.372697808697328e-06,
+      "loss": 0.2641,
+      "step": 102855
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.37047433785429e-06,
+      "loss": 0.2473,
+      "step": 102860
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.36825106995034e-06,
+      "loss": 0.2858,
+      "step": 102865
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.366028005014348e-06,
+      "loss": 0.2718,
+      "step": 102870
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.363805143075175e-06,
+      "loss": 0.2565,
+      "step": 102875
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.36158248416169e-06,
+      "loss": 0.2571,
+      "step": 102880
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.359360028302755e-06,
+      "loss": 0.2721,
+      "step": 102885
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.357137775527219e-06,
+      "loss": 0.2719,
+      "step": 102890
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.354915725863936e-06,
+      "loss": 0.2675,
+      "step": 102895
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.352693879341756e-06,
+      "loss": 0.2651,
+      "step": 102900
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.350472235989547e-06,
+      "loss": 0.2441,
+      "step": 102905
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.348250795836141e-06,
+      "loss": 0.251,
+      "step": 102910
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.346029558910375e-06,
+      "loss": 0.2562,
+      "step": 102915
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.343808525241108e-06,
+      "loss": 0.2616,
+      "step": 102920
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.341587694857168e-06,
+      "loss": 0.2518,
+      "step": 102925
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.339367067787393e-06,
+      "loss": 0.2519,
+      "step": 102930
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.33714664406061e-06,
+      "loss": 0.2531,
+      "step": 102935
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.334926423705662e-06,
+      "loss": 0.2655,
+      "step": 102940
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.33270640675137e-06,
+      "loss": 0.2587,
+      "step": 102945
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.330486593226553e-06,
+      "loss": 0.2436,
+      "step": 102950
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.32826698316004e-06,
+      "loss": 0.2479,
+      "step": 102955
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.32604757658066e-06,
+      "loss": 0.2617,
+      "step": 102960
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.323828373517224e-06,
+      "loss": 0.2575,
+      "step": 102965
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.321609373998538e-06,
+      "loss": 0.276,
+      "step": 102970
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.319390578053427e-06,
+      "loss": 0.2469,
+      "step": 102975
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.3171719857107e-06,
+      "loss": 0.2463,
+      "step": 102980
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.314953596999156e-06,
+      "loss": 0.2471,
+      "step": 102985
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.312735411947597e-06,
+      "loss": 0.2415,
+      "step": 102990
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.310517430584836e-06,
+      "loss": 0.2512,
+      "step": 102995
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.308299652939666e-06,
+      "loss": 0.2563,
+      "step": 103000
+    },
+    {
+      "epoch": 3.62,
+      "eval_loss": 0.25192898511886597,
+      "eval_runtime": 10.5381,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 103000
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.306082079040879e-06,
+      "loss": 0.2784,
+      "step": 103005
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.30386470891727e-06,
+      "loss": 0.2625,
+      "step": 103010
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.301647542597643e-06,
+      "loss": 0.2697,
+      "step": 103015
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.299430580110776e-06,
+      "loss": 0.2611,
+      "step": 103020
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.29721382148546e-06,
+      "loss": 0.2729,
+      "step": 103025
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.29499726675046e-06,
+      "loss": 0.2328,
+      "step": 103030
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.292780915934585e-06,
+      "loss": 0.2722,
+      "step": 103035
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.290564769066592e-06,
+      "loss": 0.278,
+      "step": 103040
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.288348826175266e-06,
+      "loss": 0.2406,
+      "step": 103045
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.28613308728937e-06,
+      "loss": 0.2726,
+      "step": 103050
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.283917552437684e-06,
+      "loss": 0.2624,
+      "step": 103055
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.281702221648966e-06,
+      "loss": 0.2461,
+      "step": 103060
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.279487094951991e-06,
+      "loss": 0.2531,
+      "step": 103065
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.277272172375511e-06,
+      "loss": 0.2718,
+      "step": 103070
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.275057453948297e-06,
+      "loss": 0.2636,
+      "step": 103075
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.2728429396991e-06,
+      "loss": 0.2607,
+      "step": 103080
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.270628629656663e-06,
+      "loss": 0.2376,
+      "step": 103085
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.268414523849755e-06,
+      "loss": 0.2535,
+      "step": 103090
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.26620062230712e-06,
+      "loss": 0.2569,
+      "step": 103095
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.263986925057497e-06,
+      "loss": 0.2596,
+      "step": 103100
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.261773432129628e-06,
+      "loss": 0.2691,
+      "step": 103105
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.259560143552265e-06,
+      "loss": 0.2405,
+      "step": 103110
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.257347059354132e-06,
+      "loss": 0.288,
+      "step": 103115
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.255134179563985e-06,
+      "loss": 0.2593,
+      "step": 103120
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.252921504210533e-06,
+      "loss": 0.2573,
+      "step": 103125
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.250709033322524e-06,
+      "loss": 0.2445,
+      "step": 103130
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.248496766928683e-06,
+      "loss": 0.2501,
+      "step": 103135
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.246284705057728e-06,
+      "loss": 0.2482,
+      "step": 103140
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.244072847738377e-06,
+      "loss": 0.2378,
+      "step": 103145
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.241861194999365e-06,
+      "loss": 0.2562,
+      "step": 103150
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.2396497468694e-06,
+      "loss": 0.2551,
+      "step": 103155
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.237438503377196e-06,
+      "loss": 0.2423,
+      "step": 103160
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.235227464551457e-06,
+      "loss": 0.2807,
+      "step": 103165
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.233016630420898e-06,
+      "loss": 0.2741,
+      "step": 103170
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.230806001014238e-06,
+      "loss": 0.2461,
+      "step": 103175
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.228595576360172e-06,
+      "loss": 0.2494,
+      "step": 103180
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.226385356487388e-06,
+      "loss": 0.2675,
+      "step": 103185
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.2241753414246e-06,
+      "loss": 0.2761,
+      "step": 103190
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.221965531200504e-06,
+      "loss": 0.2476,
+      "step": 103195
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.219755925843776e-06,
+      "loss": 0.2674,
+      "step": 103200
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.217546525383128e-06,
+      "loss": 0.2556,
+      "step": 103205
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.215337329847238e-06,
+      "loss": 0.263,
+      "step": 103210
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.213128339264788e-06,
+      "loss": 0.2573,
+      "step": 103215
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.210919553664454e-06,
+      "loss": 0.237,
+      "step": 103220
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.208710973074932e-06,
+      "loss": 0.2578,
+      "step": 103225
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.206502597524886e-06,
+      "loss": 0.2596,
+      "step": 103230
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.204294427043e-06,
+      "loss": 0.2492,
+      "step": 103235
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.202086461657935e-06,
+      "loss": 0.229,
+      "step": 103240
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.199878701398376e-06,
+      "loss": 0.2658,
+      "step": 103245
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.197671146292977e-06,
+      "loss": 0.2651,
+      "step": 103250
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.195463796370405e-06,
+      "loss": 0.2714,
+      "step": 103255
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.19325665165931e-06,
+      "loss": 0.2282,
+      "step": 103260
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.191049712188371e-06,
+      "loss": 0.2691,
+      "step": 103265
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.18884297798623e-06,
+      "loss": 0.2691,
+      "step": 103270
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.18663644908154e-06,
+      "loss": 0.252,
+      "step": 103275
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.184430125502958e-06,
+      "loss": 0.2643,
+      "step": 103280
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.182224007279122e-06,
+      "loss": 0.2822,
+      "step": 103285
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.18001809443869e-06,
+      "loss": 0.2773,
+      "step": 103290
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.177812387010295e-06,
+      "loss": 0.2452,
+      "step": 103295
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.175606885022575e-06,
+      "loss": 0.254,
+      "step": 103300
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.173401588504177e-06,
+      "loss": 0.278,
+      "step": 103305
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.171196497483731e-06,
+      "loss": 0.2877,
+      "step": 103310
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.168991611989855e-06,
+      "loss": 0.2475,
+      "step": 103315
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.166786932051203e-06,
+      "loss": 0.2479,
+      "step": 103320
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.164582457696384e-06,
+      "loss": 0.2354,
+      "step": 103325
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.162378188954027e-06,
+      "loss": 0.2533,
+      "step": 103330
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.160174125852745e-06,
+      "loss": 0.2438,
+      "step": 103335
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.157970268421162e-06,
+      "loss": 0.2633,
+      "step": 103340
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.155766616687902e-06,
+      "loss": 0.2715,
+      "step": 103345
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.153563170681573e-06,
+      "loss": 0.2678,
+      "step": 103350
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.151359930430772e-06,
+      "loss": 0.2665,
+      "step": 103355
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.149156895964128e-06,
+      "loss": 0.2473,
+      "step": 103360
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.146954067310237e-06,
+      "loss": 0.2512,
+      "step": 103365
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.144751444497698e-06,
+      "loss": 0.2683,
+      "step": 103370
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.142549027555103e-06,
+      "loss": 0.2704,
+      "step": 103375
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.140346816511069e-06,
+      "loss": 0.2578,
+      "step": 103380
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.138144811394178e-06,
+      "loss": 0.2615,
+      "step": 103385
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.135943012233017e-06,
+      "loss": 0.2751,
+      "step": 103390
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.13374141905618e-06,
+      "loss": 0.2699,
+      "step": 103395
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.131540031892261e-06,
+      "loss": 0.2361,
+      "step": 103400
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.12933885076984e-06,
+      "loss": 0.2633,
+      "step": 103405
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.127137875717492e-06,
+      "loss": 0.2574,
+      "step": 103410
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.124937106763793e-06,
+      "loss": 0.2722,
+      "step": 103415
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.12273654393733e-06,
+      "loss": 0.2493,
+      "step": 103420
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.120536187266667e-06,
+      "loss": 0.2745,
+      "step": 103425
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.118336036780373e-06,
+      "loss": 0.263,
+      "step": 103430
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.116136092507024e-06,
+      "loss": 0.2741,
+      "step": 103435
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.113936354475183e-06,
+      "loss": 0.2627,
+      "step": 103440
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.1117368227134e-06,
+      "loss": 0.2345,
+      "step": 103445
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.109537497250242e-06,
+      "loss": 0.2622,
+      "step": 103450
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.107338378114277e-06,
+      "loss": 0.2581,
+      "step": 103455
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.105139465334048e-06,
+      "loss": 0.2879,
+      "step": 103460
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.102940758938109e-06,
+      "loss": 0.2679,
+      "step": 103465
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.100742258955e-06,
+      "loss": 0.2528,
+      "step": 103470
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.098543965413281e-06,
+      "loss": 0.2925,
+      "step": 103475
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.096345878341492e-06,
+      "loss": 0.2465,
+      "step": 103480
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.094147997768168e-06,
+      "loss": 0.271,
+      "step": 103485
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.091950323721842e-06,
+      "loss": 0.2876,
+      "step": 103490
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.089752856231069e-06,
+      "loss": 0.2557,
+      "step": 103495
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.08755559532436e-06,
+      "loss": 0.2575,
+      "step": 103500
+    },
+    {
+      "epoch": 3.64,
+      "eval_loss": 0.2520705461502075,
+      "eval_runtime": 10.549,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 103500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.085358541030262e-06,
+      "loss": 0.2715,
+      "step": 103505
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.083161693377288e-06,
+      "loss": 0.2492,
+      "step": 103510
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.080965052393974e-06,
+      "loss": 0.2546,
+      "step": 103515
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.078768618108843e-06,
+      "loss": 0.2508,
+      "step": 103520
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.07657239055041e-06,
+      "loss": 0.2488,
+      "step": 103525
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.074376369747176e-06,
+      "loss": 0.2609,
+      "step": 103530
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.072180555727683e-06,
+      "loss": 0.2645,
+      "step": 103535
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.069984948520428e-06,
+      "loss": 0.2386,
+      "step": 103540
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.06778954815391e-06,
+      "loss": 0.2696,
+      "step": 103545
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.065594354656653e-06,
+      "loss": 0.2695,
+      "step": 103550
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.063399368057143e-06,
+      "loss": 0.2459,
+      "step": 103555
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.061204588383901e-06,
+      "loss": 0.2661,
+      "step": 103560
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.059010015665403e-06,
+      "loss": 0.275,
+      "step": 103565
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.056815649930165e-06,
+      "loss": 0.266,
+      "step": 103570
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.054621491206667e-06,
+      "loss": 0.2611,
+      "step": 103575
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.052427539523401e-06,
+      "loss": 0.249,
+      "step": 103580
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.050233794908844e-06,
+      "loss": 0.2448,
+      "step": 103585
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.048040257391501e-06,
+      "loss": 0.2596,
+      "step": 103590
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.04584692699984e-06,
+      "loss": 0.2596,
+      "step": 103595
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.043653803762345e-06,
+      "loss": 0.2496,
+      "step": 103600
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.041460887707478e-06,
+      "loss": 0.2707,
+      "step": 103605
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.039268178863727e-06,
+      "loss": 0.2813,
+      "step": 103610
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.037075677259569e-06,
+      "loss": 0.257,
+      "step": 103615
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.034883382923459e-06,
+      "loss": 0.2434,
+      "step": 103620
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.032691295883863e-06,
+      "loss": 0.2592,
+      "step": 103625
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.030499416169256e-06,
+      "loss": 0.2716,
+      "step": 103630
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.028307743808087e-06,
+      "loss": 0.2672,
+      "step": 103635
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.026116278828816e-06,
+      "loss": 0.2589,
+      "step": 103640
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.023925021259894e-06,
+      "loss": 0.244,
+      "step": 103645
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.021733971129782e-06,
+      "loss": 0.2679,
+      "step": 103650
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.019543128466922e-06,
+      "loss": 0.2594,
+      "step": 103655
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.017352493299757e-06,
+      "loss": 0.2675,
+      "step": 103660
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.01560013458205e-06,
+      "loss": 0.2589,
+      "step": 103665
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.013409872978834e-06,
+      "loss": 0.2636,
+      "step": 103670
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.011219818950945e-06,
+      "loss": 0.2685,
+      "step": 103675
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.009029972526839e-06,
+      "loss": 0.2679,
+      "step": 103680
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.00684033373494e-06,
+      "loss": 0.2409,
+      "step": 103685
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.004650902603679e-06,
+      "loss": 0.2331,
+      "step": 103690
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.002461679161475e-06,
+      "loss": 0.2469,
+      "step": 103695
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.000272663436765e-06,
+      "loss": 0.2537,
+      "step": 103700
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.998083855457979e-06,
+      "loss": 0.247,
+      "step": 103705
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.995895255253529e-06,
+      "loss": 0.2679,
+      "step": 103710
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.99370686285182e-06,
+      "loss": 0.2445,
+      "step": 103715
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.991518678281292e-06,
+      "loss": 0.2439,
+      "step": 103720
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.989330701570339e-06,
+      "loss": 0.2615,
+      "step": 103725
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.98714293274737e-06,
+      "loss": 0.2648,
+      "step": 103730
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.984955371840806e-06,
+      "loss": 0.2536,
+      "step": 103735
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.98276801887904e-06,
+      "loss": 0.2468,
+      "step": 103740
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.980580873890477e-06,
+      "loss": 0.2553,
+      "step": 103745
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.978393936903507e-06,
+      "loss": 0.2512,
+      "step": 103750
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.97620720794653e-06,
+      "loss": 0.2637,
+      "step": 103755
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.974020687047955e-06,
+      "loss": 0.2473,
+      "step": 103760
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.971834374236155e-06,
+      "loss": 0.2655,
+      "step": 103765
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.969648269539515e-06,
+      "loss": 0.2595,
+      "step": 103770
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.96746237298644e-06,
+      "loss": 0.2466,
+      "step": 103775
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.965276684605297e-06,
+      "loss": 0.2621,
+      "step": 103780
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.96309120442447e-06,
+      "loss": 0.2389,
+      "step": 103785
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.960905932472327e-06,
+      "loss": 0.2661,
+      "step": 103790
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.95872086877726e-06,
+      "loss": 0.274,
+      "step": 103795
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.956536013367628e-06,
+      "loss": 0.2816,
+      "step": 103800
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.954351366271796e-06,
+      "loss": 0.2454,
+      "step": 103805
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.952166927518146e-06,
+      "loss": 0.2581,
+      "step": 103810
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.949982697135024e-06,
+      "loss": 0.2761,
+      "step": 103815
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.947798675150806e-06,
+      "loss": 0.2835,
+      "step": 103820
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.945614861593843e-06,
+      "loss": 0.2625,
+      "step": 103825
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.943431256492484e-06,
+      "loss": 0.2508,
+      "step": 103830
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.941247859875095e-06,
+      "loss": 0.2554,
+      "step": 103835
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.939064671770023e-06,
+      "loss": 0.2452,
+      "step": 103840
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.936881692205601e-06,
+      "loss": 0.2538,
+      "step": 103845
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.934698921210195e-06,
+      "loss": 0.2836,
+      "step": 103850
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.932516358812132e-06,
+      "loss": 0.2783,
+      "step": 103855
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.93033400503975e-06,
+      "loss": 0.2501,
+      "step": 103860
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.928151859921397e-06,
+      "loss": 0.2647,
+      "step": 103865
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.925969923485391e-06,
+      "loss": 0.2741,
+      "step": 103870
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.923788195760086e-06,
+      "loss": 0.2295,
+      "step": 103875
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.921606676773791e-06,
+      "loss": 0.2456,
+      "step": 103880
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 8.919425366554832e-06,
+      "loss": 0.2479,
+      "step": 103885
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.917244265131543e-06,
+      "loss": 0.2524,
+      "step": 103890
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.915063372532242e-06,
+      "loss": 0.2658,
+      "step": 103895
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.91288268878524e-06,
+      "loss": 0.28,
+      "step": 103900
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.910702213918848e-06,
+      "loss": 0.2514,
+      "step": 103905
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.90852194796139e-06,
+      "loss": 0.2709,
+      "step": 103910
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.906341890941164e-06,
+      "loss": 0.2609,
+      "step": 103915
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.904162042886487e-06,
+      "loss": 0.2631,
+      "step": 103920
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.901982403825654e-06,
+      "loss": 0.2406,
+      "step": 103925
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.899802973786977e-06,
+      "loss": 0.2491,
+      "step": 103930
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.89762375279875e-06,
+      "loss": 0.2554,
+      "step": 103935
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.895444740889267e-06,
+      "loss": 0.2572,
+      "step": 103940
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.893265938086808e-06,
+      "loss": 0.2622,
+      "step": 103945
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.891087344419688e-06,
+      "loss": 0.2487,
+      "step": 103950
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.888908959916182e-06,
+      "loss": 0.2296,
+      "step": 103955
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.886730784604565e-06,
+      "loss": 0.2806,
+      "step": 103960
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.884552818513143e-06,
+      "loss": 0.2633,
+      "step": 103965
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.882375061670178e-06,
+      "loss": 0.2681,
+      "step": 103970
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.880197514103944e-06,
+      "loss": 0.2587,
+      "step": 103975
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.878020175842722e-06,
+      "loss": 0.2471,
+      "step": 103980
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.875843046914792e-06,
+      "loss": 0.2488,
+      "step": 103985
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.873666127348412e-06,
+      "loss": 0.258,
+      "step": 103990
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.871489417171853e-06,
+      "loss": 0.2769,
+      "step": 103995
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.869312916413363e-06,
+      "loss": 0.2779,
+      "step": 104000
+    },
+    {
+      "epoch": 3.66,
+      "eval_loss": 0.25182366371154785,
+      "eval_runtime": 10.5433,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 104000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.867136625101225e-06,
+      "loss": 0.2541,
+      "step": 104005
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.864960543263684e-06,
+      "loss": 0.2648,
+      "step": 104010
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.862784670928998e-06,
+      "loss": 0.2607,
+      "step": 104015
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.860609008125408e-06,
+      "loss": 0.2858,
+      "step": 104020
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.858433554881185e-06,
+      "loss": 0.2665,
+      "step": 104025
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.856258311224556e-06,
+      "loss": 0.2776,
+      "step": 104030
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.85408327718378e-06,
+      "loss": 0.2577,
+      "step": 104035
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.851908452787086e-06,
+      "loss": 0.2582,
+      "step": 104040
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.849733838062724e-06,
+      "loss": 0.2478,
+      "step": 104045
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.847559433038927e-06,
+      "loss": 0.2542,
+      "step": 104050
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.845385237743924e-06,
+      "loss": 0.2527,
+      "step": 104055
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.843211252205944e-06,
+      "loss": 0.2481,
+      "step": 104060
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.84103747645322e-06,
+      "loss": 0.2688,
+      "step": 104065
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.83886391051398e-06,
+      "loss": 0.2292,
+      "step": 104070
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.836690554416433e-06,
+      "loss": 0.2643,
+      "step": 104075
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.834517408188814e-06,
+      "loss": 0.2564,
+      "step": 104080
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.832344471859324e-06,
+      "loss": 0.2784,
+      "step": 104085
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.830171745456195e-06,
+      "loss": 0.2447,
+      "step": 104090
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.827999229007622e-06,
+      "loss": 0.2457,
+      "step": 104095
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.825826922541827e-06,
+      "loss": 0.2656,
+      "step": 104100
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.823654826087014e-06,
+      "loss": 0.2321,
+      "step": 104105
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.82148293967138e-06,
+      "loss": 0.251,
+      "step": 104110
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.819311263323121e-06,
+      "loss": 0.27,
+      "step": 104115
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.817139797070447e-06,
+      "loss": 0.2412,
+      "step": 104120
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.814968540941551e-06,
+      "loss": 0.2398,
+      "step": 104125
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.812797494964622e-06,
+      "loss": 0.2572,
+      "step": 104130
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.810626659167842e-06,
+      "loss": 0.2635,
+      "step": 104135
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.808456033579405e-06,
+      "loss": 0.2611,
+      "step": 104140
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.806285618227503e-06,
+      "loss": 0.2458,
+      "step": 104145
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.80411541314031e-06,
+      "loss": 0.2607,
+      "step": 104150
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.801945418345997e-06,
+      "loss": 0.2331,
+      "step": 104155
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.799775633872756e-06,
+      "loss": 0.2686,
+      "step": 104160
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.797606059748754e-06,
+      "loss": 0.2482,
+      "step": 104165
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 8.795436696002157e-06,
+      "loss": 0.2513,
+      "step": 104170
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.79326754266113e-06,
+      "loss": 0.2615,
+      "step": 104175
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.791098599753848e-06,
+      "loss": 0.2588,
+      "step": 104180
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.78892986730847e-06,
+      "loss": 0.2491,
+      "step": 104185
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.786761345353148e-06,
+      "loss": 0.2622,
+      "step": 104190
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.784593033916041e-06,
+      "loss": 0.2478,
+      "step": 104195
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.782424933025317e-06,
+      "loss": 0.2468,
+      "step": 104200
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.78025704270912e-06,
+      "loss": 0.2527,
+      "step": 104205
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.778089362995585e-06,
+      "loss": 0.2665,
+      "step": 104210
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.775921893912878e-06,
+      "loss": 0.2505,
+      "step": 104215
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.773754635489135e-06,
+      "loss": 0.2476,
+      "step": 104220
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.771587587752491e-06,
+      "loss": 0.2766,
+      "step": 104225
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.76942075073108e-06,
+      "loss": 0.2894,
+      "step": 104230
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.767254124453052e-06,
+      "loss": 0.2504,
+      "step": 104235
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.765087708946534e-06,
+      "loss": 0.2704,
+      "step": 104240
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.762921504239643e-06,
+      "loss": 0.2435,
+      "step": 104245
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.760755510360525e-06,
+      "loss": 0.2439,
+      "step": 104250
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.758589727337286e-06,
+      "loss": 0.2392,
+      "step": 104255
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.756424155198064e-06,
+      "loss": 0.2783,
+      "step": 104260
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.754258793970968e-06,
+      "loss": 0.2382,
+      "step": 104265
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.752093643684109e-06,
+      "loss": 0.258,
+      "step": 104270
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.749928704365611e-06,
+      "loss": 0.2632,
+      "step": 104275
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.747763976043585e-06,
+      "loss": 0.2571,
+      "step": 104280
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.74559945874613e-06,
+      "loss": 0.2561,
+      "step": 104285
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.743435152501344e-06,
+      "loss": 0.2723,
+      "step": 104290
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.741271057337352e-06,
+      "loss": 0.2665,
+      "step": 104295
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.73910717328223e-06,
+      "loss": 0.2651,
+      "step": 104300
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.736943500364092e-06,
+      "loss": 0.2573,
+      "step": 104305
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.734780038611018e-06,
+      "loss": 0.2369,
+      "step": 104310
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.732616788051113e-06,
+      "loss": 0.2747,
+      "step": 104315
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.730453748712458e-06,
+      "loss": 0.2685,
+      "step": 104320
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.728290920623133e-06,
+      "loss": 0.2375,
+      "step": 104325
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.726128303811234e-06,
+      "loss": 0.2635,
+      "step": 104330
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.723965898304833e-06,
+      "loss": 0.2421,
+      "step": 104335
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.721803704132011e-06,
+      "loss": 0.2507,
+      "step": 104340
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.719641721320831e-06,
+      "loss": 0.2881,
+      "step": 104345
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.717479949899381e-06,
+      "loss": 0.2453,
+      "step": 104350
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.715318389895715e-06,
+      "loss": 0.2716,
+      "step": 104355
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.713157041337919e-06,
+      "loss": 0.2413,
+      "step": 104360
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.710995904254036e-06,
+      "loss": 0.2537,
+      "step": 104365
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.708834978672142e-06,
+      "loss": 0.2658,
+      "step": 104370
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.70667426462029e-06,
+      "loss": 0.2662,
+      "step": 104375
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.704513762126537e-06,
+      "loss": 0.2634,
+      "step": 104380
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.702353471218924e-06,
+      "loss": 0.2611,
+      "step": 104385
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.70019339192552e-06,
+      "loss": 0.2609,
+      "step": 104390
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.698033524274362e-06,
+      "loss": 0.2498,
+      "step": 104395
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.695873868293495e-06,
+      "loss": 0.2725,
+      "step": 104400
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.693714424010954e-06,
+      "loss": 0.2618,
+      "step": 104405
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.691555191454793e-06,
+      "loss": 0.2633,
+      "step": 104410
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.68939617065303e-06,
+      "loss": 0.2531,
+      "step": 104415
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.687237361633719e-06,
+      "loss": 0.2509,
+      "step": 104420
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.68507876442487e-06,
+      "loss": 0.2635,
+      "step": 104425
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.682920379054533e-06,
+      "loss": 0.2755,
+      "step": 104430
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.680762205550718e-06,
+      "loss": 0.2635,
+      "step": 104435
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.678604243941446e-06,
+      "loss": 0.2531,
+      "step": 104440
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.676446494254748e-06,
+      "loss": 0.2754,
+      "step": 104445
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 8.674288956518633e-06,
+      "loss": 0.2745,
+      "step": 104450
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.672131630761119e-06,
+      "loss": 0.2658,
+      "step": 104455
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.669974517010207e-06,
+      "loss": 0.2687,
+      "step": 104460
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.667817615293922e-06,
+      "loss": 0.2609,
+      "step": 104465
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.665660925640254e-06,
+      "loss": 0.246,
+      "step": 104470
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.663504448077222e-06,
+      "loss": 0.2521,
+      "step": 104475
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.661348182632812e-06,
+      "loss": 0.2602,
+      "step": 104480
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.659192129335034e-06,
+      "loss": 0.2542,
+      "step": 104485
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.657036288211878e-06,
+      "loss": 0.2705,
+      "step": 104490
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.654880659291335e-06,
+      "loss": 0.2608,
+      "step": 104495
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.652725242601386e-06,
+      "loss": 0.256,
+      "step": 104500
+    },
+    {
+      "epoch": 3.68,
+      "eval_loss": 0.2519655227661133,
+      "eval_runtime": 10.5307,
+      "eval_samples_per_second": 9.496,
+      "eval_steps_per_second": 9.496,
+      "step": 104500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.650570038170037e-06,
+      "loss": 0.2585,
+      "step": 104505
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.64841504602526e-06,
+      "loss": 0.2414,
+      "step": 104510
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.646260266195036e-06,
+      "loss": 0.2293,
+      "step": 104515
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.644105698707339e-06,
+      "loss": 0.2631,
+      "step": 104520
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.641951343590149e-06,
+      "loss": 0.2891,
+      "step": 104525
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.63979720087145e-06,
+      "loss": 0.2713,
+      "step": 104530
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.637643270579204e-06,
+      "loss": 0.2612,
+      "step": 104535
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.635489552741366e-06,
+      "loss": 0.241,
+      "step": 104540
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.633336047385921e-06,
+      "loss": 0.2513,
+      "step": 104545
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.63118275454082e-06,
+      "loss": 0.2451,
+      "step": 104550
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.629029674234015e-06,
+      "loss": 0.2679,
+      "step": 104555
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.62687680649348e-06,
+      "loss": 0.2696,
+      "step": 104560
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.62472415134716e-06,
+      "loss": 0.2559,
+      "step": 104565
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.622571708823007e-06,
+      "loss": 0.2507,
+      "step": 104570
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.620419478948957e-06,
+      "loss": 0.2527,
+      "step": 104575
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.618267461752965e-06,
+      "loss": 0.2299,
+      "step": 104580
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.616115657262982e-06,
+      "loss": 0.2722,
+      "step": 104585
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.61396406550694e-06,
+      "loss": 0.2582,
+      "step": 104590
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.611812686512769e-06,
+      "loss": 0.273,
+      "step": 104595
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.609661520308417e-06,
+      "loss": 0.2449,
+      "step": 104600
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.607510566921809e-06,
+      "loss": 0.2516,
+      "step": 104605
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.605359826380873e-06,
+      "loss": 0.2548,
+      "step": 104610
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.603209298713527e-06,
+      "loss": 0.2399,
+      "step": 104615
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.60105898394771e-06,
+      "loss": 0.232,
+      "step": 104620
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.598908882111336e-06,
+      "loss": 0.2542,
+      "step": 104625
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.596758993232312e-06,
+      "loss": 0.257,
+      "step": 104630
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.594609317338565e-06,
+      "loss": 0.2478,
+      "step": 104635
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.59245985445801e-06,
+      "loss": 0.255,
+      "step": 104640
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.59031060461855e-06,
+      "loss": 0.2627,
+      "step": 104645
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.588161567848095e-06,
+      "loss": 0.2289,
+      "step": 104650
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.586012744174538e-06,
+      "loss": 0.255,
+      "step": 104655
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.583864133625794e-06,
+      "loss": 0.2671,
+      "step": 104660
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.581715736229756e-06,
+      "loss": 0.2716,
+      "step": 104665
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.579567552014312e-06,
+      "loss": 0.2696,
+      "step": 104670
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.57741958100737e-06,
+      "loss": 0.2524,
+      "step": 104675
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.575271823236813e-06,
+      "loss": 0.265,
+      "step": 104680
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.573124278730516e-06,
+      "loss": 0.2618,
+      "step": 104685
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.570976947516385e-06,
+      "loss": 0.2743,
+      "step": 104690
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.568829829622283e-06,
+      "loss": 0.2527,
+      "step": 104695
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.566682925076106e-06,
+      "loss": 0.2584,
+      "step": 104700
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.564536233905718e-06,
+      "loss": 0.2596,
+      "step": 104705
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.562389756138988e-06,
+      "loss": 0.276,
+      "step": 104710
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.560243491803805e-06,
+      "loss": 0.2643,
+      "step": 104715
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.558097440928026e-06,
+      "loss": 0.2524,
+      "step": 104720
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.555951603539513e-06,
+      "loss": 0.2529,
+      "step": 104725
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.553805979666126e-06,
+      "loss": 0.2391,
+      "step": 104730
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 8.551660569335737e-06,
+      "loss": 0.2702,
+      "step": 104735
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.54951537257619e-06,
+      "loss": 0.2369,
+      "step": 104740
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.547370389415349e-06,
+      "loss": 0.2496,
+      "step": 104745
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.545225619881053e-06,
+      "loss": 0.2909,
+      "step": 104750
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.543081064001168e-06,
+      "loss": 0.2438,
+      "step": 104755
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.54093672180353e-06,
+      "loss": 0.2567,
+      "step": 104760
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.538792593315981e-06,
+      "loss": 0.2717,
+      "step": 104765
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.536648678566353e-06,
+      "loss": 0.2731,
+      "step": 104770
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.534504977582502e-06,
+      "loss": 0.2562,
+      "step": 104775
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.532361490392252e-06,
+      "loss": 0.2409,
+      "step": 104780
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.530218217023428e-06,
+      "loss": 0.2501,
+      "step": 104785
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.528075157503873e-06,
+      "loss": 0.254,
+      "step": 104790
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.525932311861399e-06,
+      "loss": 0.2703,
+      "step": 104795
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.523789680123844e-06,
+      "loss": 0.2657,
+      "step": 104800
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.521647262319016e-06,
+      "loss": 0.2476,
+      "step": 104805
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.519505058474747e-06,
+      "loss": 0.254,
+      "step": 104810
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.517363068618842e-06,
+      "loss": 0.2672,
+      "step": 104815
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.515221292779115e-06,
+      "loss": 0.2628,
+      "step": 104820
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.51307973098337e-06,
+      "loss": 0.2794,
+      "step": 104825
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.510938383259426e-06,
+      "loss": 0.267,
+      "step": 104830
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.508797249635078e-06,
+      "loss": 0.2309,
+      "step": 104835
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.506656330138132e-06,
+      "loss": 0.2529,
+      "step": 104840
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.504515624796378e-06,
+      "loss": 0.2605,
+      "step": 104845
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.502375133637621e-06,
+      "loss": 0.257,
+      "step": 104850
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.500234856689646e-06,
+      "loss": 0.2644,
+      "step": 104855
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.498094793980254e-06,
+      "loss": 0.247,
+      "step": 104860
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.495954945537218e-06,
+      "loss": 0.2457,
+      "step": 104865
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.49381531138834e-06,
+      "loss": 0.2419,
+      "step": 104870
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.491675891561393e-06,
+      "loss": 0.2537,
+      "step": 104875
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.489536686084152e-06,
+      "loss": 0.2623,
+      "step": 104880
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.487397694984388e-06,
+      "loss": 0.2529,
+      "step": 104885
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.485258918289895e-06,
+      "loss": 0.2585,
+      "step": 104890
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.483120356028429e-06,
+      "loss": 0.2664,
+      "step": 104895
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.48098200822775e-06,
+      "loss": 0.2778,
+      "step": 104900
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.478843874915646e-06,
+      "loss": 0.2379,
+      "step": 104905
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.476705956119854e-06,
+      "loss": 0.2932,
+      "step": 104910
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.474568251868157e-06,
+      "loss": 0.2642,
+      "step": 104915
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.472430762188289e-06,
+      "loss": 0.2508,
+      "step": 104920
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.470293487108024e-06,
+      "loss": 0.2744,
+      "step": 104925
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.468156426655108e-06,
+      "loss": 0.275,
+      "step": 104930
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.466019580857284e-06,
+      "loss": 0.2619,
+      "step": 104935
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.463882949742292e-06,
+      "loss": 0.2579,
+      "step": 104940
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.461746533337888e-06,
+      "loss": 0.2501,
+      "step": 104945
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.459610331671807e-06,
+      "loss": 0.2773,
+      "step": 104950
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.457474344771787e-06,
+      "loss": 0.2684,
+      "step": 104955
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.455338572665553e-06,
+      "loss": 0.256,
+      "step": 104960
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.453203015380842e-06,
+      "loss": 0.2457,
+      "step": 104965
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.451067672945395e-06,
+      "loss": 0.2383,
+      "step": 104970
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.448932545386928e-06,
+      "loss": 0.2567,
+      "step": 104975
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.446797632733156e-06,
+      "loss": 0.2452,
+      "step": 104980
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.444662935011815e-06,
+      "loss": 0.2533,
+      "step": 104985
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.442528452250614e-06,
+      "loss": 0.2415,
+      "step": 104990
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.440394184477274e-06,
+      "loss": 0.2855,
+      "step": 104995
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.438260131719489e-06,
+      "loss": 0.2472,
+      "step": 105000
+    },
+    {
+      "epoch": 3.69,
+      "eval_loss": 0.25183290243148804,
+      "eval_runtime": 10.5414,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 105000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.436126294004992e-06,
+      "loss": 0.2619,
+      "step": 105005
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.43399267136148e-06,
+      "loss": 0.2468,
+      "step": 105010
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.431859263816647e-06,
+      "loss": 0.2627,
+      "step": 105015
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 8.429726071398202e-06,
+      "loss": 0.2629,
+      "step": 105020
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.42759309413385e-06,
+      "loss": 0.2444,
+      "step": 105025
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.425460332051286e-06,
+      "loss": 0.2447,
+      "step": 105030
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.423327785178183e-06,
+      "loss": 0.2685,
+      "step": 105035
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.421195453542255e-06,
+      "loss": 0.2732,
+      "step": 105040
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.419063337171176e-06,
+      "loss": 0.2691,
+      "step": 105045
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.416931436092635e-06,
+      "loss": 0.2742,
+      "step": 105050
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.4147997503343e-06,
+      "loss": 0.2566,
+      "step": 105055
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.412668279923868e-06,
+      "loss": 0.248,
+      "step": 105060
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.41053702488901e-06,
+      "loss": 0.2583,
+      "step": 105065
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.408405985257387e-06,
+      "loss": 0.2467,
+      "step": 105070
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.406275161056675e-06,
+      "loss": 0.2544,
+      "step": 105075
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.404144552314555e-06,
+      "loss": 0.2703,
+      "step": 105080
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.402014159058683e-06,
+      "loss": 0.2647,
+      "step": 105085
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.399883981316713e-06,
+      "loss": 0.2664,
+      "step": 105090
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.397754019116307e-06,
+      "loss": 0.2453,
+      "step": 105095
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.395624272485128e-06,
+      "loss": 0.2837,
+      "step": 105100
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.393494741450828e-06,
+      "loss": 0.2511,
+      "step": 105105
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.391365426041054e-06,
+      "loss": 0.2686,
+      "step": 105110
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.389236326283445e-06,
+      "loss": 0.2467,
+      "step": 105115
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.387107442205663e-06,
+      "loss": 0.2503,
+      "step": 105120
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.384978773835336e-06,
+      "loss": 0.247,
+      "step": 105125
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.38285032120012e-06,
+      "loss": 0.2505,
+      "step": 105130
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.38072208432763e-06,
+      "loss": 0.2562,
+      "step": 105135
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.37859406324552e-06,
+      "loss": 0.2503,
+      "step": 105140
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.376466257981413e-06,
+      "loss": 0.259,
+      "step": 105145
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.374338668562928e-06,
+      "loss": 0.2572,
+      "step": 105150
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.372211295017707e-06,
+      "loss": 0.262,
+      "step": 105155
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.370084137373363e-06,
+      "loss": 0.2496,
+      "step": 105160
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.367957195657517e-06,
+      "loss": 0.2508,
+      "step": 105165
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.365830469897778e-06,
+      "loss": 0.2497,
+      "step": 105170
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.363703960121777e-06,
+      "loss": 0.2672,
+      "step": 105175
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.36157766635711e-06,
+      "loss": 0.2408,
+      "step": 105180
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.359451588631397e-06,
+      "loss": 0.2604,
+      "step": 105185
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.357325726972231e-06,
+      "loss": 0.2554,
+      "step": 105190
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.35520008140723e-06,
+      "loss": 0.2394,
+      "step": 105195
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.353074651963987e-06,
+      "loss": 0.239,
+      "step": 105200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.3509494386701e-06,
+      "loss": 0.2537,
+      "step": 105205
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.348824441553155e-06,
+      "loss": 0.2547,
+      "step": 105210
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.346699660640759e-06,
+      "loss": 0.2668,
+      "step": 105215
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.344575095960491e-06,
+      "loss": 0.2744,
+      "step": 105220
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.34245074753994e-06,
+      "loss": 0.2562,
+      "step": 105225
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.34032661540668e-06,
+      "loss": 0.2451,
+      "step": 105230
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.338202699588299e-06,
+      "loss": 0.2601,
+      "step": 105235
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.336079000112385e-06,
+      "loss": 0.244,
+      "step": 105240
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.333955517006496e-06,
+      "loss": 0.266,
+      "step": 105245
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.331832250298215e-06,
+      "loss": 0.2506,
+      "step": 105250
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.329709200015111e-06,
+      "loss": 0.269,
+      "step": 105255
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.327586366184747e-06,
+      "loss": 0.2455,
+      "step": 105260
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.325463748834677e-06,
+      "loss": 0.2603,
+      "step": 105265
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.323341347992481e-06,
+      "loss": 0.2528,
+      "step": 105270
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.321219163685706e-06,
+      "loss": 0.2537,
+      "step": 105275
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.31909719594191e-06,
+      "loss": 0.2582,
+      "step": 105280
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.316975444788632e-06,
+      "loss": 0.254,
+      "step": 105285
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.314853910253445e-06,
+      "loss": 0.2834,
+      "step": 105290
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.312732592363876e-06,
+      "loss": 0.2707,
+      "step": 105295
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.310611491147482e-06,
+      "loss": 0.2676,
+      "step": 105300
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 8.308490606631794e-06,
+      "loss": 0.2672,
+      "step": 105305
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.306369938844364e-06,
+      "loss": 0.2639,
+      "step": 105310
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.304249487812715e-06,
+      "loss": 0.2749,
+      "step": 105315
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.302129253564385e-06,
+      "loss": 0.2716,
+      "step": 105320
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.300009236126897e-06,
+      "loss": 0.2357,
+      "step": 105325
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.29788943552779e-06,
+      "loss": 0.2526,
+      "step": 105330
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.29576985179458e-06,
+      "loss": 0.2675,
+      "step": 105335
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.293650484954782e-06,
+      "loss": 0.2506,
+      "step": 105340
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.291531335035932e-06,
+      "loss": 0.2853,
+      "step": 105345
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.289412402065527e-06,
+      "loss": 0.2501,
+      "step": 105350
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.287293686071099e-06,
+      "loss": 0.2818,
+      "step": 105355
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.28517518708014e-06,
+      "loss": 0.2614,
+      "step": 105360
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.283056905120173e-06,
+      "loss": 0.2488,
+      "step": 105365
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.280938840218697e-06,
+      "loss": 0.2743,
+      "step": 105370
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.278820992403207e-06,
+      "loss": 0.2463,
+      "step": 105375
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.276703361701205e-06,
+      "loss": 0.2606,
+      "step": 105380
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.274585948140193e-06,
+      "loss": 0.262,
+      "step": 105385
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.272468751747661e-06,
+      "loss": 0.2712,
+      "step": 105390
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.270351772551096e-06,
+      "loss": 0.2589,
+      "step": 105395
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.268235010577985e-06,
+      "loss": 0.2597,
+      "step": 105400
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.266118465855813e-06,
+      "loss": 0.2744,
+      "step": 105405
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.264002138412071e-06,
+      "loss": 0.2636,
+      "step": 105410
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.261886028274232e-06,
+      "loss": 0.2698,
+      "step": 105415
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.259770135469765e-06,
+      "loss": 0.255,
+      "step": 105420
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.257654460026162e-06,
+      "loss": 0.2355,
+      "step": 105425
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.255539001970877e-06,
+      "loss": 0.2424,
+      "step": 105430
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.253423761331386e-06,
+      "loss": 0.2442,
+      "step": 105435
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.251308738135144e-06,
+      "loss": 0.2664,
+      "step": 105440
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.249193932409626e-06,
+      "loss": 0.262,
+      "step": 105445
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.247079344182284e-06,
+      "loss": 0.2698,
+      "step": 105450
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.244964973480573e-06,
+      "loss": 0.2772,
+      "step": 105455
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.242850820331946e-06,
+      "loss": 0.2527,
+      "step": 105460
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.240736884763866e-06,
+      "loss": 0.2283,
+      "step": 105465
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.238623166803775e-06,
+      "loss": 0.251,
+      "step": 105470
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.236509666479106e-06,
+      "loss": 0.2775,
+      "step": 105475
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.23439638381732e-06,
+      "loss": 0.2506,
+      "step": 105480
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.232283318845849e-06,
+      "loss": 0.2641,
+      "step": 105485
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.230170471592127e-06,
+      "loss": 0.2681,
+      "step": 105490
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.228057842083584e-06,
+      "loss": 0.2476,
+      "step": 105495
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.225945430347662e-06,
+      "loss": 0.2553,
+      "step": 105500
+    },
+    {
+      "epoch": 3.71,
+      "eval_loss": 0.2516160011291504,
+      "eval_runtime": 10.5476,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 105500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.223833236411784e-06,
+      "loss": 0.2675,
+      "step": 105505
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.221721260303373e-06,
+      "loss": 0.2427,
+      "step": 105510
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.219609502049846e-06,
+      "loss": 0.242,
+      "step": 105515
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.217497961678644e-06,
+      "loss": 0.2726,
+      "step": 105520
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.21538663921717e-06,
+      "loss": 0.2623,
+      "step": 105525
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.213275534692841e-06,
+      "loss": 0.2397,
+      "step": 105530
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.211164648133055e-06,
+      "loss": 0.2779,
+      "step": 105535
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.20905397956524e-06,
+      "loss": 0.2635,
+      "step": 105540
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.206943529016794e-06,
+      "loss": 0.2533,
+      "step": 105545
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.204833296515122e-06,
+      "loss": 0.2796,
+      "step": 105550
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.20272328208761e-06,
+      "loss": 0.2774,
+      "step": 105555
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.200613485761677e-06,
+      "loss": 0.2724,
+      "step": 105560
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.198503907564694e-06,
+      "loss": 0.2661,
+      "step": 105565
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.196394547524078e-06,
+      "loss": 0.2391,
+      "step": 105570
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.194285405667193e-06,
+      "loss": 0.2699,
+      "step": 105575
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.192176482021446e-06,
+      "loss": 0.2464,
+      "step": 105580
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.190067776614214e-06,
+      "loss": 0.2504,
+      "step": 105585
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 8.187959289472869e-06,
+      "loss": 0.256,
+      "step": 105590
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.185851020624788e-06,
+      "loss": 0.2432,
+      "step": 105595
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.183742970097357e-06,
+      "loss": 0.2638,
+      "step": 105600
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.181635137917942e-06,
+      "loss": 0.2533,
+      "step": 105605
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.179527524113904e-06,
+      "loss": 0.244,
+      "step": 105610
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.17742012871262e-06,
+      "loss": 0.2672,
+      "step": 105615
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.175312951741446e-06,
+      "loss": 0.2632,
+      "step": 105620
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.173205993227751e-06,
+      "loss": 0.259,
+      "step": 105625
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.17109925319888e-06,
+      "loss": 0.2579,
+      "step": 105630
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.168992731682202e-06,
+      "loss": 0.2476,
+      "step": 105635
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.166886428705058e-06,
+      "loss": 0.2701,
+      "step": 105640
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.164780344294806e-06,
+      "loss": 0.2374,
+      "step": 105645
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.162674478478774e-06,
+      "loss": 0.2599,
+      "step": 105650
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.16056883128433e-06,
+      "loss": 0.2632,
+      "step": 105655
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.158463402738798e-06,
+      "loss": 0.227,
+      "step": 105660
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.156358192869521e-06,
+      "loss": 0.2211,
+      "step": 105665
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.154253201703824e-06,
+      "loss": 0.2356,
+      "step": 105670
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.15214842926905e-06,
+      "loss": 0.2578,
+      "step": 105675
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.15004387559253e-06,
+      "loss": 0.2664,
+      "step": 105680
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.147939540701588e-06,
+      "loss": 0.2439,
+      "step": 105685
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.145835424623539e-06,
+      "loss": 0.2558,
+      "step": 105690
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.143731527385711e-06,
+      "loss": 0.2669,
+      "step": 105695
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.141627849015427e-06,
+      "loss": 0.2658,
+      "step": 105700
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.139524389539993e-06,
+      "loss": 0.2569,
+      "step": 105705
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.137421148986716e-06,
+      "loss": 0.251,
+      "step": 105710
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.135318127382917e-06,
+      "loss": 0.2741,
+      "step": 105715
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.133215324755903e-06,
+      "loss": 0.2686,
+      "step": 105720
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.131112741132962e-06,
+      "loss": 0.2404,
+      "step": 105725
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.129010376541412e-06,
+      "loss": 0.271,
+      "step": 105730
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.126908231008536e-06,
+      "loss": 0.2655,
+      "step": 105735
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.124806304561647e-06,
+      "loss": 0.2301,
+      "step": 105740
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.122704597228017e-06,
+      "loss": 0.2741,
+      "step": 105745
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.120603109034954e-06,
+      "loss": 0.2468,
+      "step": 105750
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.118501840009738e-06,
+      "loss": 0.269,
+      "step": 105755
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.116400790179648e-06,
+      "loss": 0.2589,
+      "step": 105760
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.114299959571959e-06,
+      "loss": 0.2432,
+      "step": 105765
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.112199348213965e-06,
+      "loss": 0.2401,
+      "step": 105770
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.110098956132934e-06,
+      "loss": 0.2524,
+      "step": 105775
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.107998783356128e-06,
+      "loss": 0.2679,
+      "step": 105780
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.105898829910833e-06,
+      "loss": 0.2746,
+      "step": 105785
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.1037990958243e-06,
+      "loss": 0.253,
+      "step": 105790
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.101699581123809e-06,
+      "loss": 0.2622,
+      "step": 105795
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.099600285836614e-06,
+      "loss": 0.2589,
+      "step": 105800
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.097501209989963e-06,
+      "loss": 0.2516,
+      "step": 105805
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.095402353611125e-06,
+      "loss": 0.2557,
+      "step": 105810
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.093303716727351e-06,
+      "loss": 0.2641,
+      "step": 105815
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.091205299365884e-06,
+      "loss": 0.2774,
+      "step": 105820
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.089107101553964e-06,
+      "loss": 0.2497,
+      "step": 105825
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.087009123318853e-06,
+      "loss": 0.2542,
+      "step": 105830
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.084911364687784e-06,
+      "loss": 0.2485,
+      "step": 105835
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.082813825687982e-06,
+      "loss": 0.2743,
+      "step": 105840
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.080716506346696e-06,
+      "loss": 0.2506,
+      "step": 105845
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.078619406691163e-06,
+      "loss": 0.2794,
+      "step": 105850
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.076522526748605e-06,
+      "loss": 0.261,
+      "step": 105855
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.074425866546242e-06,
+      "loss": 0.238,
+      "step": 105860
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.072329426111313e-06,
+      "loss": 0.2335,
+      "step": 105865
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.070233205471031e-06,
+      "loss": 0.2563,
+      "step": 105870
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 8.068137204652612e-06,
+      "loss": 0.2484,
+      "step": 105875
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.066041423683265e-06,
+      "loss": 0.2721,
+      "step": 105880
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.063945862590219e-06,
+      "loss": 0.2546,
+      "step": 105885
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.061850521400676e-06,
+      "loss": 0.2546,
+      "step": 105890
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.059755400141833e-06,
+      "loss": 0.2708,
+      "step": 105895
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.0576604988409e-06,
+      "loss": 0.2522,
+      "step": 105900
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.055565817525088e-06,
+      "loss": 0.2467,
+      "step": 105905
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.053471356221589e-06,
+      "loss": 0.2535,
+      "step": 105910
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.051377114957595e-06,
+      "loss": 0.2481,
+      "step": 105915
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.049283093760294e-06,
+      "loss": 0.2395,
+      "step": 105920
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.047189292656885e-06,
+      "loss": 0.222,
+      "step": 105925
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.045095711674553e-06,
+      "loss": 0.278,
+      "step": 105930
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.043002350840479e-06,
+      "loss": 0.2369,
+      "step": 105935
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.040909210181838e-06,
+      "loss": 0.2597,
+      "step": 105940
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.038816289725821e-06,
+      "loss": 0.2459,
+      "step": 105945
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.036723589499586e-06,
+      "loss": 0.2616,
+      "step": 105950
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.034631109530319e-06,
+      "loss": 0.2529,
+      "step": 105955
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.032538849845195e-06,
+      "loss": 0.2854,
+      "step": 105960
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.030446810471371e-06,
+      "loss": 0.2783,
+      "step": 105965
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.028354991436014e-06,
+      "loss": 0.2637,
+      "step": 105970
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.02626339276627e-06,
+      "loss": 0.2745,
+      "step": 105975
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.02417201448932e-06,
+      "loss": 0.2596,
+      "step": 105980
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.022080856632312e-06,
+      "loss": 0.261,
+      "step": 105985
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.019989919222393e-06,
+      "loss": 0.2877,
+      "step": 105990
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.017899202286705e-06,
+      "loss": 0.2326,
+      "step": 105995
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.015808705852412e-06,
+      "loss": 0.2643,
+      "step": 106000
+    },
+    {
+      "epoch": 3.73,
+      "eval_loss": 0.25173264741897583,
+      "eval_runtime": 10.5541,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 9.475,
+      "step": 106000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.013718429946645e-06,
+      "loss": 0.2656,
+      "step": 106005
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.011628374596555e-06,
+      "loss": 0.2496,
+      "step": 106010
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.009538539829267e-06,
+      "loss": 0.2472,
+      "step": 106015
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.007448925671934e-06,
+      "loss": 0.2544,
+      "step": 106020
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.005359532151677e-06,
+      "loss": 0.2508,
+      "step": 106025
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.003270359295628e-06,
+      "loss": 0.2504,
+      "step": 106030
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 8.001181407130903e-06,
+      "loss": 0.2553,
+      "step": 106035
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.99909267568464e-06,
+      "loss": 0.2288,
+      "step": 106040
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.99700416498396e-06,
+      "loss": 0.2656,
+      "step": 106045
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.994915875055971e-06,
+      "loss": 0.2688,
+      "step": 106050
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.992827805927785e-06,
+      "loss": 0.2497,
+      "step": 106055
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.990739957626525e-06,
+      "loss": 0.2433,
+      "step": 106060
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.988652330179303e-06,
+      "loss": 0.268,
+      "step": 106065
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.986564923613214e-06,
+      "loss": 0.2484,
+      "step": 106070
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.98447773795537e-06,
+      "loss": 0.2794,
+      "step": 106075
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.982390773232873e-06,
+      "loss": 0.2717,
+      "step": 106080
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.980304029472816e-06,
+      "loss": 0.2683,
+      "step": 106085
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.978217506702284e-06,
+      "loss": 0.2694,
+      "step": 106090
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.976131204948393e-06,
+      "loss": 0.2623,
+      "step": 106095
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.974045124238217e-06,
+      "loss": 0.2564,
+      "step": 106100
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.971959264598844e-06,
+      "loss": 0.2586,
+      "step": 106105
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.969873626057348e-06,
+      "loss": 0.2623,
+      "step": 106110
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.967788208640823e-06,
+      "loss": 0.2551,
+      "step": 106115
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.96570301237635e-06,
+      "loss": 0.2527,
+      "step": 106120
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.963618037290999e-06,
+      "loss": 0.2826,
+      "step": 106125
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.961533283411829e-06,
+      "loss": 0.2554,
+      "step": 106130
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.95944875076593e-06,
+      "loss": 0.2638,
+      "step": 106135
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.95736443938036e-06,
+      "loss": 0.27,
+      "step": 106140
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.955280349282183e-06,
+      "loss": 0.2426,
+      "step": 106145
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.953196480498446e-06,
+      "loss": 0.247,
+      "step": 106150
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 7.951112833056226e-06,
+      "loss": 0.2563,
+      "step": 106155
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.949029406982574e-06,
+      "loss": 0.2629,
+      "step": 106160
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.946946202304528e-06,
+      "loss": 0.253,
+      "step": 106165
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.944863219049155e-06,
+      "loss": 0.2575,
+      "step": 106170
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.942780457243487e-06,
+      "loss": 0.268,
+      "step": 106175
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.94069791691458e-06,
+      "loss": 0.2613,
+      "step": 106180
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.93861559808946e-06,
+      "loss": 0.2466,
+      "step": 106185
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.936533500795182e-06,
+      "loss": 0.2552,
+      "step": 106190
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.934451625058769e-06,
+      "loss": 0.2545,
+      "step": 106195
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.932369970907256e-06,
+      "loss": 0.2565,
+      "step": 106200
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.930288538367665e-06,
+      "loss": 0.2552,
+      "step": 106205
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.928207327467033e-06,
+      "loss": 0.2505,
+      "step": 106210
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.926126338232378e-06,
+      "loss": 0.2556,
+      "step": 106215
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.924045570690714e-06,
+      "loss": 0.2675,
+      "step": 106220
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.921965024869074e-06,
+      "loss": 0.2662,
+      "step": 106225
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.919884700794452e-06,
+      "loss": 0.2472,
+      "step": 106230
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.91780459849388e-06,
+      "loss": 0.2526,
+      "step": 106235
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.91572471799436e-06,
+      "loss": 0.2511,
+      "step": 106240
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.913645059322886e-06,
+      "loss": 0.244,
+      "step": 106245
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.911565622506478e-06,
+      "loss": 0.2755,
+      "step": 106250
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.909486407572132e-06,
+      "loss": 0.2588,
+      "step": 106255
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.907407414546836e-06,
+      "loss": 0.2589,
+      "step": 106260
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.905328643457586e-06,
+      "loss": 0.2673,
+      "step": 106265
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.903250094331385e-06,
+      "loss": 0.2476,
+      "step": 106270
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.901171767195215e-06,
+      "loss": 0.2565,
+      "step": 106275
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.899093662076052e-06,
+      "loss": 0.2415,
+      "step": 106280
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.897015779000887e-06,
+      "loss": 0.2595,
+      "step": 106285
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.894938117996711e-06,
+      "loss": 0.2557,
+      "step": 106290
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.89286067909049e-06,
+      "loss": 0.2347,
+      "step": 106295
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.890783462309193e-06,
+      "loss": 0.2701,
+      "step": 106300
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.888706467679805e-06,
+      "loss": 0.2512,
+      "step": 106305
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.886629695229284e-06,
+      "loss": 0.2537,
+      "step": 106310
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.884553144984599e-06,
+      "loss": 0.264,
+      "step": 106315
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.882476816972706e-06,
+      "loss": 0.2409,
+      "step": 106320
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.880400711220576e-06,
+      "loss": 0.2468,
+      "step": 106325
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.878324827755163e-06,
+      "loss": 0.2689,
+      "step": 106330
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.876249166603408e-06,
+      "loss": 0.2537,
+      "step": 106335
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.874173727792275e-06,
+      "loss": 0.2725,
+      "step": 106340
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.872098511348716e-06,
+      "loss": 0.26,
+      "step": 106345
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.87002351729967e-06,
+      "loss": 0.258,
+      "step": 106350
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.86794874567208e-06,
+      "loss": 0.2454,
+      "step": 106355
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.865874196492878e-06,
+      "loss": 0.2593,
+      "step": 106360
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.863799869789016e-06,
+      "loss": 0.2799,
+      "step": 106365
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.861725765587419e-06,
+      "loss": 0.2782,
+      "step": 106370
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.859651883915018e-06,
+      "loss": 0.2602,
+      "step": 106375
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.857578224798731e-06,
+      "loss": 0.25,
+      "step": 106380
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.855504788265505e-06,
+      "loss": 0.2351,
+      "step": 106385
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.85343157434224e-06,
+      "loss": 0.2632,
+      "step": 106390
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.851358583055875e-06,
+      "loss": 0.2442,
+      "step": 106395
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.849285814433312e-06,
+      "loss": 0.2653,
+      "step": 106400
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.847213268501478e-06,
+      "loss": 0.2266,
+      "step": 106405
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.845140945287271e-06,
+      "loss": 0.2704,
+      "step": 106410
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.843068844817597e-06,
+      "loss": 0.2662,
+      "step": 106415
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.840996967119377e-06,
+      "loss": 0.2551,
+      "step": 106420
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.838925312219503e-06,
+      "loss": 0.2598,
+      "step": 106425
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.836853880144874e-06,
+      "loss": 0.2275,
+      "step": 106430
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.834782670922377e-06,
+      "loss": 0.2677,
+      "step": 106435
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 7.832711684578925e-06,
+      "loss": 0.279,
+      "step": 106440
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.830640921141388e-06,
+      "loss": 0.2623,
+      "step": 106445
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.828570380636672e-06,
+      "loss": 0.2451,
+      "step": 106450
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.826500063091646e-06,
+      "loss": 0.2776,
+      "step": 106455
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.824429968533207e-06,
+      "loss": 0.2688,
+      "step": 106460
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.822360096988224e-06,
+      "loss": 0.2786,
+      "step": 106465
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.820290448483574e-06,
+      "loss": 0.2598,
+      "step": 106470
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.818221023046127e-06,
+      "loss": 0.2687,
+      "step": 106475
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.816151820702761e-06,
+      "loss": 0.2457,
+      "step": 106480
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.81408284148034e-06,
+      "loss": 0.2553,
+      "step": 106485
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.812014085405728e-06,
+      "loss": 0.2522,
+      "step": 106490
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.809945552505779e-06,
+      "loss": 0.2643,
+      "step": 106495
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.807877242807357e-06,
+      "loss": 0.2496,
+      "step": 106500
+    },
+    {
+      "epoch": 3.75,
+      "eval_loss": 0.2516409158706665,
+      "eval_runtime": 10.5352,
+      "eval_samples_per_second": 9.492,
+      "eval_steps_per_second": 9.492,
+      "step": 106500
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.805809156337327e-06,
+      "loss": 0.2422,
+      "step": 106505
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.803741293122533e-06,
+      "loss": 0.2345,
+      "step": 106510
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.801673653189818e-06,
+      "loss": 0.2534,
+      "step": 106515
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.799606236566045e-06,
+      "loss": 0.271,
+      "step": 106520
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.797539043278048e-06,
+      "loss": 0.2369,
+      "step": 106525
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.795472073352664e-06,
+      "loss": 0.2472,
+      "step": 106530
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.793405326816744e-06,
+      "loss": 0.2607,
+      "step": 106535
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.791338803697115e-06,
+      "loss": 0.2395,
+      "step": 106540
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.789272504020611e-06,
+      "loss": 0.2539,
+      "step": 106545
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.787206427814053e-06,
+      "loss": 0.2468,
+      "step": 106550
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.785140575104277e-06,
+      "loss": 0.2856,
+      "step": 106555
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.783074945918115e-06,
+      "loss": 0.2525,
+      "step": 106560
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.781009540282378e-06,
+      "loss": 0.2338,
+      "step": 106565
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.778944358223875e-06,
+      "loss": 0.2603,
+      "step": 106570
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.776879399769438e-06,
+      "loss": 0.2385,
+      "step": 106575
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.77481466494587e-06,
+      "loss": 0.2633,
+      "step": 106580
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.772750153779985e-06,
+      "loss": 0.2625,
+      "step": 106585
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.770685866298577e-06,
+      "loss": 0.2417,
+      "step": 106590
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.768621802528467e-06,
+      "loss": 0.2505,
+      "step": 106595
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.766557962496445e-06,
+      "loss": 0.2446,
+      "step": 106600
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.764494346229307e-06,
+      "loss": 0.2627,
+      "step": 106605
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.762430953753855e-06,
+      "loss": 0.2617,
+      "step": 106610
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.760367785096869e-06,
+      "loss": 0.2554,
+      "step": 106615
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.758304840285158e-06,
+      "loss": 0.253,
+      "step": 106620
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.75624211934549e-06,
+      "loss": 0.2655,
+      "step": 106625
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.75417962230465e-06,
+      "loss": 0.2666,
+      "step": 106630
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.75211734918943e-06,
+      "loss": 0.2492,
+      "step": 106635
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.750055300026599e-06,
+      "loss": 0.2651,
+      "step": 106640
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.747993474842923e-06,
+      "loss": 0.286,
+      "step": 106645
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.745931873665194e-06,
+      "loss": 0.2526,
+      "step": 106650
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.743870496520165e-06,
+      "loss": 0.246,
+      "step": 106655
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.741809343434598e-06,
+      "loss": 0.254,
+      "step": 106660
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.739748414435272e-06,
+      "loss": 0.2589,
+      "step": 106665
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.737687709548927e-06,
+      "loss": 0.2474,
+      "step": 106670
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.735627228802342e-06,
+      "loss": 0.2452,
+      "step": 106675
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.733566972222259e-06,
+      "loss": 0.2682,
+      "step": 106680
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.73150693983542e-06,
+      "loss": 0.2653,
+      "step": 106685
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.729447131668593e-06,
+      "loss": 0.2691,
+      "step": 106690
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.727387547748513e-06,
+      "loss": 0.2659,
+      "step": 106695
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.725328188101925e-06,
+      "loss": 0.2828,
+      "step": 106700
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.723269052755553e-06,
+      "loss": 0.2698,
+      "step": 106705
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.721210141736156e-06,
+      "loss": 0.2539,
+      "step": 106710
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.719151455070462e-06,
+      "loss": 0.2575,
+      "step": 106715
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.717092992785185e-06,
+      "loss": 0.2708,
+      "step": 106720
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 7.715034754907068e-06,
+      "loss": 0.2411,
+      "step": 106725
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.71297674146284e-06,
+      "loss": 0.2614,
+      "step": 106730
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.71091895247922e-06,
+      "loss": 0.2691,
+      "step": 106735
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.708861387982921e-06,
+      "loss": 0.2591,
+      "step": 106740
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.706804048000654e-06,
+      "loss": 0.2456,
+      "step": 106745
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.704746932559148e-06,
+      "loss": 0.2298,
+      "step": 106750
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.702690041685105e-06,
+      "loss": 0.2538,
+      "step": 106755
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.70063337540523e-06,
+      "loss": 0.2592,
+      "step": 106760
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.698576933746225e-06,
+      "loss": 0.2535,
+      "step": 106765
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.696520716734804e-06,
+      "loss": 0.2577,
+      "step": 106770
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.694464724397649e-06,
+      "loss": 0.256,
+      "step": 106775
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.692408956761466e-06,
+      "loss": 0.2563,
+      "step": 106780
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.690353413852952e-06,
+      "loss": 0.2454,
+      "step": 106785
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.688298095698792e-06,
+      "loss": 0.2379,
+      "step": 106790
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.686243002325675e-06,
+      "loss": 0.2398,
+      "step": 106795
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.684188133760272e-06,
+      "loss": 0.2724,
+      "step": 106800
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.682133490029286e-06,
+      "loss": 0.246,
+      "step": 106805
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.680079071159379e-06,
+      "loss": 0.2528,
+      "step": 106810
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.678024877177234e-06,
+      "loss": 0.26,
+      "step": 106815
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.675970908109512e-06,
+      "loss": 0.2595,
+      "step": 106820
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.6739171639829e-06,
+      "loss": 0.2548,
+      "step": 106825
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.671863644824045e-06,
+      "loss": 0.2683,
+      "step": 106830
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.669810350659629e-06,
+      "loss": 0.2656,
+      "step": 106835
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.667757281516296e-06,
+      "loss": 0.2655,
+      "step": 106840
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.665704437420724e-06,
+      "loss": 0.2698,
+      "step": 106845
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.663651818399553e-06,
+      "loss": 0.2581,
+      "step": 106850
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.661599424479437e-06,
+      "loss": 0.2746,
+      "step": 106855
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.659547255687023e-06,
+      "loss": 0.2813,
+      "step": 106860
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.657495312048963e-06,
+      "loss": 0.2596,
+      "step": 106865
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.6554435935919e-06,
+      "loss": 0.2277,
+      "step": 106870
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.65339210034247e-06,
+      "loss": 0.2635,
+      "step": 106875
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.651340832327305e-06,
+      "loss": 0.2561,
+      "step": 106880
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.649289789573044e-06,
+      "loss": 0.2629,
+      "step": 106885
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.64723897210633e-06,
+      "loss": 0.2495,
+      "step": 106890
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.645188379953774e-06,
+      "loss": 0.2622,
+      "step": 106895
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.643138013142018e-06,
+      "loss": 0.2551,
+      "step": 106900
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.641087871697677e-06,
+      "loss": 0.2712,
+      "step": 106905
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.63903795564737e-06,
+      "loss": 0.2408,
+      "step": 106910
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.636988265017702e-06,
+      "loss": 0.2498,
+      "step": 106915
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.634938799835312e-06,
+      "loss": 0.2667,
+      "step": 106920
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.632889560126794e-06,
+      "loss": 0.2521,
+      "step": 106925
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.630840545918761e-06,
+      "loss": 0.2465,
+      "step": 106930
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.628791757237811e-06,
+      "loss": 0.2646,
+      "step": 106935
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.626743194110547e-06,
+      "loss": 0.241,
+      "step": 106940
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.624694856563585e-06,
+      "loss": 0.2578,
+      "step": 106945
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.6226467446235075e-06,
+      "loss": 0.2587,
+      "step": 106950
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.620598858316905e-06,
+      "loss": 0.263,
+      "step": 106955
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.618551197670376e-06,
+      "loss": 0.2499,
+      "step": 106960
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.616503762710509e-06,
+      "loss": 0.2554,
+      "step": 106965
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.6144565534638835e-06,
+      "loss": 0.2798,
+      "step": 106970
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.612409569957074e-06,
+      "loss": 0.2447,
+      "step": 106975
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.610362812216676e-06,
+      "loss": 0.2339,
+      "step": 106980
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.608316280269253e-06,
+      "loss": 0.236,
+      "step": 106985
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.6062699741413775e-06,
+      "loss": 0.2403,
+      "step": 106990
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.604223893859622e-06,
+      "loss": 0.2719,
+      "step": 106995
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.602178039450563e-06,
+      "loss": 0.2621,
+      "step": 107000
+    },
+    {
+      "epoch": 3.76,
+      "eval_loss": 0.251350462436676,
+      "eval_runtime": 10.5372,
+      "eval_samples_per_second": 9.49,
+      "eval_steps_per_second": 9.49,
+      "step": 107000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.600132410940758e-06,
+      "loss": 0.271,
+      "step": 107005
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 7.598087008356758e-06,
+      "loss": 0.2491,
+      "step": 107010
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.596041831725137e-06,
+      "loss": 0.2713,
+      "step": 107015
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.593996881072443e-06,
+      "loss": 0.2486,
+      "step": 107020
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.591952156425231e-06,
+      "loss": 0.2624,
+      "step": 107025
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.589907657810039e-06,
+      "loss": 0.2571,
+      "step": 107030
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.587863385253427e-06,
+      "loss": 0.2679,
+      "step": 107035
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.585819338781938e-06,
+      "loss": 0.2466,
+      "step": 107040
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.583775518422098e-06,
+      "loss": 0.2415,
+      "step": 107045
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.581731924200464e-06,
+      "loss": 0.2564,
+      "step": 107050
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.579688556143555e-06,
+      "loss": 0.2617,
+      "step": 107055
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.577645414277917e-06,
+      "loss": 0.2695,
+      "step": 107060
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.575602498630072e-06,
+      "loss": 0.2421,
+      "step": 107065
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.573559809226536e-06,
+      "loss": 0.2467,
+      "step": 107070
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.571517346093851e-06,
+      "loss": 0.2511,
+      "step": 107075
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.569475109258525e-06,
+      "loss": 0.2721,
+      "step": 107080
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.567433098747079e-06,
+      "loss": 0.2477,
+      "step": 107085
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.565391314586018e-06,
+      "loss": 0.2662,
+      "step": 107090
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.563349756801869e-06,
+      "loss": 0.2606,
+      "step": 107095
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.561308425421124e-06,
+      "loss": 0.2439,
+      "step": 107100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.559267320470304e-06,
+      "loss": 0.2411,
+      "step": 107105
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.557226441975898e-06,
+      "loss": 0.2375,
+      "step": 107110
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.555185789964417e-06,
+      "loss": 0.258,
+      "step": 107115
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.5531453644623525e-06,
+      "loss": 0.2532,
+      "step": 107120
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.5511051654961915e-06,
+      "loss": 0.2477,
+      "step": 107125
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.549065193092436e-06,
+      "loss": 0.2492,
+      "step": 107130
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.54702544727757e-06,
+      "loss": 0.2551,
+      "step": 107135
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.544985928078077e-06,
+      "loss": 0.2316,
+      "step": 107140
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.542946635520429e-06,
+      "loss": 0.2486,
+      "step": 107145
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.5409075696311225e-06,
+      "loss": 0.2684,
+      "step": 107150
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.538868730436627e-06,
+      "loss": 0.2515,
+      "step": 107155
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.536830117963403e-06,
+      "loss": 0.2779,
+      "step": 107160
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.534791732237934e-06,
+      "loss": 0.2644,
+      "step": 107165
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.532753573286691e-06,
+      "loss": 0.2374,
+      "step": 107170
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.530715641136129e-06,
+      "loss": 0.2743,
+      "step": 107175
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.528677935812714e-06,
+      "loss": 0.2512,
+      "step": 107180
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.526640457342896e-06,
+      "loss": 0.2593,
+      "step": 107185
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.5246032057531415e-06,
+      "loss": 0.246,
+      "step": 107190
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.522566181069899e-06,
+      "loss": 0.2391,
+      "step": 107195
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.520529383319616e-06,
+      "loss": 0.2643,
+      "step": 107200
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.518492812528732e-06,
+      "loss": 0.2446,
+      "step": 107205
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.516456468723707e-06,
+      "loss": 0.2642,
+      "step": 107210
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.514420351930962e-06,
+      "loss": 0.284,
+      "step": 107215
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.512384462176955e-06,
+      "loss": 0.2778,
+      "step": 107220
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.510348799488104e-06,
+      "loss": 0.2793,
+      "step": 107225
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.508313363890853e-06,
+      "loss": 0.2586,
+      "step": 107230
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.506278155411628e-06,
+      "loss": 0.2428,
+      "step": 107235
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.504243174076841e-06,
+      "loss": 0.2601,
+      "step": 107240
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.502208419912937e-06,
+      "loss": 0.2664,
+      "step": 107245
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.500173892946328e-06,
+      "loss": 0.2744,
+      "step": 107250
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.498139593203424e-06,
+      "loss": 0.2545,
+      "step": 107255
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.496105520710636e-06,
+      "loss": 0.2582,
+      "step": 107260
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.494071675494391e-06,
+      "loss": 0.263,
+      "step": 107265
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.49203805758108e-06,
+      "loss": 0.2693,
+      "step": 107270
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.490004666997125e-06,
+      "loss": 0.2386,
+      "step": 107275
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.487971503768909e-06,
+      "loss": 0.2503,
+      "step": 107280
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.485938567922854e-06,
+      "loss": 0.2404,
+      "step": 107285
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.483905859485344e-06,
+      "loss": 0.2657,
+      "step": 107290
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 7.481873378482771e-06,
+      "loss": 0.2587,
+      "step": 107295
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.47984112494152e-06,
+      "loss": 0.2653,
+      "step": 107300
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.477809098887994e-06,
+      "loss": 0.2493,
+      "step": 107305
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.475777300348568e-06,
+      "loss": 0.2677,
+      "step": 107310
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.473745729349627e-06,
+      "loss": 0.2736,
+      "step": 107315
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.471714385917539e-06,
+      "loss": 0.253,
+      "step": 107320
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.46968327007869e-06,
+      "loss": 0.2603,
+      "step": 107325
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.467652381859455e-06,
+      "loss": 0.2642,
+      "step": 107330
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.4656217212862e-06,
+      "loss": 0.241,
+      "step": 107335
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.463591288385285e-06,
+      "loss": 0.2435,
+      "step": 107340
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.461561083183088e-06,
+      "loss": 0.2613,
+      "step": 107345
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.459531105705964e-06,
+      "loss": 0.2122,
+      "step": 107350
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.457501355980256e-06,
+      "loss": 0.2452,
+      "step": 107355
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.455471834032343e-06,
+      "loss": 0.2441,
+      "step": 107360
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.453442539888566e-06,
+      "loss": 0.2422,
+      "step": 107365
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.451413473575275e-06,
+      "loss": 0.258,
+      "step": 107370
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.449384635118805e-06,
+      "loss": 0.2488,
+      "step": 107375
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.44735602454551e-06,
+      "loss": 0.2523,
+      "step": 107380
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.445327641881739e-06,
+      "loss": 0.2399,
+      "step": 107385
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.443299487153818e-06,
+      "loss": 0.2626,
+      "step": 107390
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.4412715603880745e-06,
+      "loss": 0.274,
+      "step": 107395
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.439243861610856e-06,
+      "loss": 0.2693,
+      "step": 107400
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.437216390848484e-06,
+      "loss": 0.2787,
+      "step": 107405
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.435189148127283e-06,
+      "loss": 0.2647,
+      "step": 107410
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.433162133473568e-06,
+      "loss": 0.2889,
+      "step": 107415
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.43113534691367e-06,
+      "loss": 0.2834,
+      "step": 107420
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.429108788473904e-06,
+      "loss": 0.2552,
+      "step": 107425
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.427082458180573e-06,
+      "loss": 0.2517,
+      "step": 107430
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.425056356059995e-06,
+      "loss": 0.2573,
+      "step": 107435
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.423030482138485e-06,
+      "loss": 0.2585,
+      "step": 107440
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.421004836442341e-06,
+      "loss": 0.2431,
+      "step": 107445
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.418979418997865e-06,
+      "loss": 0.2421,
+      "step": 107450
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.416954229831347e-06,
+      "loss": 0.2353,
+      "step": 107455
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.414929268969095e-06,
+      "loss": 0.2701,
+      "step": 107460
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.412904536437401e-06,
+      "loss": 0.2685,
+      "step": 107465
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.410880032262543e-06,
+      "loss": 0.2601,
+      "step": 107470
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.408855756470823e-06,
+      "loss": 0.249,
+      "step": 107475
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.406831709088519e-06,
+      "loss": 0.2613,
+      "step": 107480
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.4048078901419045e-06,
+      "loss": 0.2371,
+      "step": 107485
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.40278429965727e-06,
+      "loss": 0.2581,
+      "step": 107490
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.400760937660875e-06,
+      "loss": 0.28,
+      "step": 107495
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.39873780417901e-06,
+      "loss": 0.2608,
+      "step": 107500
+    },
+    {
+      "epoch": 3.78,
+      "eval_loss": 0.25113674998283386,
+      "eval_runtime": 10.5445,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 107500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.3967148992379356e-06,
+      "loss": 0.2622,
+      "step": 107505
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.394692222863908e-06,
+      "loss": 0.249,
+      "step": 107510
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.3926697750832105e-06,
+      "loss": 0.2459,
+      "step": 107515
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.390647555922089e-06,
+      "loss": 0.2483,
+      "step": 107520
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.388625565406806e-06,
+      "loss": 0.2719,
+      "step": 107525
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.386603803563605e-06,
+      "loss": 0.2514,
+      "step": 107530
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.384582270418753e-06,
+      "loss": 0.2525,
+      "step": 107535
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.382560965998483e-06,
+      "loss": 0.2545,
+      "step": 107540
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.380539890329058e-06,
+      "loss": 0.2962,
+      "step": 107545
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.378519043436704e-06,
+      "loss": 0.2555,
+      "step": 107550
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.3764984253476724e-06,
+      "loss": 0.2605,
+      "step": 107555
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.374478036088195e-06,
+      "loss": 0.2702,
+      "step": 107560
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.3724578756845076e-06,
+      "loss": 0.2675,
+      "step": 107565
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.370437944162825e-06,
+      "loss": 0.2398,
+      "step": 107570
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.368418241549399e-06,
+      "loss": 0.272,
+      "step": 107575
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 7.366398767870441e-06,
+      "loss": 0.2472,
+      "step": 107580
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.3643795231521675e-06,
+      "loss": 0.2548,
+      "step": 107585
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.362360507420809e-06,
+      "loss": 0.2573,
+      "step": 107590
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.360341720702577e-06,
+      "loss": 0.2434,
+      "step": 107595
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.358323163023675e-06,
+      "loss": 0.2234,
+      "step": 107600
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.356304834410321e-06,
+      "loss": 0.2456,
+      "step": 107605
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.354286734888729e-06,
+      "loss": 0.2644,
+      "step": 107610
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.352268864485093e-06,
+      "loss": 0.2681,
+      "step": 107615
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.350251223225619e-06,
+      "loss": 0.2773,
+      "step": 107620
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.34823381113649e-06,
+      "loss": 0.2726,
+      "step": 107625
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.346216628243921e-06,
+      "loss": 0.2355,
+      "step": 107630
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.344199674574093e-06,
+      "loss": 0.2691,
+      "step": 107635
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.342182950153198e-06,
+      "loss": 0.2433,
+      "step": 107640
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.340166455007414e-06,
+      "loss": 0.2368,
+      "step": 107645
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.338150189162937e-06,
+      "loss": 0.2437,
+      "step": 107650
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.336134152645932e-06,
+      "loss": 0.264,
+      "step": 107655
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.334118345482593e-06,
+      "loss": 0.2433,
+      "step": 107660
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.332102767699076e-06,
+      "loss": 0.2492,
+      "step": 107665
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.3300874193215715e-06,
+      "loss": 0.2647,
+      "step": 107670
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.328072300376234e-06,
+      "loss": 0.2439,
+      "step": 107675
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.326057410889231e-06,
+      "loss": 0.2655,
+      "step": 107680
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.32404275088672e-06,
+      "loss": 0.2685,
+      "step": 107685
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.32202832039487e-06,
+      "loss": 0.2541,
+      "step": 107690
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.320014119439833e-06,
+      "loss": 0.254,
+      "step": 107695
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.318000148047754e-06,
+      "loss": 0.2618,
+      "step": 107700
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.315986406244798e-06,
+      "loss": 0.255,
+      "step": 107705
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.313972894057094e-06,
+      "loss": 0.2475,
+      "step": 107710
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.311959611510807e-06,
+      "loss": 0.2343,
+      "step": 107715
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.309946558632061e-06,
+      "loss": 0.2428,
+      "step": 107720
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.307933735447006e-06,
+      "loss": 0.2493,
+      "step": 107725
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.305921141981772e-06,
+      "loss": 0.2648,
+      "step": 107730
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.303908778262491e-06,
+      "loss": 0.2529,
+      "step": 107735
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.301896644315287e-06,
+      "loss": 0.2375,
+      "step": 107740
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.299884740166297e-06,
+      "loss": 0.2622,
+      "step": 107745
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.29787306584164e-06,
+      "loss": 0.2582,
+      "step": 107750
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.295861621367434e-06,
+      "loss": 0.2783,
+      "step": 107755
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.293850406769789e-06,
+      "loss": 0.2627,
+      "step": 107760
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.2918394220748306e-06,
+      "loss": 0.2553,
+      "step": 107765
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.289828667308673e-06,
+      "loss": 0.2623,
+      "step": 107770
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.287818142497418e-06,
+      "loss": 0.2408,
+      "step": 107775
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.285807847667167e-06,
+      "loss": 0.2554,
+      "step": 107780
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.283797782844032e-06,
+      "loss": 0.2503,
+      "step": 107785
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.281787948054109e-06,
+      "loss": 0.2431,
+      "step": 107790
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.27977834332349e-06,
+      "loss": 0.2696,
+      "step": 107795
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.277768968678264e-06,
+      "loss": 0.2607,
+      "step": 107800
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.27575982414454e-06,
+      "loss": 0.2608,
+      "step": 107805
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.2737509097483894e-06,
+      "loss": 0.274,
+      "step": 107810
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.271742225515896e-06,
+      "loss": 0.2475,
+      "step": 107815
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.269733771473147e-06,
+      "loss": 0.2381,
+      "step": 107820
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.2677255476462255e-06,
+      "loss": 0.2625,
+      "step": 107825
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.265717554061205e-06,
+      "loss": 0.248,
+      "step": 107830
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.2637097907441444e-06,
+      "loss": 0.2609,
+      "step": 107835
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.261702257721131e-06,
+      "loss": 0.2537,
+      "step": 107840
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.259694955018226e-06,
+      "loss": 0.2211,
+      "step": 107845
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.25768788266149e-06,
+      "loss": 0.2568,
+      "step": 107850
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.255681040676976e-06,
+      "loss": 0.2581,
+      "step": 107855
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.253674429090756e-06,
+      "loss": 0.2605,
+      "step": 107860
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 7.25166804792888e-06,
+      "loss": 0.2435,
+      "step": 107865
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.249661897217391e-06,
+      "loss": 0.26,
+      "step": 107870
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.247655976982343e-06,
+      "loss": 0.2477,
+      "step": 107875
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.245650287249791e-06,
+      "loss": 0.2548,
+      "step": 107880
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.243644828045768e-06,
+      "loss": 0.2486,
+      "step": 107885
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.241639599396316e-06,
+      "loss": 0.2293,
+      "step": 107890
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.2396346013274635e-06,
+      "loss": 0.2371,
+      "step": 107895
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.237629833865259e-06,
+      "loss": 0.2498,
+      "step": 107900
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.235625297035725e-06,
+      "loss": 0.2501,
+      "step": 107905
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.233620990864889e-06,
+      "loss": 0.2624,
+      "step": 107910
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.231616915378769e-06,
+      "loss": 0.2747,
+      "step": 107915
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.2296130706033996e-06,
+      "loss": 0.2571,
+      "step": 107920
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.227609456564788e-06,
+      "loss": 0.2528,
+      "step": 107925
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.225606073288962e-06,
+      "loss": 0.2426,
+      "step": 107930
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.223602920801919e-06,
+      "loss": 0.2591,
+      "step": 107935
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.221599999129686e-06,
+      "loss": 0.2634,
+      "step": 107940
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.219597308298259e-06,
+      "loss": 0.2832,
+      "step": 107945
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.217594848333636e-06,
+      "loss": 0.2459,
+      "step": 107950
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.215592619261832e-06,
+      "loss": 0.257,
+      "step": 107955
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.21359062110884e-06,
+      "loss": 0.2516,
+      "step": 107960
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.21158885390065e-06,
+      "loss": 0.2523,
+      "step": 107965
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.209587317663246e-06,
+      "loss": 0.2596,
+      "step": 107970
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.207586012422637e-06,
+      "loss": 0.2645,
+      "step": 107975
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.20558493820479e-06,
+      "loss": 0.2577,
+      "step": 107980
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.203584095035704e-06,
+      "loss": 0.2718,
+      "step": 107985
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.2015834829413406e-06,
+      "loss": 0.246,
+      "step": 107990
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.199583101947696e-06,
+      "loss": 0.2511,
+      "step": 107995
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.197582952080731e-06,
+      "loss": 0.2522,
+      "step": 108000
+    },
+    {
+      "epoch": 3.8,
+      "eval_loss": 0.25110575556755066,
+      "eval_runtime": 10.5564,
+      "eval_samples_per_second": 9.473,
+      "eval_steps_per_second": 9.473,
+      "step": 108000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.195583033366421e-06,
+      "loss": 0.252,
+      "step": 108005
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.193583345830723e-06,
+      "loss": 0.2597,
+      "step": 108010
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.19158388949962e-06,
+      "loss": 0.2421,
+      "step": 108015
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.189584664399063e-06,
+      "loss": 0.2783,
+      "step": 108020
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.187585670555011e-06,
+      "loss": 0.2669,
+      "step": 108025
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.185586907993413e-06,
+      "loss": 0.2566,
+      "step": 108030
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1835883767402385e-06,
+      "loss": 0.2678,
+      "step": 108035
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.181590076821418e-06,
+      "loss": 0.2312,
+      "step": 108040
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1795920082629144e-06,
+      "loss": 0.2733,
+      "step": 108045
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.177594171090657e-06,
+      "loss": 0.2529,
+      "step": 108050
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1755965653306026e-06,
+      "loss": 0.2586,
+      "step": 108055
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.173599191008678e-06,
+      "loss": 0.2694,
+      "step": 108060
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.171602048150816e-06,
+      "loss": 0.2616,
+      "step": 108065
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.169605136782956e-06,
+      "loss": 0.2726,
+      "step": 108070
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.167608456931024e-06,
+      "loss": 0.2387,
+      "step": 108075
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.165612008620945e-06,
+      "loss": 0.2768,
+      "step": 108080
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.163615791878634e-06,
+      "loss": 0.2449,
+      "step": 108085
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1616198067300215e-06,
+      "loss": 0.2485,
+      "step": 108090
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1596240532010136e-06,
+      "loss": 0.2662,
+      "step": 108095
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.157628531317539e-06,
+      "loss": 0.239,
+      "step": 108100
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1556332411054915e-06,
+      "loss": 0.2711,
+      "step": 108105
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.153638182590794e-06,
+      "loss": 0.2437,
+      "step": 108110
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1516433557993426e-06,
+      "loss": 0.2689,
+      "step": 108115
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.149648760757041e-06,
+      "loss": 0.2491,
+      "step": 108120
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.147654397489778e-06,
+      "loss": 0.2512,
+      "step": 108125
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.145660266023466e-06,
+      "loss": 0.2689,
+      "step": 108130
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1436663663839855e-06,
+      "loss": 0.2484,
+      "step": 108135
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.1416726985972264e-06,
+      "loss": 0.2558,
+      "step": 108140
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 7.139679262689083e-06,
+      "loss": 0.274,
+      "step": 108145
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.137686058685428e-06,
+      "loss": 0.2741,
+      "step": 108150
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.135693086612155e-06,
+      "loss": 0.2981,
+      "step": 108155
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.133700346495134e-06,
+      "loss": 0.2628,
+      "step": 108160
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.131707838360233e-06,
+      "loss": 0.2518,
+      "step": 108165
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.1297155622333356e-06,
+      "loss": 0.2673,
+      "step": 108170
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.1277235181403054e-06,
+      "loss": 0.2507,
+      "step": 108175
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.1257317061070004e-06,
+      "loss": 0.2517,
+      "step": 108180
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.123740126159298e-06,
+      "loss": 0.2719,
+      "step": 108185
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.121748778323048e-06,
+      "loss": 0.2713,
+      "step": 108190
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.119757662624107e-06,
+      "loss": 0.2425,
+      "step": 108195
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.117766779088322e-06,
+      "loss": 0.2567,
+      "step": 108200
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.115776127741553e-06,
+      "loss": 0.2699,
+      "step": 108205
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.113785708609649e-06,
+      "loss": 0.2453,
+      "step": 108210
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.11179552171845e-06,
+      "loss": 0.277,
+      "step": 108215
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.109805567093792e-06,
+      "loss": 0.2807,
+      "step": 108220
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.107815844761525e-06,
+      "loss": 0.2494,
+      "step": 108225
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.10582635474748e-06,
+      "loss": 0.255,
+      "step": 108230
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.103837097077484e-06,
+      "loss": 0.2396,
+      "step": 108235
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.101848071777364e-06,
+      "loss": 0.2441,
+      "step": 108240
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.099859278872958e-06,
+      "loss": 0.2754,
+      "step": 108245
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.097870718390084e-06,
+      "loss": 0.2485,
+      "step": 108250
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.095882390354553e-06,
+      "loss": 0.2541,
+      "step": 108255
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.093894294792186e-06,
+      "loss": 0.2545,
+      "step": 108260
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.091906431728812e-06,
+      "loss": 0.2585,
+      "step": 108265
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.089918801190232e-06,
+      "loss": 0.2477,
+      "step": 108270
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.087931403202253e-06,
+      "loss": 0.2772,
+      "step": 108275
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.08594423779067e-06,
+      "loss": 0.2597,
+      "step": 108280
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.083957304981303e-06,
+      "loss": 0.2318,
+      "step": 108285
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.081970604799945e-06,
+      "loss": 0.2652,
+      "step": 108290
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.079984137272383e-06,
+      "loss": 0.2515,
+      "step": 108295
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.0779979024244245e-06,
+      "loss": 0.2409,
+      "step": 108300
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.076011900281851e-06,
+      "loss": 0.2472,
+      "step": 108305
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.074026130870445e-06,
+      "loss": 0.264,
+      "step": 108310
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.072040594215995e-06,
+      "loss": 0.2683,
+      "step": 108315
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.070055290344291e-06,
+      "loss": 0.2634,
+      "step": 108320
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.0680702192811e-06,
+      "loss": 0.2489,
+      "step": 108325
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.066085381052204e-06,
+      "loss": 0.2795,
+      "step": 108330
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.064100775683358e-06,
+      "loss": 0.24,
+      "step": 108335
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.0621164032003555e-06,
+      "loss": 0.2562,
+      "step": 108340
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.060132263628946e-06,
+      "loss": 0.261,
+      "step": 108345
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.058148356994898e-06,
+      "loss": 0.2728,
+      "step": 108350
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.056164683323962e-06,
+      "loss": 0.2721,
+      "step": 108355
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.054181242641911e-06,
+      "loss": 0.276,
+      "step": 108360
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.05219803497448e-06,
+      "loss": 0.2347,
+      "step": 108365
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.05021506034744e-06,
+      "loss": 0.2393,
+      "step": 108370
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.048232318786519e-06,
+      "loss": 0.2607,
+      "step": 108375
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.046249810317476e-06,
+      "loss": 0.2788,
+      "step": 108380
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.044267534966051e-06,
+      "loss": 0.248,
+      "step": 108385
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.042285492757975e-06,
+      "loss": 0.2547,
+      "step": 108390
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.040303683718982e-06,
+      "loss": 0.2323,
+      "step": 108395
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.038322107874817e-06,
+      "loss": 0.2419,
+      "step": 108400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.036340765251204e-06,
+      "loss": 0.2638,
+      "step": 108405
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.034359655873857e-06,
+      "loss": 0.2656,
+      "step": 108410
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.03237877976852e-06,
+      "loss": 0.2576,
+      "step": 108415
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.030398136960894e-06,
+      "loss": 0.2746,
+      "step": 108420
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.0284177274767135e-06,
+      "loss": 0.2601,
+      "step": 108425
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 7.0264375513416776e-06,
+      "loss": 0.2409,
+      "step": 108430
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.024457608581514e-06,
+      "loss": 0.2737,
+      "step": 108435
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.0224778992219214e-06,
+      "loss": 0.2879,
+      "step": 108440
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.020498423288604e-06,
+      "loss": 0.2708,
+      "step": 108445
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.018519180807262e-06,
+      "loss": 0.2499,
+      "step": 108450
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.016540171803601e-06,
+      "loss": 0.2447,
+      "step": 108455
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.014561396303318e-06,
+      "loss": 0.2602,
+      "step": 108460
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.012582854332101e-06,
+      "loss": 0.2647,
+      "step": 108465
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.0106045459156334e-06,
+      "loss": 0.2633,
+      "step": 108470
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.008626471079619e-06,
+      "loss": 0.2714,
+      "step": 108475
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.006648629849724e-06,
+      "loss": 0.2677,
+      "step": 108480
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.004671022251649e-06,
+      "loss": 0.2508,
+      "step": 108485
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.002693648311051e-06,
+      "loss": 0.2567,
+      "step": 108490
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 7.000716508053623e-06,
+      "loss": 0.2636,
+      "step": 108495
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.99873960150503e-06,
+      "loss": 0.2607,
+      "step": 108500
+    },
+    {
+      "epoch": 3.82,
+      "eval_loss": 0.25093138217926025,
+      "eval_runtime": 10.5444,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 108500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.9967629286909416e-06,
+      "loss": 0.241,
+      "step": 108505
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.994786489637012e-06,
+      "loss": 0.26,
+      "step": 108510
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.9928102843689225e-06,
+      "loss": 0.2456,
+      "step": 108515
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.990834312912323e-06,
+      "loss": 0.2521,
+      "step": 108520
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.988858575292867e-06,
+      "loss": 0.2463,
+      "step": 108525
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.9868830715362185e-06,
+      "loss": 0.2718,
+      "step": 108530
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.98530283692935e-06,
+      "loss": 0.2599,
+      "step": 108535
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.983327754190374e-06,
+      "loss": 0.2566,
+      "step": 108540
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.981352905386021e-06,
+      "loss": 0.2568,
+      "step": 108545
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.979378290541927e-06,
+      "loss": 0.2647,
+      "step": 108550
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.97740390968373e-06,
+      "loss": 0.2363,
+      "step": 108555
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.975429762837058e-06,
+      "loss": 0.2646,
+      "step": 108560
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.97345585002755e-06,
+      "loss": 0.258,
+      "step": 108565
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.971482171280846e-06,
+      "loss": 0.2637,
+      "step": 108570
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.969508726622565e-06,
+      "loss": 0.2563,
+      "step": 108575
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.967535516078325e-06,
+      "loss": 0.2778,
+      "step": 108580
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.965562539673756e-06,
+      "loss": 0.2618,
+      "step": 108585
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.963589797434475e-06,
+      "loss": 0.2527,
+      "step": 108590
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.9616172893860885e-06,
+      "loss": 0.2589,
+      "step": 108595
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.959645015554217e-06,
+      "loss": 0.2867,
+      "step": 108600
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.957672975964468e-06,
+      "loss": 0.2487,
+      "step": 108605
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.955701170642445e-06,
+      "loss": 0.267,
+      "step": 108610
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.953729599613743e-06,
+      "loss": 0.2538,
+      "step": 108615
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.951758262903976e-06,
+      "loss": 0.2294,
+      "step": 108620
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.949787160538729e-06,
+      "loss": 0.2474,
+      "step": 108625
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.947816292543605e-06,
+      "loss": 0.2506,
+      "step": 108630
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.945845658944183e-06,
+      "loss": 0.2478,
+      "step": 108635
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.943875259766064e-06,
+      "loss": 0.295,
+      "step": 108640
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.941905095034826e-06,
+      "loss": 0.2573,
+      "step": 108645
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.9399351647760485e-06,
+      "loss": 0.253,
+      "step": 108650
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.937965469015304e-06,
+      "loss": 0.2792,
+      "step": 108655
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.935996007778184e-06,
+      "loss": 0.251,
+      "step": 108660
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.93402678109025e-06,
+      "loss": 0.2515,
+      "step": 108665
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.932057788977062e-06,
+      "loss": 0.264,
+      "step": 108670
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.930089031464207e-06,
+      "loss": 0.2479,
+      "step": 108675
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.928120508577229e-06,
+      "loss": 0.2568,
+      "step": 108680
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.926152220341703e-06,
+      "loss": 0.2518,
+      "step": 108685
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.924184166783182e-06,
+      "loss": 0.2506,
+      "step": 108690
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.922216347927208e-06,
+      "loss": 0.2643,
+      "step": 108695
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.920248763799348e-06,
+      "loss": 0.2458,
+      "step": 108700
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.918281414425143e-06,
+      "loss": 0.2576,
+      "step": 108705
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.916314299830129e-06,
+      "loss": 0.2654,
+      "step": 108710
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 6.914347420039865e-06,
+      "loss": 0.2416,
+      "step": 108715
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.912380775079882e-06,
+      "loss": 0.2236,
+      "step": 108720
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.9104143649757065e-06,
+      "loss": 0.2305,
+      "step": 108725
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.9084481897528835e-06,
+      "loss": 0.2546,
+      "step": 108730
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.906482249436933e-06,
+      "loss": 0.2442,
+      "step": 108735
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.904516544053394e-06,
+      "loss": 0.2563,
+      "step": 108740
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.902551073627783e-06,
+      "loss": 0.2573,
+      "step": 108745
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.9005858381856105e-06,
+      "loss": 0.2586,
+      "step": 108750
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.898620837752412e-06,
+      "loss": 0.2463,
+      "step": 108755
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.896656072353691e-06,
+      "loss": 0.2651,
+      "step": 108760
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.894691542014961e-06,
+      "loss": 0.2895,
+      "step": 108765
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.892727246761721e-06,
+      "loss": 0.2465,
+      "step": 108770
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.890763186619495e-06,
+      "loss": 0.2639,
+      "step": 108775
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.888799361613771e-06,
+      "loss": 0.2673,
+      "step": 108780
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.8868357717700446e-06,
+      "loss": 0.2721,
+      "step": 108785
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.884872417113822e-06,
+      "loss": 0.2598,
+      "step": 108790
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.882909297670598e-06,
+      "loss": 0.2458,
+      "step": 108795
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.880946413465858e-06,
+      "loss": 0.2658,
+      "step": 108800
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.878983764525085e-06,
+      "loss": 0.2494,
+      "step": 108805
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.87702135087376e-06,
+      "loss": 0.256,
+      "step": 108810
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.875059172537379e-06,
+      "loss": 0.2424,
+      "step": 108815
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.873097229541406e-06,
+      "loss": 0.2528,
+      "step": 108820
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.871135521911315e-06,
+      "loss": 0.2658,
+      "step": 108825
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.869174049672589e-06,
+      "loss": 0.2566,
+      "step": 108830
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.8672128128506874e-06,
+      "loss": 0.2662,
+      "step": 108835
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.865251811471071e-06,
+      "loss": 0.2603,
+      "step": 108840
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.863291045559209e-06,
+      "loss": 0.2534,
+      "step": 108845
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.861330515140568e-06,
+      "loss": 0.257,
+      "step": 108850
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.859370220240594e-06,
+      "loss": 0.2316,
+      "step": 108855
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.857410160884745e-06,
+      "loss": 0.2485,
+      "step": 108860
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.8554503370984595e-06,
+      "loss": 0.2623,
+      "step": 108865
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.853490748907204e-06,
+      "loss": 0.264,
+      "step": 108870
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.8515313963364105e-06,
+      "loss": 0.2575,
+      "step": 108875
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.84957227941152e-06,
+      "loss": 0.269,
+      "step": 108880
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.847613398157965e-06,
+      "loss": 0.2572,
+      "step": 108885
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.845654752601197e-06,
+      "loss": 0.2508,
+      "step": 108890
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.843696342766628e-06,
+      "loss": 0.247,
+      "step": 108895
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.841738168679704e-06,
+      "loss": 0.2632,
+      "step": 108900
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.839780230365836e-06,
+      "loss": 0.2564,
+      "step": 108905
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.837822527850463e-06,
+      "loss": 0.2442,
+      "step": 108910
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.835865061158994e-06,
+      "loss": 0.2493,
+      "step": 108915
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.833907830316846e-06,
+      "loss": 0.2528,
+      "step": 108920
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.831950835349427e-06,
+      "loss": 0.2608,
+      "step": 108925
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.829994076282159e-06,
+      "loss": 0.2436,
+      "step": 108930
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.828037553140443e-06,
+      "loss": 0.2438,
+      "step": 108935
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.826081265949677e-06,
+      "loss": 0.2773,
+      "step": 108940
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.824125214735274e-06,
+      "loss": 0.2594,
+      "step": 108945
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.822169399522624e-06,
+      "loss": 0.2415,
+      "step": 108950
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.820213820337129e-06,
+      "loss": 0.272,
+      "step": 108955
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.81825847720417e-06,
+      "loss": 0.2607,
+      "step": 108960
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.816303370149149e-06,
+      "loss": 0.2469,
+      "step": 108965
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.814348499197445e-06,
+      "loss": 0.2684,
+      "step": 108970
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.812393864374442e-06,
+      "loss": 0.2724,
+      "step": 108975
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.810439465705512e-06,
+      "loss": 0.2911,
+      "step": 108980
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.808485303216044e-06,
+      "loss": 0.2469,
+      "step": 108985
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.806531376931408e-06,
+      "loss": 0.2637,
+      "step": 108990
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.80457768687697e-06,
+      "loss": 0.2581,
+      "step": 108995
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 6.802624233078092e-06,
+      "loss": 0.2617,
+      "step": 109000
+    },
+    {
+      "epoch": 3.83,
+      "eval_loss": 0.2511043846607208,
+      "eval_runtime": 10.5525,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 109000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.800671015560145e-06,
+      "loss": 0.2452,
+      "step": 109005
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7987180343485005e-06,
+      "loss": 0.25,
+      "step": 109010
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.796765289468507e-06,
+      "loss": 0.26,
+      "step": 109015
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7948127809455115e-06,
+      "loss": 0.2422,
+      "step": 109020
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.792860508804882e-06,
+      "loss": 0.2714,
+      "step": 109025
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.790908473071961e-06,
+      "loss": 0.3057,
+      "step": 109030
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7889566737720896e-06,
+      "loss": 0.2673,
+      "step": 109035
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.787005110930611e-06,
+      "loss": 0.2768,
+      "step": 109040
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.785053784572873e-06,
+      "loss": 0.2606,
+      "step": 109045
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.783102694724211e-06,
+      "loss": 0.2553,
+      "step": 109050
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.781151841409944e-06,
+      "loss": 0.2538,
+      "step": 109055
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7792012246554235e-06,
+      "loss": 0.2512,
+      "step": 109060
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7772508444859575e-06,
+      "loss": 0.2475,
+      "step": 109065
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.77530070092689e-06,
+      "loss": 0.2504,
+      "step": 109070
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.773350794003522e-06,
+      "loss": 0.2552,
+      "step": 109075
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.771401123741192e-06,
+      "loss": 0.2618,
+      "step": 109080
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7694516901652045e-06,
+      "loss": 0.2591,
+      "step": 109085
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.767502493300873e-06,
+      "loss": 0.2501,
+      "step": 109090
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.765553533173497e-06,
+      "loss": 0.2763,
+      "step": 109095
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7636048098084e-06,
+      "loss": 0.2461,
+      "step": 109100
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.761656323230875e-06,
+      "loss": 0.2704,
+      "step": 109105
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.759708073466217e-06,
+      "loss": 0.2636,
+      "step": 109110
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7577600605397365e-06,
+      "loss": 0.2704,
+      "step": 109115
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.755812284476712e-06,
+      "loss": 0.2373,
+      "step": 109120
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.753864745302449e-06,
+      "loss": 0.2521,
+      "step": 109125
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7519174430422275e-06,
+      "loss": 0.2599,
+      "step": 109130
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.749970377721323e-06,
+      "loss": 0.2376,
+      "step": 109135
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.748023549365034e-06,
+      "loss": 0.2462,
+      "step": 109140
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.746076957998631e-06,
+      "loss": 0.2321,
+      "step": 109145
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.744130603647392e-06,
+      "loss": 0.2397,
+      "step": 109150
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.742184486336576e-06,
+      "loss": 0.2366,
+      "step": 109155
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.740238606091468e-06,
+      "loss": 0.2545,
+      "step": 109160
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.738292962937323e-06,
+      "loss": 0.2597,
+      "step": 109165
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.736347556899414e-06,
+      "loss": 0.2432,
+      "step": 109170
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.734402388002992e-06,
+      "loss": 0.2496,
+      "step": 109175
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7324574562733225e-06,
+      "loss": 0.2766,
+      "step": 109180
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.730512761735655e-06,
+      "loss": 0.242,
+      "step": 109185
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.728568304415231e-06,
+      "loss": 0.262,
+      "step": 109190
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.726624084337313e-06,
+      "loss": 0.2766,
+      "step": 109195
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.724680101527139e-06,
+      "loss": 0.253,
+      "step": 109200
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.722736356009951e-06,
+      "loss": 0.2514,
+      "step": 109205
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.720792847810977e-06,
+      "loss": 0.2806,
+      "step": 109210
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.718849576955471e-06,
+      "loss": 0.2738,
+      "step": 109215
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.716906543468657e-06,
+      "loss": 0.2753,
+      "step": 109220
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.714963747375752e-06,
+      "loss": 0.2665,
+      "step": 109225
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.713021188701993e-06,
+      "loss": 0.2478,
+      "step": 109230
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.711078867472612e-06,
+      "loss": 0.2545,
+      "step": 109235
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.709136783712816e-06,
+      "loss": 0.2623,
+      "step": 109240
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.707194937447825e-06,
+      "loss": 0.2383,
+      "step": 109245
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.705253328702846e-06,
+      "loss": 0.2522,
+      "step": 109250
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.703311957503103e-06,
+      "loss": 0.2479,
+      "step": 109255
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.7013708238737954e-06,
+      "loss": 0.2549,
+      "step": 109260
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.699429927840131e-06,
+      "loss": 0.254,
+      "step": 109265
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.697489269427298e-06,
+      "loss": 0.2509,
+      "step": 109270
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.695548848660515e-06,
+      "loss": 0.249,
+      "step": 109275
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.693608665564957e-06,
+      "loss": 0.2779,
+      "step": 109280
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 6.691668720165836e-06,
+      "loss": 0.2439,
+      "step": 109285
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6897290124883216e-06,
+      "loss": 0.2579,
+      "step": 109290
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.687789542557618e-06,
+      "loss": 0.2478,
+      "step": 109295
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6858503103989e-06,
+      "loss": 0.2513,
+      "step": 109300
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.683911316037336e-06,
+      "loss": 0.2413,
+      "step": 109305
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.681972559498123e-06,
+      "loss": 0.2327,
+      "step": 109310
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.680034040806421e-06,
+      "loss": 0.256,
+      "step": 109315
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6780957599874075e-06,
+      "loss": 0.2294,
+      "step": 109320
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.676157717066237e-06,
+      "loss": 0.2734,
+      "step": 109325
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6742199120680885e-06,
+      "loss": 0.2602,
+      "step": 109330
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.672282345018111e-06,
+      "loss": 0.2599,
+      "step": 109335
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.670345015941476e-06,
+      "loss": 0.2409,
+      "step": 109340
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.668407924863324e-06,
+      "loss": 0.2501,
+      "step": 109345
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.666471071808822e-06,
+      "loss": 0.2568,
+      "step": 109350
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.664534456803109e-06,
+      "loss": 0.2652,
+      "step": 109355
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.662598079871335e-06,
+      "loss": 0.2532,
+      "step": 109360
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.660661941038632e-06,
+      "loss": 0.2475,
+      "step": 109365
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.658726040330155e-06,
+      "loss": 0.2356,
+      "step": 109370
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6567903777710326e-06,
+      "loss": 0.2341,
+      "step": 109375
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6548549533863995e-06,
+      "loss": 0.2409,
+      "step": 109380
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.652919767201377e-06,
+      "loss": 0.2327,
+      "step": 109385
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.650984819241099e-06,
+      "loss": 0.2545,
+      "step": 109390
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6490501095307e-06,
+      "loss": 0.2491,
+      "step": 109395
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.647115638095288e-06,
+      "loss": 0.2579,
+      "step": 109400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6451814049599825e-06,
+      "loss": 0.2572,
+      "step": 109405
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.643247410149903e-06,
+      "loss": 0.2585,
+      "step": 109410
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.641313653690162e-06,
+      "loss": 0.2468,
+      "step": 109415
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.639380135605855e-06,
+      "loss": 0.2684,
+      "step": 109420
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6374468559221025e-06,
+      "loss": 0.2608,
+      "step": 109425
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.635513814664004e-06,
+      "loss": 0.255,
+      "step": 109430
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6335810118566545e-06,
+      "loss": 0.2507,
+      "step": 109435
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.631648447525144e-06,
+      "loss": 0.2536,
+      "step": 109440
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6297161216945735e-06,
+      "loss": 0.2414,
+      "step": 109445
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6277840343900415e-06,
+      "loss": 0.2569,
+      "step": 109450
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.625852185636625e-06,
+      "loss": 0.2659,
+      "step": 109455
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.623920575459403e-06,
+      "loss": 0.2423,
+      "step": 109460
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6219892038834666e-06,
+      "loss": 0.2552,
+      "step": 109465
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.62005807093389e-06,
+      "loss": 0.2629,
+      "step": 109470
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.618127176635747e-06,
+      "loss": 0.2408,
+      "step": 109475
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6161965210141e-06,
+      "loss": 0.2507,
+      "step": 109480
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.614266104094035e-06,
+      "loss": 0.2448,
+      "step": 109485
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.612335925900606e-06,
+      "loss": 0.2761,
+      "step": 109490
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6104059864588705e-06,
+      "loss": 0.2613,
+      "step": 109495
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.608476285793902e-06,
+      "loss": 0.2673,
+      "step": 109500
+    },
+    {
+      "epoch": 3.85,
+      "eval_loss": 0.2506560981273651,
+      "eval_runtime": 10.5745,
+      "eval_samples_per_second": 9.457,
+      "eval_steps_per_second": 9.457,
+      "step": 109500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.60654682393074e-06,
+      "loss": 0.2786,
+      "step": 109505
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.6046176008944555e-06,
+      "loss": 0.2617,
+      "step": 109510
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.602688616710087e-06,
+      "loss": 0.2634,
+      "step": 109515
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.600759871402673e-06,
+      "loss": 0.251,
+      "step": 109520
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.598831364997276e-06,
+      "loss": 0.2717,
+      "step": 109525
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.596903097518928e-06,
+      "loss": 0.2611,
+      "step": 109530
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.594975068992657e-06,
+      "loss": 0.2771,
+      "step": 109535
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.593047279443512e-06,
+      "loss": 0.2661,
+      "step": 109540
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.5911197288965165e-06,
+      "loss": 0.2525,
+      "step": 109545
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.589192417376694e-06,
+      "loss": 0.2556,
+      "step": 109550
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.58726534490908e-06,
+      "loss": 0.2301,
+      "step": 109555
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.585338511518685e-06,
+      "loss": 0.2549,
+      "step": 109560
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.583411917230542e-06,
+      "loss": 0.2957,
+      "step": 109565
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.581485562069656e-06,
+      "loss": 0.227,
+      "step": 109570
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.579559446061037e-06,
+      "loss": 0.2694,
+      "step": 109575
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.577633569229705e-06,
+      "loss": 0.2556,
+      "step": 109580
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.575707931600661e-06,
+      "loss": 0.2774,
+      "step": 109585
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.573782533198908e-06,
+      "loss": 0.2739,
+      "step": 109590
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5718573740494365e-06,
+      "loss": 0.2586,
+      "step": 109595
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.569932454177261e-06,
+      "loss": 0.2453,
+      "step": 109600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.568007773607357e-06,
+      "loss": 0.2472,
+      "step": 109605
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.566083332364734e-06,
+      "loss": 0.277,
+      "step": 109610
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5641591304743625e-06,
+      "loss": 0.2531,
+      "step": 109615
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5622351679612424e-06,
+      "loss": 0.2533,
+      "step": 109620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.560311444850351e-06,
+      "loss": 0.2605,
+      "step": 109625
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.558387961166659e-06,
+      "loss": 0.2622,
+      "step": 109630
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.55646471693514e-06,
+      "loss": 0.2792,
+      "step": 109635
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.554541712180778e-06,
+      "loss": 0.2643,
+      "step": 109640
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5526189469285375e-06,
+      "loss": 0.2605,
+      "step": 109645
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5506964212033754e-06,
+      "loss": 0.2747,
+      "step": 109650
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.548774135030269e-06,
+      "loss": 0.2637,
+      "step": 109655
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.546852088434172e-06,
+      "loss": 0.2543,
+      "step": 109660
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.544930281440028e-06,
+      "loss": 0.2289,
+      "step": 109665
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.543008714072804e-06,
+      "loss": 0.2327,
+      "step": 109670
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.541087386357459e-06,
+      "loss": 0.2571,
+      "step": 109675
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.539166298318924e-06,
+      "loss": 0.2668,
+      "step": 109680
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5372454499821515e-06,
+      "loss": 0.246,
+      "step": 109685
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.53532484137207e-06,
+      "loss": 0.2797,
+      "step": 109690
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5334044725136375e-06,
+      "loss": 0.2634,
+      "step": 109695
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5314843434317764e-06,
+      "loss": 0.2428,
+      "step": 109700
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.529564454151418e-06,
+      "loss": 0.2481,
+      "step": 109705
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.527644804697489e-06,
+      "loss": 0.2572,
+      "step": 109710
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.525725395094923e-06,
+      "loss": 0.254,
+      "step": 109715
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5238062253686314e-06,
+      "loss": 0.2514,
+      "step": 109720
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5218872955435465e-06,
+      "loss": 0.2573,
+      "step": 109725
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.519968605644569e-06,
+      "loss": 0.2663,
+      "step": 109730
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.518050155696631e-06,
+      "loss": 0.2713,
+      "step": 109735
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.516131945724629e-06,
+      "loss": 0.279,
+      "step": 109740
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.514213975753472e-06,
+      "loss": 0.2773,
+      "step": 109745
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5122962458080586e-06,
+      "loss": 0.2522,
+      "step": 109750
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.510378755913302e-06,
+      "loss": 0.2602,
+      "step": 109755
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5084615060940905e-06,
+      "loss": 0.2345,
+      "step": 109760
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.5065444963753134e-06,
+      "loss": 0.2481,
+      "step": 109765
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.504627726781875e-06,
+      "loss": 0.2813,
+      "step": 109770
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.502711197338651e-06,
+      "loss": 0.2379,
+      "step": 109775
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.500794908070537e-06,
+      "loss": 0.247,
+      "step": 109780
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.498878859002405e-06,
+      "loss": 0.2438,
+      "step": 109785
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.496963050159144e-06,
+      "loss": 0.2464,
+      "step": 109790
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.495047481565625e-06,
+      "loss": 0.253,
+      "step": 109795
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.493132153246717e-06,
+      "loss": 0.2706,
+      "step": 109800
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.491217065227287e-06,
+      "loss": 0.2471,
+      "step": 109805
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.489302217532212e-06,
+      "loss": 0.25,
+      "step": 109810
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.487387610186346e-06,
+      "loss": 0.2396,
+      "step": 109815
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.485473243214554e-06,
+      "loss": 0.2484,
+      "step": 109820
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.483559116641683e-06,
+      "loss": 0.2551,
+      "step": 109825
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.481645230492594e-06,
+      "loss": 0.272,
+      "step": 109830
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.479731584792145e-06,
+      "loss": 0.2526,
+      "step": 109835
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.477818179565176e-06,
+      "loss": 0.2607,
+      "step": 109840
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.475905014836525e-06,
+      "loss": 0.2703,
+      "step": 109845
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.4739920906310456e-06,
+      "loss": 0.2606,
+      "step": 109850
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.472079406973569e-06,
+      "loss": 0.2238,
+      "step": 109855
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.470166963888932e-06,
+      "loss": 0.2683,
+      "step": 109860
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.4682547614019596e-06,
+      "loss": 0.2836,
+      "step": 109865
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.466342799537492e-06,
+      "loss": 0.2543,
+      "step": 109870
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.4644310783203485e-06,
+      "loss": 0.2591,
+      "step": 109875
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.462519597775346e-06,
+      "loss": 0.2443,
+      "step": 109880
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.460608357927309e-06,
+      "loss": 0.2622,
+      "step": 109885
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.458697358801061e-06,
+      "loss": 0.2791,
+      "step": 109890
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.456786600421411e-06,
+      "loss": 0.2643,
+      "step": 109895
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.454876082813158e-06,
+      "loss": 0.2539,
+      "step": 109900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.452965806001124e-06,
+      "loss": 0.2467,
+      "step": 109905
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.451055770010106e-06,
+      "loss": 0.277,
+      "step": 109910
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.449145974864906e-06,
+      "loss": 0.2755,
+      "step": 109915
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.447236420590311e-06,
+      "loss": 0.2534,
+      "step": 109920
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.445327107211132e-06,
+      "loss": 0.2553,
+      "step": 109925
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.443418034752155e-06,
+      "loss": 0.277,
+      "step": 109930
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.441509203238155e-06,
+      "loss": 0.2739,
+      "step": 109935
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.439600612693935e-06,
+      "loss": 0.2784,
+      "step": 109940
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.437692263144266e-06,
+      "loss": 0.2548,
+      "step": 109945
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.435784154613933e-06,
+      "loss": 0.2584,
+      "step": 109950
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.43387628712771e-06,
+      "loss": 0.2587,
+      "step": 109955
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.431968660710361e-06,
+      "loss": 0.2411,
+      "step": 109960
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.430061275386667e-06,
+      "loss": 0.2401,
+      "step": 109965
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.428154131181393e-06,
+      "loss": 0.2648,
+      "step": 109970
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.426247228119297e-06,
+      "loss": 0.252,
+      "step": 109975
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.424340566225132e-06,
+      "loss": 0.2508,
+      "step": 109980
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.422434145523671e-06,
+      "loss": 0.2496,
+      "step": 109985
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.420527966039652e-06,
+      "loss": 0.2349,
+      "step": 109990
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.418622027797838e-06,
+      "loss": 0.2563,
+      "step": 109995
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.416716330822967e-06,
+      "loss": 0.2675,
+      "step": 110000
+    },
+    {
+      "epoch": 3.87,
+      "eval_loss": 0.2505649924278259,
+      "eval_runtime": 10.5578,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 9.472,
+      "step": 110000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.414810875139796e-06,
+      "loss": 0.2671,
+      "step": 110005
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.4129056607730555e-06,
+      "loss": 0.2459,
+      "step": 110010
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.411000687747478e-06,
+      "loss": 0.2513,
+      "step": 110015
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.409095956087813e-06,
+      "loss": 0.2715,
+      "step": 110020
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.407191465818788e-06,
+      "loss": 0.2547,
+      "step": 110025
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.405287216965125e-06,
+      "loss": 0.2419,
+      "step": 110030
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.403383209551547e-06,
+      "loss": 0.2474,
+      "step": 110035
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.401479443602792e-06,
+      "loss": 0.2438,
+      "step": 110040
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.39957591914356e-06,
+      "loss": 0.2548,
+      "step": 110045
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.397672636198584e-06,
+      "loss": 0.2529,
+      "step": 110050
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.395769594792564e-06,
+      "loss": 0.2649,
+      "step": 110055
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.3938667949502205e-06,
+      "loss": 0.2417,
+      "step": 110060
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.3919642366962545e-06,
+      "loss": 0.2702,
+      "step": 110065
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.3900619200553705e-06,
+      "loss": 0.2615,
+      "step": 110070
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.38815984505226e-06,
+      "loss": 0.2561,
+      "step": 110075
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.386258011711638e-06,
+      "loss": 0.2494,
+      "step": 110080
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.384356420058188e-06,
+      "loss": 0.2514,
+      "step": 110085
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.382455070116599e-06,
+      "loss": 0.2737,
+      "step": 110090
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.380553961911556e-06,
+      "loss": 0.2449,
+      "step": 110095
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.378653095467757e-06,
+      "loss": 0.2506,
+      "step": 110100
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.376752470809869e-06,
+      "loss": 0.2576,
+      "step": 110105
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.374852087962574e-06,
+      "loss": 0.243,
+      "step": 110110
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.372951946950559e-06,
+      "loss": 0.2739,
+      "step": 110115
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.37105204779849e-06,
+      "loss": 0.2525,
+      "step": 110120
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.369152390531031e-06,
+      "loss": 0.2615,
+      "step": 110125
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.367252975172841e-06,
+      "loss": 0.2457,
+      "step": 110130
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 6.3653538017486e-06,
+      "loss": 0.2682,
+      "step": 110135
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.363454870282959e-06,
+      "loss": 0.266,
+      "step": 110140
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.361556180800577e-06,
+      "loss": 0.2501,
+      "step": 110145
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.3596577333260935e-06,
+      "loss": 0.261,
+      "step": 110150
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.35775952788418e-06,
+      "loss": 0.2517,
+      "step": 110155
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.355861564499466e-06,
+      "loss": 0.283,
+      "step": 110160
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.353963843196609e-06,
+      "loss": 0.2687,
+      "step": 110165
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.352066364000236e-06,
+      "loss": 0.2655,
+      "step": 110170
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.350169126935001e-06,
+      "loss": 0.2467,
+      "step": 110175
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.348272132025526e-06,
+      "loss": 0.2478,
+      "step": 110180
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.3463753792964485e-06,
+      "loss": 0.2508,
+      "step": 110185
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.344478868772386e-06,
+      "loss": 0.2584,
+      "step": 110190
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.342582600477975e-06,
+      "loss": 0.2522,
+      "step": 110195
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.340686574437835e-06,
+      "loss": 0.2353,
+      "step": 110200
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.338790790676585e-06,
+      "loss": 0.2505,
+      "step": 110205
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.336895249218827e-06,
+      "loss": 0.2479,
+      "step": 110210
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.334999950089188e-06,
+      "loss": 0.2476,
+      "step": 110215
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.33310489331228e-06,
+      "loss": 0.2702,
+      "step": 110220
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.3312100789126975e-06,
+      "loss": 0.2398,
+      "step": 110225
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.329315506915054e-06,
+      "loss": 0.2507,
+      "step": 110230
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.327421177343945e-06,
+      "loss": 0.2615,
+      "step": 110235
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.325527090223965e-06,
+      "loss": 0.2568,
+      "step": 110240
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.323633245579702e-06,
+      "loss": 0.2687,
+      "step": 110245
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.321739643435762e-06,
+      "loss": 0.2595,
+      "step": 110250
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.319846283816721e-06,
+      "loss": 0.2578,
+      "step": 110255
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.317953166747168e-06,
+      "loss": 0.2479,
+      "step": 110260
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.316060292251675e-06,
+      "loss": 0.2583,
+      "step": 110265
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.3141676603548226e-06,
+      "loss": 0.2528,
+      "step": 110270
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.312275271081197e-06,
+      "loss": 0.2585,
+      "step": 110275
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.31038312445536e-06,
+      "loss": 0.2335,
+      "step": 110280
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.308491220501874e-06,
+      "loss": 0.2517,
+      "step": 110285
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.306599559245322e-06,
+      "loss": 0.2758,
+      "step": 110290
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.304708140710253e-06,
+      "loss": 0.2717,
+      "step": 110295
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.302816964921226e-06,
+      "loss": 0.2278,
+      "step": 110300
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.300926031902796e-06,
+      "loss": 0.2578,
+      "step": 110305
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.299035341679524e-06,
+      "loss": 0.2574,
+      "step": 110310
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.297144894275952e-06,
+      "loss": 0.2487,
+      "step": 110315
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.29525468971662e-06,
+      "loss": 0.2582,
+      "step": 110320
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.293364728026082e-06,
+      "loss": 0.2558,
+      "step": 110325
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.291475009228881e-06,
+      "loss": 0.2433,
+      "step": 110330
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.289585533349548e-06,
+      "loss": 0.2345,
+      "step": 110335
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.287696300412607e-06,
+      "loss": 0.2536,
+      "step": 110340
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.2858073104426066e-06,
+      "loss": 0.2563,
+      "step": 110345
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.283918563464067e-06,
+      "loss": 0.2594,
+      "step": 110350
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.282030059501506e-06,
+      "loss": 0.2484,
+      "step": 110355
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.280141798579445e-06,
+      "loss": 0.2491,
+      "step": 110360
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.2782537807224125e-06,
+      "loss": 0.2373,
+      "step": 110365
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.276366005954917e-06,
+      "loss": 0.2352,
+      "step": 110370
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.274478474301463e-06,
+      "loss": 0.2756,
+      "step": 110375
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.27259118578657e-06,
+      "loss": 0.2509,
+      "step": 110380
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.270704140434733e-06,
+      "loss": 0.2575,
+      "step": 110385
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.268817338270466e-06,
+      "loss": 0.2689,
+      "step": 110390
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.266930779318262e-06,
+      "loss": 0.2609,
+      "step": 110395
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.26504446360261e-06,
+      "loss": 0.2579,
+      "step": 110400
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.2631583911480164e-06,
+      "loss": 0.2455,
+      "step": 110405
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.261272561978962e-06,
+      "loss": 0.2613,
+      "step": 110410
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.259386976119932e-06,
+      "loss": 0.2372,
+      "step": 110415
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 6.257501633595408e-06,
+      "loss": 0.2716,
+      "step": 110420
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.2556165344298785e-06,
+      "loss": 0.2526,
+      "step": 110425
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.253731678647809e-06,
+      "loss": 0.2588,
+      "step": 110430
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.251847066273686e-06,
+      "loss": 0.2479,
+      "step": 110435
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.249962697331968e-06,
+      "loss": 0.2687,
+      "step": 110440
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.248078571847133e-06,
+      "loss": 0.2849,
+      "step": 110445
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.24619468984364e-06,
+      "loss": 0.2562,
+      "step": 110450
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.244311051345941e-06,
+      "loss": 0.2657,
+      "step": 110455
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.242427656378514e-06,
+      "loss": 0.2758,
+      "step": 110460
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.240544504965798e-06,
+      "loss": 0.2533,
+      "step": 110465
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.23866159713225e-06,
+      "loss": 0.2474,
+      "step": 110470
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.23677893290231e-06,
+      "loss": 0.2443,
+      "step": 110475
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.234896512300437e-06,
+      "loss": 0.2682,
+      "step": 110480
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.233014335351059e-06,
+      "loss": 0.2568,
+      "step": 110485
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.231132402078629e-06,
+      "loss": 0.2572,
+      "step": 110490
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.22925071250757e-06,
+      "loss": 0.262,
+      "step": 110495
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.2273692666623255e-06,
+      "loss": 0.269,
+      "step": 110500
+    },
+    {
+      "epoch": 3.89,
+      "eval_loss": 0.25054579973220825,
+      "eval_runtime": 10.5376,
+      "eval_samples_per_second": 9.49,
+      "eval_steps_per_second": 9.49,
+      "step": 110500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.225488064567322e-06,
+      "loss": 0.2626,
+      "step": 110505
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.223607106246981e-06,
+      "loss": 0.2529,
+      "step": 110510
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.2217263917257215e-06,
+      "loss": 0.2658,
+      "step": 110515
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.219845921027975e-06,
+      "loss": 0.2449,
+      "step": 110520
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.217965694178152e-06,
+      "loss": 0.263,
+      "step": 110525
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.21608571120067e-06,
+      "loss": 0.2426,
+      "step": 110530
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.214205972119924e-06,
+      "loss": 0.2637,
+      "step": 110535
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.212326476960345e-06,
+      "loss": 0.267,
+      "step": 110540
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.210447225746316e-06,
+      "loss": 0.2568,
+      "step": 110545
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.208568218502253e-06,
+      "loss": 0.2487,
+      "step": 110550
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.206689455252543e-06,
+      "loss": 0.2381,
+      "step": 110555
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.20481093602159e-06,
+      "loss": 0.2342,
+      "step": 110560
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.202932660833783e-06,
+      "loss": 0.2762,
+      "step": 110565
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.201054629713507e-06,
+      "loss": 0.2464,
+      "step": 110570
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.199176842685139e-06,
+      "loss": 0.2611,
+      "step": 110575
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.197299299773077e-06,
+      "loss": 0.2453,
+      "step": 110580
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.195422001001694e-06,
+      "loss": 0.2531,
+      "step": 110585
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.193544946395355e-06,
+      "loss": 0.2641,
+      "step": 110590
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.191668135978448e-06,
+      "loss": 0.2675,
+      "step": 110595
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.189791569775327e-06,
+      "loss": 0.2526,
+      "step": 110600
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.187915247810375e-06,
+      "loss": 0.2349,
+      "step": 110605
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.1860391701079376e-06,
+      "loss": 0.253,
+      "step": 110610
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.184163336692392e-06,
+      "loss": 0.2703,
+      "step": 110615
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.182287747588084e-06,
+      "loss": 0.2514,
+      "step": 110620
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.180412402819366e-06,
+      "loss": 0.2467,
+      "step": 110625
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.178537302410587e-06,
+      "loss": 0.2518,
+      "step": 110630
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.1766624463861e-06,
+      "loss": 0.2674,
+      "step": 110635
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.174787834770251e-06,
+      "loss": 0.2545,
+      "step": 110640
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.172913467587374e-06,
+      "loss": 0.259,
+      "step": 110645
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.171039344861798e-06,
+      "loss": 0.2691,
+      "step": 110650
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.169165466617871e-06,
+      "loss": 0.2591,
+      "step": 110655
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.167291832879926e-06,
+      "loss": 0.2503,
+      "step": 110660
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.165418443672285e-06,
+      "loss": 0.2583,
+      "step": 110665
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.163545299019266e-06,
+      "loss": 0.2442,
+      "step": 110670
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.1616723989452064e-06,
+      "loss": 0.282,
+      "step": 110675
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.159799743474415e-06,
+      "loss": 0.2496,
+      "step": 110680
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.1579273326312085e-06,
+      "loss": 0.2592,
+      "step": 110685
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.156055166439889e-06,
+      "loss": 0.248,
+      "step": 110690
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.154183244924783e-06,
+      "loss": 0.271,
+      "step": 110695
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.152311568110186e-06,
+      "loss": 0.2695,
+      "step": 110700
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 6.150440136020397e-06,
+      "loss": 0.2529,
+      "step": 110705
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.14856894867972e-06,
+      "loss": 0.2594,
+      "step": 110710
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.1466980061124586e-06,
+      "loss": 0.2701,
+      "step": 110715
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.144827308342899e-06,
+      "loss": 0.2438,
+      "step": 110720
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.142956855395324e-06,
+      "loss": 0.2585,
+      "step": 110725
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.141086647294034e-06,
+      "loss": 0.2527,
+      "step": 110730
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.139216684063307e-06,
+      "loss": 0.2485,
+      "step": 110735
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.137346965727417e-06,
+      "loss": 0.2606,
+      "step": 110740
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.135477492310643e-06,
+      "loss": 0.257,
+      "step": 110745
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.133608263837265e-06,
+      "loss": 0.2391,
+      "step": 110750
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.131739280331553e-06,
+      "loss": 0.2296,
+      "step": 110755
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.129870541817765e-06,
+      "loss": 0.2693,
+      "step": 110760
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.128002048320169e-06,
+      "loss": 0.2424,
+      "step": 110765
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.12613379986304e-06,
+      "loss": 0.2491,
+      "step": 110770
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.124265796470624e-06,
+      "loss": 0.2508,
+      "step": 110775
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.122398038167176e-06,
+      "loss": 0.2627,
+      "step": 110780
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.120530524976942e-06,
+      "loss": 0.2383,
+      "step": 110785
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.118663256924187e-06,
+      "loss": 0.262,
+      "step": 110790
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.116796234033143e-06,
+      "loss": 0.2547,
+      "step": 110795
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.114929456328056e-06,
+      "loss": 0.2659,
+      "step": 110800
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.113062923833157e-06,
+      "loss": 0.2456,
+      "step": 110805
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.111196636572697e-06,
+      "loss": 0.2617,
+      "step": 110810
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.109330594570892e-06,
+      "loss": 0.2244,
+      "step": 110815
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.107464797851989e-06,
+      "loss": 0.2801,
+      "step": 110820
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.105599246440197e-06,
+      "loss": 0.2624,
+      "step": 110825
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.103733940359754e-06,
+      "loss": 0.2427,
+      "step": 110830
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.101868879634873e-06,
+      "loss": 0.265,
+      "step": 110835
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.10000406428976e-06,
+      "loss": 0.2538,
+      "step": 110840
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.098139494348651e-06,
+      "loss": 0.249,
+      "step": 110845
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.096275169835742e-06,
+      "loss": 0.2571,
+      "step": 110850
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.09441109077524e-06,
+      "loss": 0.2541,
+      "step": 110855
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.092547257191344e-06,
+      "loss": 0.2631,
+      "step": 110860
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.0906836691082685e-06,
+      "loss": 0.2344,
+      "step": 110865
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.088820326550199e-06,
+      "loss": 0.2398,
+      "step": 110870
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.0869572295413405e-06,
+      "loss": 0.2803,
+      "step": 110875
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.085094378105868e-06,
+      "loss": 0.2657,
+      "step": 110880
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.083231772267991e-06,
+      "loss": 0.2532,
+      "step": 110885
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.081369412051882e-06,
+      "loss": 0.2559,
+      "step": 110890
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.079507297481723e-06,
+      "loss": 0.2637,
+      "step": 110895
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.0776454285816856e-06,
+      "loss": 0.2566,
+      "step": 110900
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.075783805375956e-06,
+      "loss": 0.2534,
+      "step": 110905
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.073922427888707e-06,
+      "loss": 0.2611,
+      "step": 110910
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.072061296144099e-06,
+      "loss": 0.2659,
+      "step": 110915
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.070200410166297e-06,
+      "loss": 0.2797,
+      "step": 110920
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.068339769979465e-06,
+      "loss": 0.2757,
+      "step": 110925
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.066479375607773e-06,
+      "loss": 0.2475,
+      "step": 110930
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.06461922707536e-06,
+      "loss": 0.2344,
+      "step": 110935
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.062759324406397e-06,
+      "loss": 0.2698,
+      "step": 110940
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.060899667625023e-06,
+      "loss": 0.2541,
+      "step": 110945
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.059040256755386e-06,
+      "loss": 0.2707,
+      "step": 110950
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.057181091821621e-06,
+      "loss": 0.2364,
+      "step": 110955
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.055322172847882e-06,
+      "loss": 0.2434,
+      "step": 110960
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.0534634998583e-06,
+      "loss": 0.2586,
+      "step": 110965
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.051605072877009e-06,
+      "loss": 0.2427,
+      "step": 110970
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.049746891928132e-06,
+      "loss": 0.2499,
+      "step": 110975
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.047888957035808e-06,
+      "loss": 0.2687,
+      "step": 110980
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.04603126822415e-06,
+      "loss": 0.2442,
+      "step": 110985
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 6.0441738255172905e-06,
+      "loss": 0.2566,
+      "step": 110990
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.042316628939337e-06,
+      "loss": 0.2521,
+      "step": 110995
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.0404596785144155e-06,
+      "loss": 0.2526,
+      "step": 111000
+    },
+    {
+      "epoch": 3.91,
+      "eval_loss": 0.25022944808006287,
+      "eval_runtime": 10.5381,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 111000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.038602974266627e-06,
+      "loss": 0.2202,
+      "step": 111005
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.036746516220088e-06,
+      "loss": 0.2684,
+      "step": 111010
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.034890304398888e-06,
+      "loss": 0.2788,
+      "step": 111015
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.033034338827146e-06,
+      "loss": 0.2533,
+      "step": 111020
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.031178619528951e-06,
+      "loss": 0.2626,
+      "step": 111025
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.0293231465283986e-06,
+      "loss": 0.2392,
+      "step": 111030
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.0274679198495856e-06,
+      "loss": 0.236,
+      "step": 111035
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.025612939516595e-06,
+      "loss": 0.2892,
+      "step": 111040
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.0237582055535205e-06,
+      "loss": 0.2554,
+      "step": 111045
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.021903717984434e-06,
+      "loss": 0.2808,
+      "step": 111050
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.020049476833428e-06,
+      "loss": 0.2824,
+      "step": 111055
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.0181954821245695e-06,
+      "loss": 0.2596,
+      "step": 111060
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.016341733881936e-06,
+      "loss": 0.2509,
+      "step": 111065
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.014488232129584e-06,
+      "loss": 0.2612,
+      "step": 111070
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.012634976891598e-06,
+      "loss": 0.2241,
+      "step": 111075
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.010781968192037e-06,
+      "loss": 0.2892,
+      "step": 111080
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.008929206054956e-06,
+      "loss": 0.2441,
+      "step": 111085
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.007076690504404e-06,
+      "loss": 0.2606,
+      "step": 111090
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.0052244215644484e-06,
+      "loss": 0.2838,
+      "step": 111095
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.003372399259138e-06,
+      "loss": 0.2421,
+      "step": 111100
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 6.001520623612522e-06,
+      "loss": 0.2662,
+      "step": 111105
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.999669094648633e-06,
+      "loss": 0.2496,
+      "step": 111110
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.997817812391523e-06,
+      "loss": 0.2536,
+      "step": 111115
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.995966776865228e-06,
+      "loss": 0.2571,
+      "step": 111120
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.994115988093782e-06,
+      "loss": 0.2573,
+      "step": 111125
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.992265446101203e-06,
+      "loss": 0.2578,
+      "step": 111130
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.990415150911541e-06,
+      "loss": 0.297,
+      "step": 111135
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.98856510254881e-06,
+      "loss": 0.2555,
+      "step": 111140
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.986715301037024e-06,
+      "loss": 0.2434,
+      "step": 111145
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.984865746400211e-06,
+      "loss": 0.266,
+      "step": 111150
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.983016438662392e-06,
+      "loss": 0.2509,
+      "step": 111155
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.981167377847571e-06,
+      "loss": 0.2096,
+      "step": 111160
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.97931856397975e-06,
+      "loss": 0.2772,
+      "step": 111165
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.977469997082949e-06,
+      "loss": 0.2639,
+      "step": 111170
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.975621677181165e-06,
+      "loss": 0.2814,
+      "step": 111175
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.973773604298397e-06,
+      "loss": 0.2605,
+      "step": 111180
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.9719257784586316e-06,
+      "loss": 0.265,
+      "step": 111185
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.970078199685877e-06,
+      "loss": 0.2361,
+      "step": 111190
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.968230868004115e-06,
+      "loss": 0.258,
+      "step": 111195
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.966383783437329e-06,
+      "loss": 0.2375,
+      "step": 111200
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.964536946009502e-06,
+      "loss": 0.241,
+      "step": 111205
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.962690355744627e-06,
+      "loss": 0.2537,
+      "step": 111210
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.960844012666672e-06,
+      "loss": 0.2603,
+      "step": 111215
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.958997916799611e-06,
+      "loss": 0.2443,
+      "step": 111220
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.957152068167404e-06,
+      "loss": 0.2563,
+      "step": 111225
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.9553064667940366e-06,
+      "loss": 0.2483,
+      "step": 111230
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.953461112703462e-06,
+      "loss": 0.2582,
+      "step": 111235
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.951616005919644e-06,
+      "loss": 0.2488,
+      "step": 111240
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.949771146466532e-06,
+      "loss": 0.2573,
+      "step": 111245
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.947926534368093e-06,
+      "loss": 0.2767,
+      "step": 111250
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.9460821696482645e-06,
+      "loss": 0.2423,
+      "step": 111255
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.944238052331011e-06,
+      "loss": 0.2467,
+      "step": 111260
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.942394182440261e-06,
+      "loss": 0.2543,
+      "step": 111265
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.940550559999969e-06,
+      "loss": 0.2854,
+      "step": 111270
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.938707185034067e-06,
+      "loss": 0.2625,
+      "step": 111275
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.936864057566485e-06,
+      "loss": 0.255,
+      "step": 111280
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.9350211776211665e-06,
+      "loss": 0.2553,
+      "step": 111285
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.933178545222034e-06,
+      "loss": 0.2503,
+      "step": 111290
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.931336160393014e-06,
+      "loss": 0.2601,
+      "step": 111295
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.929494023158016e-06,
+      "loss": 0.2451,
+      "step": 111300
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.927652133540981e-06,
+      "loss": 0.2553,
+      "step": 111305
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.9258104915658054e-06,
+      "loss": 0.2729,
+      "step": 111310
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.9239690972564184e-06,
+      "loss": 0.2575,
+      "step": 111315
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.922127950636711e-06,
+      "loss": 0.2593,
+      "step": 111320
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.920287051730608e-06,
+      "loss": 0.2671,
+      "step": 111325
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.918446400562005e-06,
+      "loss": 0.2497,
+      "step": 111330
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.9166059971548e-06,
+      "loss": 0.2913,
+      "step": 111335
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.914765841532883e-06,
+      "loss": 0.2321,
+      "step": 111340
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.912925933720159e-06,
+      "loss": 0.2588,
+      "step": 111345
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.911086273740512e-06,
+      "loss": 0.2605,
+      "step": 111350
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.909246861617834e-06,
+      "loss": 0.2494,
+      "step": 111355
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.907407697375994e-06,
+      "loss": 0.2646,
+      "step": 111360
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.905568781038884e-06,
+      "loss": 0.2306,
+      "step": 111365
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.903730112630385e-06,
+      "loss": 0.2534,
+      "step": 111370
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.901891692174366e-06,
+      "loss": 0.258,
+      "step": 111375
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.90005351969469e-06,
+      "loss": 0.2528,
+      "step": 111380
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.898215595215239e-06,
+      "loss": 0.2637,
+      "step": 111385
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.896377918759871e-06,
+      "loss": 0.2638,
+      "step": 111390
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.8945404903524404e-06,
+      "loss": 0.268,
+      "step": 111395
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.892703310016814e-06,
+      "loss": 0.2561,
+      "step": 111400
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.890866377776844e-06,
+      "loss": 0.247,
+      "step": 111405
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.889029693656381e-06,
+      "loss": 0.2453,
+      "step": 111410
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.887193257679266e-06,
+      "loss": 0.2482,
+      "step": 111415
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.88535706986936e-06,
+      "loss": 0.2626,
+      "step": 111420
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.883521130250485e-06,
+      "loss": 0.2708,
+      "step": 111425
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.881685438846498e-06,
+      "loss": 0.253,
+      "step": 111430
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.879849995681219e-06,
+      "loss": 0.2609,
+      "step": 111435
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.878014800778495e-06,
+      "loss": 0.2458,
+      "step": 111440
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.876179854162145e-06,
+      "loss": 0.2702,
+      "step": 111445
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.874345155855998e-06,
+      "loss": 0.277,
+      "step": 111450
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.872510705883868e-06,
+      "loss": 0.2534,
+      "step": 111455
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.870676504269587e-06,
+      "loss": 0.275,
+      "step": 111460
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.868842551036966e-06,
+      "loss": 0.2626,
+      "step": 111465
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.867008846209809e-06,
+      "loss": 0.2749,
+      "step": 111470
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.865175389811942e-06,
+      "loss": 0.2634,
+      "step": 111475
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.8633421818671515e-06,
+      "loss": 0.2567,
+      "step": 111480
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.861509222399261e-06,
+      "loss": 0.2652,
+      "step": 111485
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.8596765114320605e-06,
+      "loss": 0.2739,
+      "step": 111490
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.8578440489893385e-06,
+      "loss": 0.2396,
+      "step": 111495
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.856011835094902e-06,
+      "loss": 0.2528,
+      "step": 111500
+    },
+    {
+      "epoch": 3.92,
+      "eval_loss": 0.2502204477787018,
+      "eval_runtime": 10.5596,
+      "eval_samples_per_second": 9.47,
+      "eval_steps_per_second": 9.47,
+      "step": 111500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.8541798697725384e-06,
+      "loss": 0.2533,
+      "step": 111505
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.852348153046022e-06,
+      "loss": 0.2447,
+      "step": 111510
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.8505166849391555e-06,
+      "loss": 0.259,
+      "step": 111515
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.848685465475709e-06,
+      "loss": 0.2621,
+      "step": 111520
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.84685449467946e-06,
+      "loss": 0.2557,
+      "step": 111525
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.845023772574177e-06,
+      "loss": 0.2671,
+      "step": 111530
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.843193299183636e-06,
+      "loss": 0.2392,
+      "step": 111535
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.841363074531611e-06,
+      "loss": 0.2612,
+      "step": 111540
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.8398990739176875e-06,
+      "loss": 0.2569,
+      "step": 111545
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.838069297054866e-06,
+      "loss": 0.2386,
+      "step": 111550
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.836239768997084e-06,
+      "loss": 0.2383,
+      "step": 111555
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.834410489768097e-06,
+      "loss": 0.252,
+      "step": 111560
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.832581459391662e-06,
+      "loss": 0.2548,
+      "step": 111565
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.830752677891521e-06,
+      "loss": 0.2536,
+      "step": 111570
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.8289241452914285e-06,
+      "loss": 0.246,
+      "step": 111575
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.827095861615114e-06,
+      "loss": 0.2682,
+      "step": 111580
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.825267826886333e-06,
+      "loss": 0.2444,
+      "step": 111585
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.823440041128813e-06,
+      "loss": 0.2722,
+      "step": 111590
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.821612504366286e-06,
+      "loss": 0.2466,
+      "step": 111595
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.819785216622473e-06,
+      "loss": 0.2542,
+      "step": 111600
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.817958177921118e-06,
+      "loss": 0.2506,
+      "step": 111605
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.816131388285934e-06,
+      "loss": 0.2846,
+      "step": 111610
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.814304847740631e-06,
+      "loss": 0.2556,
+      "step": 111615
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.812478556308948e-06,
+      "loss": 0.2741,
+      "step": 111620
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.810652514014575e-06,
+      "loss": 0.2559,
+      "step": 111625
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.808826720881242e-06,
+      "loss": 0.236,
+      "step": 111630
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.807001176932644e-06,
+      "loss": 0.2694,
+      "step": 111635
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.805175882192484e-06,
+      "loss": 0.259,
+      "step": 111640
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.8033508366844705e-06,
+      "loss": 0.2497,
+      "step": 111645
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.801526040432295e-06,
+      "loss": 0.2416,
+      "step": 111650
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.79970149345965e-06,
+      "loss": 0.243,
+      "step": 111655
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.797877195790222e-06,
+      "loss": 0.2511,
+      "step": 111660
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.796053147447711e-06,
+      "loss": 0.2293,
+      "step": 111665
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.794229348455796e-06,
+      "loss": 0.2502,
+      "step": 111670
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.792405798838147e-06,
+      "loss": 0.2516,
+      "step": 111675
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.790582498618449e-06,
+      "loss": 0.2522,
+      "step": 111680
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.788759447820386e-06,
+      "loss": 0.2582,
+      "step": 111685
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.786936646467622e-06,
+      "loss": 0.2405,
+      "step": 111690
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.785114094583815e-06,
+      "loss": 0.2332,
+      "step": 111695
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.783291792192647e-06,
+      "loss": 0.2697,
+      "step": 111700
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.781469739317769e-06,
+      "loss": 0.2611,
+      "step": 111705
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.779647935982843e-06,
+      "loss": 0.2595,
+      "step": 111710
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.777826382211513e-06,
+      "loss": 0.2782,
+      "step": 111715
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.776005078027447e-06,
+      "loss": 0.2558,
+      "step": 111720
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.774184023454287e-06,
+      "loss": 0.2604,
+      "step": 111725
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.77236321851567e-06,
+      "loss": 0.2573,
+      "step": 111730
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.7705426632352426e-06,
+      "loss": 0.2599,
+      "step": 111735
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.768722357636658e-06,
+      "loss": 0.2361,
+      "step": 111740
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.766902301743534e-06,
+      "loss": 0.2601,
+      "step": 111745
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.765082495579513e-06,
+      "loss": 0.2411,
+      "step": 111750
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.76326293916821e-06,
+      "loss": 0.2592,
+      "step": 111755
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.761443632533267e-06,
+      "loss": 0.2562,
+      "step": 111760
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.759624575698302e-06,
+      "loss": 0.2676,
+      "step": 111765
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.75780576868693e-06,
+      "loss": 0.2544,
+      "step": 111770
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.755987211522765e-06,
+      "loss": 0.2397,
+      "step": 111775
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.754168904229426e-06,
+      "loss": 0.2674,
+      "step": 111780
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.7523508468305175e-06,
+      "loss": 0.2483,
+      "step": 111785
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.750533039349654e-06,
+      "loss": 0.255,
+      "step": 111790
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.748715481810427e-06,
+      "loss": 0.2782,
+      "step": 111795
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.746898174236448e-06,
+      "loss": 0.2647,
+      "step": 111800
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.745081116651307e-06,
+      "loss": 0.2554,
+      "step": 111805
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.7432643090786e-06,
+      "loss": 0.2611,
+      "step": 111810
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.741447751541909e-06,
+      "loss": 0.2397,
+      "step": 111815
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.739631444064833e-06,
+      "loss": 0.2665,
+      "step": 111820
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.7378153866709476e-06,
+      "loss": 0.2722,
+      "step": 111825
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.73599957938383e-06,
+      "loss": 0.2563,
+      "step": 111830
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.734184022227071e-06,
+      "loss": 0.2429,
+      "step": 111835
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.732368715224226e-06,
+      "loss": 0.2568,
+      "step": 111840
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.730553658398885e-06,
+      "loss": 0.2663,
+      "step": 111845
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.728738851774598e-06,
+      "loss": 0.2366,
+      "step": 111850
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.7269242953749416e-06,
+      "loss": 0.2618,
+      "step": 111855
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.725109989223474e-06,
+      "loss": 0.2438,
+      "step": 111860
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.72329593334375e-06,
+      "loss": 0.237,
+      "step": 111865
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.721482127759317e-06,
+      "loss": 0.2486,
+      "step": 111870
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.719668572493742e-06,
+      "loss": 0.2308,
+      "step": 111875
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.717855267570563e-06,
+      "loss": 0.2412,
+      "step": 111880
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.716042213013326e-06,
+      "loss": 0.2283,
+      "step": 111885
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.714229408845567e-06,
+      "loss": 0.2597,
+      "step": 111890
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.712416855090827e-06,
+      "loss": 0.2631,
+      "step": 111895
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.7106045517726525e-06,
+      "loss": 0.2607,
+      "step": 111900
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.708792498914564e-06,
+      "loss": 0.2574,
+      "step": 111905
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.706980696540084e-06,
+      "loss": 0.2504,
+      "step": 111910
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.705169144672753e-06,
+      "loss": 0.2306,
+      "step": 111915
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.703357843336085e-06,
+      "loss": 0.2684,
+      "step": 111920
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.701546792553597e-06,
+      "loss": 0.2615,
+      "step": 111925
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.6997359923488005e-06,
+      "loss": 0.2572,
+      "step": 111930
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.6979254427452195e-06,
+      "loss": 0.2564,
+      "step": 111935
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.696115143766356e-06,
+      "loss": 0.2539,
+      "step": 111940
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.694305095435709e-06,
+      "loss": 0.2691,
+      "step": 111945
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.692495297776787e-06,
+      "loss": 0.2637,
+      "step": 111950
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.690685750813096e-06,
+      "loss": 0.255,
+      "step": 111955
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.688876454568126e-06,
+      "loss": 0.2362,
+      "step": 111960
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.6870674090653604e-06,
+      "loss": 0.2559,
+      "step": 111965
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.685258614328304e-06,
+      "loss": 0.2687,
+      "step": 111970
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.683450070380436e-06,
+      "loss": 0.2709,
+      "step": 111975
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.681641777245239e-06,
+      "loss": 0.2485,
+      "step": 111980
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.679833734946186e-06,
+      "loss": 0.2666,
+      "step": 111985
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.678025943506765e-06,
+      "loss": 0.2412,
+      "step": 111990
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.676218402950445e-06,
+      "loss": 0.2448,
+      "step": 111995
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.674411113300688e-06,
+      "loss": 0.2586,
+      "step": 112000
+    },
+    {
+      "epoch": 3.94,
+      "eval_loss": 0.2502914071083069,
+      "eval_runtime": 10.543,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 112000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.67260407458097e-06,
+      "loss": 0.26,
+      "step": 112005
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.670797286814747e-06,
+      "loss": 0.2595,
+      "step": 112010
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.668990750025488e-06,
+      "loss": 0.2258,
+      "step": 112015
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.667184464236644e-06,
+      "loss": 0.2686,
+      "step": 112020
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.665378429471663e-06,
+      "loss": 0.274,
+      "step": 112025
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.663572645754009e-06,
+      "loss": 0.2668,
+      "step": 112030
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.661767113107119e-06,
+      "loss": 0.2795,
+      "step": 112035
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.659961831554442e-06,
+      "loss": 0.2584,
+      "step": 112040
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.658156801119404e-06,
+      "loss": 0.25,
+      "step": 112045
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.656352021825464e-06,
+      "loss": 0.2491,
+      "step": 112050
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.654547493696036e-06,
+      "loss": 0.2243,
+      "step": 112055
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.652743216754566e-06,
+      "loss": 0.2689,
+      "step": 112060
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.650939191024468e-06,
+      "loss": 0.2536,
+      "step": 112065
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.649135416529183e-06,
+      "loss": 0.2666,
+      "step": 112070
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.6473318932921185e-06,
+      "loss": 0.2499,
+      "step": 112075
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.6455286213366905e-06,
+      "loss": 0.2671,
+      "step": 112080
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.643725600686326e-06,
+      "loss": 0.2577,
+      "step": 112085
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.641922831364424e-06,
+      "loss": 0.2648,
+      "step": 112090
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.6401203133944005e-06,
+      "loss": 0.2578,
+      "step": 112095
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.6383180467996486e-06,
+      "loss": 0.2655,
+      "step": 112100
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.636516031603583e-06,
+      "loss": 0.2529,
+      "step": 112105
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.6347142678295966e-06,
+      "loss": 0.2483,
+      "step": 112110
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.632912755501077e-06,
+      "loss": 0.2485,
+      "step": 112115
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.63111149464142e-06,
+      "loss": 0.2457,
+      "step": 112120
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.629310485274022e-06,
+      "loss": 0.2464,
+      "step": 112125
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.627509727422265e-06,
+      "loss": 0.2581,
+      "step": 112130
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.6257092211095236e-06,
+      "loss": 0.2541,
+      "step": 112135
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.623908966359176e-06,
+      "loss": 0.2483,
+      "step": 112140
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.622108963194605e-06,
+      "loss": 0.259,
+      "step": 112145
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.620309211639179e-06,
+      "loss": 0.2469,
+      "step": 112150
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.618509711716269e-06,
+      "loss": 0.2391,
+      "step": 112155
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.616710463449229e-06,
+      "loss": 0.2729,
+      "step": 112160
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.614911466861436e-06,
+      "loss": 0.2461,
+      "step": 112165
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.6131127219762355e-06,
+      "loss": 0.2695,
+      "step": 112170
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.6113142288169916e-06,
+      "loss": 0.2568,
+      "step": 112175
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.60951598740706e-06,
+      "loss": 0.2366,
+      "step": 112180
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.607717997769785e-06,
+      "loss": 0.2724,
+      "step": 112185
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.605920259928512e-06,
+      "loss": 0.261,
+      "step": 112190
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.604122773906576e-06,
+      "loss": 0.2338,
+      "step": 112195
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.6023255397273295e-06,
+      "loss": 0.2412,
+      "step": 112200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.600528557414103e-06,
+      "loss": 0.2643,
+      "step": 112205
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.598731826990231e-06,
+      "loss": 0.2565,
+      "step": 112210
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.596935348479029e-06,
+      "loss": 0.2487,
+      "step": 112215
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.595139121903844e-06,
+      "loss": 0.2487,
+      "step": 112220
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.593343147287983e-06,
+      "loss": 0.2763,
+      "step": 112225
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.591547424654775e-06,
+      "loss": 0.2291,
+      "step": 112230
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5897519540275305e-06,
+      "loss": 0.2449,
+      "step": 112235
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.587956735429569e-06,
+      "loss": 0.27,
+      "step": 112240
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.586161768884196e-06,
+      "loss": 0.2645,
+      "step": 112245
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.584367054414721e-06,
+      "loss": 0.2603,
+      "step": 112250
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.582572592044436e-06,
+      "loss": 0.2299,
+      "step": 112255
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.580778381796656e-06,
+      "loss": 0.2727,
+      "step": 112260
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.578984423694672e-06,
+      "loss": 0.2459,
+      "step": 112265
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.577190717761774e-06,
+      "loss": 0.244,
+      "step": 112270
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.575397264021248e-06,
+      "loss": 0.2651,
+      "step": 112275
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.57360406249639e-06,
+      "loss": 0.281,
+      "step": 112280
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.571811113210484e-06,
+      "loss": 0.2392,
+      "step": 112285
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.570018416186804e-06,
+      "loss": 0.2478,
+      "step": 112290
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.568225971448634e-06,
+      "loss": 0.2503,
+      "step": 112295
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.566433779019245e-06,
+      "loss": 0.2471,
+      "step": 112300
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.564641838921905e-06,
+      "loss": 0.2399,
+      "step": 112305
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.562850151179877e-06,
+      "loss": 0.2553,
+      "step": 112310
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.561058715816436e-06,
+      "loss": 0.2661,
+      "step": 112315
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5592675328548365e-06,
+      "loss": 0.2591,
+      "step": 112320
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.557476602318337e-06,
+      "loss": 0.2437,
+      "step": 112325
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5556859242301865e-06,
+      "loss": 0.2619,
+      "step": 112330
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5538954986136346e-06,
+      "loss": 0.2565,
+      "step": 112335
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5521053254919435e-06,
+      "loss": 0.252,
+      "step": 112340
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.550315404888348e-06,
+      "loss": 0.2521,
+      "step": 112345
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5485257368260855e-06,
+      "loss": 0.2571,
+      "step": 112350
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5467363213283995e-06,
+      "loss": 0.2457,
+      "step": 112355
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5449471584185234e-06,
+      "loss": 0.2381,
+      "step": 112360
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.54315824811969e-06,
+      "loss": 0.2456,
+      "step": 112365
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.541369590455114e-06,
+      "loss": 0.2348,
+      "step": 112370
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.5395811854480365e-06,
+      "loss": 0.2713,
+      "step": 112375
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.537793033121674e-06,
+      "loss": 0.2595,
+      "step": 112380
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.536005133499236e-06,
+      "loss": 0.2677,
+      "step": 112385
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.534217486603946e-06,
+      "loss": 0.2486,
+      "step": 112390
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.532430092459017e-06,
+      "loss": 0.2567,
+      "step": 112395
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.530642951087656e-06,
+      "loss": 0.2534,
+      "step": 112400
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.528856062513058e-06,
+      "loss": 0.2319,
+      "step": 112405
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.527069426758438e-06,
+      "loss": 0.2461,
+      "step": 112410
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.525283043846988e-06,
+      "loss": 0.2407,
+      "step": 112415
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.523496913801904e-06,
+      "loss": 0.2517,
+      "step": 112420
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.521711036646371e-06,
+      "loss": 0.2763,
+      "step": 112425
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.51992541240359e-06,
+      "loss": 0.243,
+      "step": 112430
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.518140041096739e-06,
+      "loss": 0.2605,
+      "step": 112435
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.516354922748996e-06,
+      "loss": 0.2414,
+      "step": 112440
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.514570057383547e-06,
+      "loss": 0.2687,
+      "step": 112445
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.512785445023561e-06,
+      "loss": 0.2659,
+      "step": 112450
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.511001085692219e-06,
+      "loss": 0.2549,
+      "step": 112455
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.509216979412685e-06,
+      "loss": 0.2681,
+      "step": 112460
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.507433126208114e-06,
+      "loss": 0.2681,
+      "step": 112465
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.5056495261016886e-06,
+      "loss": 0.2617,
+      "step": 112470
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.5038661791165545e-06,
+      "loss": 0.2559,
+      "step": 112475
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.502083085275869e-06,
+      "loss": 0.2439,
+      "step": 112480
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.500300244602779e-06,
+      "loss": 0.2442,
+      "step": 112485
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.498517657120447e-06,
+      "loss": 0.2508,
+      "step": 112490
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.496735322852004e-06,
+      "loss": 0.2434,
+      "step": 112495
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.494953241820608e-06,
+      "loss": 0.2763,
+      "step": 112500
+    },
+    {
+      "epoch": 3.96,
+      "eval_loss": 0.25041884183883667,
+      "eval_runtime": 10.5467,
+      "eval_samples_per_second": 9.482,
+      "eval_steps_per_second": 9.482,
+      "step": 112500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.49317141404938e-06,
+      "loss": 0.2521,
+      "step": 112505
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.491389839561476e-06,
+      "loss": 0.2585,
+      "step": 112510
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.489608518380016e-06,
+      "loss": 0.2675,
+      "step": 112515
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.487827450528124e-06,
+      "loss": 0.2587,
+      "step": 112520
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.486046636028941e-06,
+      "loss": 0.2654,
+      "step": 112525
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.484266074905584e-06,
+      "loss": 0.2562,
+      "step": 112530
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4824857671811695e-06,
+      "loss": 0.2806,
+      "step": 112535
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4807057128788054e-06,
+      "loss": 0.2583,
+      "step": 112540
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.478925912021623e-06,
+      "loss": 0.253,
+      "step": 112545
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.477146364632724e-06,
+      "loss": 0.2511,
+      "step": 112550
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.475367070735207e-06,
+      "loss": 0.25,
+      "step": 112555
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.473588030352178e-06,
+      "loss": 0.2861,
+      "step": 112560
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.471809243506751e-06,
+      "loss": 0.2481,
+      "step": 112565
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.470030710222007e-06,
+      "loss": 0.2609,
+      "step": 112570
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.468252430521048e-06,
+      "loss": 0.2488,
+      "step": 112575
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4664744044269495e-06,
+      "loss": 0.279,
+      "step": 112580
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.464696631962815e-06,
+      "loss": 0.2666,
+      "step": 112585
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.462919113151721e-06,
+      "loss": 0.2647,
+      "step": 112590
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.461141848016746e-06,
+      "loss": 0.2778,
+      "step": 112595
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.45936483658096e-06,
+      "loss": 0.2522,
+      "step": 112600
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.4575880788674534e-06,
+      "loss": 0.2357,
+      "step": 112605
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.455811574899278e-06,
+      "loss": 0.2347,
+      "step": 112610
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.454035324699517e-06,
+      "loss": 0.2538,
+      "step": 112615
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.452259328291218e-06,
+      "loss": 0.2651,
+      "step": 112620
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.450483585697458e-06,
+      "loss": 0.2506,
+      "step": 112625
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.448708096941282e-06,
+      "loss": 0.2621,
+      "step": 112630
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.446932862045742e-06,
+      "loss": 0.2529,
+      "step": 112635
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.445157881033899e-06,
+      "loss": 0.2539,
+      "step": 112640
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.443383153928794e-06,
+      "loss": 0.2768,
+      "step": 112645
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.441608680753471e-06,
+      "loss": 0.2464,
+      "step": 112650
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.439834461530963e-06,
+      "loss": 0.2607,
+      "step": 112655
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.43806049628432e-06,
+      "loss": 0.2392,
+      "step": 112660
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.436286785036562e-06,
+      "loss": 0.2353,
+      "step": 112665
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.434513327810733e-06,
+      "loss": 0.2523,
+      "step": 112670
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.432740124629851e-06,
+      "loss": 0.2582,
+      "step": 112675
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.430967175516949e-06,
+      "loss": 0.2597,
+      "step": 112680
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.42919448049504e-06,
+      "loss": 0.2584,
+      "step": 112685
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.427422039587144e-06,
+      "loss": 0.2459,
+      "step": 112690
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.425649852816267e-06,
+      "loss": 0.2611,
+      "step": 112695
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.423877920205434e-06,
+      "loss": 0.2471,
+      "step": 112700
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.422106241777644e-06,
+      "loss": 0.2433,
+      "step": 112705
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.420334817555903e-06,
+      "loss": 0.2592,
+      "step": 112710
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.418563647563205e-06,
+      "loss": 0.2581,
+      "step": 112715
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.416792731822551e-06,
+      "loss": 0.269,
+      "step": 112720
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.415022070356948e-06,
+      "loss": 0.2512,
+      "step": 112725
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.413251663189373e-06,
+      "loss": 0.249,
+      "step": 112730
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.411481510342811e-06,
+      "loss": 0.257,
+      "step": 112735
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.40971161184026e-06,
+      "loss": 0.2505,
+      "step": 112740
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.407941967704694e-06,
+      "loss": 0.2469,
+      "step": 112745
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.406172577959082e-06,
+      "loss": 0.2598,
+      "step": 112750
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.404403442626413e-06,
+      "loss": 0.2777,
+      "step": 112755
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.4026345617296495e-06,
+      "loss": 0.2368,
+      "step": 112760
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.400865935291763e-06,
+      "loss": 0.2426,
+      "step": 112765
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.399097563335709e-06,
+      "loss": 0.251,
+      "step": 112770
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.397329445884456e-06,
+      "loss": 0.2764,
+      "step": 112775
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.395561582960968e-06,
+      "loss": 0.2656,
+      "step": 112780
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.3937939745881935e-06,
+      "loss": 0.2502,
+      "step": 112785
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.392026620789075e-06,
+      "loss": 0.2647,
+      "step": 112790
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.390259521586577e-06,
+      "loss": 0.2678,
+      "step": 112795
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.388492677003634e-06,
+      "loss": 0.2412,
+      "step": 112800
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.38672608706319e-06,
+      "loss": 0.2462,
+      "step": 112805
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.384959751788176e-06,
+      "loss": 0.2595,
+      "step": 112810
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.383193671201539e-06,
+      "loss": 0.2423,
+      "step": 112815
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.381427845326206e-06,
+      "loss": 0.2729,
+      "step": 112820
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.379662274185093e-06,
+      "loss": 0.2904,
+      "step": 112825
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.377896957801137e-06,
+      "loss": 0.2528,
+      "step": 112830
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.376131896197267e-06,
+      "loss": 0.2373,
+      "step": 112835
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.374367089396387e-06,
+      "loss": 0.2667,
+      "step": 112840
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.37260253742142e-06,
+      "loss": 0.271,
+      "step": 112845
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.370838240295267e-06,
+      "loss": 0.2607,
+      "step": 112850
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.36907419804085e-06,
+      "loss": 0.2656,
+      "step": 112855
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.367310410681064e-06,
+      "loss": 0.2527,
+      "step": 112860
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.365546878238811e-06,
+      "loss": 0.2566,
+      "step": 112865
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.363783600736997e-06,
+      "loss": 0.239,
+      "step": 112870
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.362020578198512e-06,
+      "loss": 0.2389,
+      "step": 112875
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.360257810646244e-06,
+      "loss": 0.2692,
+      "step": 112880
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.358495298103089e-06,
+      "loss": 0.2588,
+      "step": 112885
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.356733040591921e-06,
+      "loss": 0.2516,
+      "step": 112890
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.354971038135639e-06,
+      "loss": 0.2483,
+      "step": 112895
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.353209290757108e-06,
+      "loss": 0.2648,
+      "step": 112900
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.351447798479203e-06,
+      "loss": 0.2546,
+      "step": 112905
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.349686561324805e-06,
+      "loss": 0.2499,
+      "step": 112910
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.347925579316776e-06,
+      "loss": 0.2448,
+      "step": 112915
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.346164852477984e-06,
+      "loss": 0.2545,
+      "step": 112920
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.34440438083128e-06,
+      "loss": 0.2552,
+      "step": 112925
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.34264416439954e-06,
+      "loss": 0.2438,
+      "step": 112930
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.340884203205604e-06,
+      "loss": 0.2661,
+      "step": 112935
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.339124497272338e-06,
+      "loss": 0.2654,
+      "step": 112940
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.337365046622575e-06,
+      "loss": 0.2698,
+      "step": 112945
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.3356058512791765e-06,
+      "loss": 0.235,
+      "step": 112950
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.333846911264975e-06,
+      "loss": 0.248,
+      "step": 112955
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.3320882266028135e-06,
+      "loss": 0.2477,
+      "step": 112960
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.330329797315517e-06,
+      "loss": 0.2536,
+      "step": 112965
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.3285716234259316e-06,
+      "loss": 0.2584,
+      "step": 112970
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.326813704956879e-06,
+      "loss": 0.2535,
+      "step": 112975
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.325056041931187e-06,
+      "loss": 0.2336,
+      "step": 112980
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.323298634371668e-06,
+      "loss": 0.2672,
+      "step": 112985
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.321541482301157e-06,
+      "loss": 0.2798,
+      "step": 112990
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.319784585742455e-06,
+      "loss": 0.2321,
+      "step": 112995
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.318027944718379e-06,
+      "loss": 0.2488,
+      "step": 113000
+    },
+    {
+      "epoch": 3.98,
+      "eval_loss": 0.25009918212890625,
+      "eval_runtime": 10.5513,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 113000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.316271559251745e-06,
+      "loss": 0.243,
+      "step": 113005
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.314515429365355e-06,
+      "loss": 0.2529,
+      "step": 113010
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.312759555082006e-06,
+      "loss": 0.2655,
+      "step": 113015
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.311003936424497e-06,
+      "loss": 0.2353,
+      "step": 113020
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.309248573415629e-06,
+      "loss": 0.2534,
+      "step": 113025
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.307493466078195e-06,
+      "loss": 0.2704,
+      "step": 113030
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.305738614434977e-06,
+      "loss": 0.262,
+      "step": 113035
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.303984018508759e-06,
+      "loss": 0.254,
+      "step": 113040
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.302229678322338e-06,
+      "loss": 0.2676,
+      "step": 113045
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.300475593898471e-06,
+      "loss": 0.2601,
+      "step": 113050
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.298721765259956e-06,
+      "loss": 0.27,
+      "step": 113055
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.296968192429549e-06,
+      "loss": 0.2447,
+      "step": 113060
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.295214875430032e-06,
+      "loss": 0.249,
+      "step": 113065
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.293461814284162e-06,
+      "loss": 0.25,
+      "step": 113070
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.291709009014706e-06,
+      "loss": 0.2598,
+      "step": 113075
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.28995645964441e-06,
+      "loss": 0.2613,
+      "step": 113080
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.288204166196048e-06,
+      "loss": 0.2532,
+      "step": 113085
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.286452128692365e-06,
+      "loss": 0.2574,
+      "step": 113090
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.284700347156105e-06,
+      "loss": 0.2614,
+      "step": 113095
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.28294882161002e-06,
+      "loss": 0.2657,
+      "step": 113100
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.281197552076847e-06,
+      "loss": 0.2509,
+      "step": 113105
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.279446538579336e-06,
+      "loss": 0.2641,
+      "step": 113110
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.2776957811402044e-06,
+      "loss": 0.2451,
+      "step": 113115
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.275945279782207e-06,
+      "loss": 0.259,
+      "step": 113120
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.27419503452806e-06,
+      "loss": 0.2733,
+      "step": 113125
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.272445045400489e-06,
+      "loss": 0.273,
+      "step": 113130
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.270695312422214e-06,
+      "loss": 0.2471,
+      "step": 113135
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.268945835615962e-06,
+      "loss": 0.247,
+      "step": 113140
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.267196615004446e-06,
+      "loss": 0.2455,
+      "step": 113145
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.265447650610381e-06,
+      "loss": 0.2811,
+      "step": 113150
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.263698942456463e-06,
+      "loss": 0.255,
+      "step": 113155
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.261950490565407e-06,
+      "loss": 0.2532,
+      "step": 113160
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.260202294959923e-06,
+      "loss": 0.2497,
+      "step": 113165
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.258454355662704e-06,
+      "loss": 0.2666,
+      "step": 113170
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.256706672696438e-06,
+      "loss": 0.2644,
+      "step": 113175
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.254959246083832e-06,
+      "loss": 0.2605,
+      "step": 113180
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.253212075847566e-06,
+      "loss": 0.2268,
+      "step": 113185
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.251465162010327e-06,
+      "loss": 0.271,
+      "step": 113190
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.249718504594794e-06,
+      "loss": 0.2515,
+      "step": 113195
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.247972103623655e-06,
+      "loss": 0.2726,
+      "step": 113200
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.246225959119583e-06,
+      "loss": 0.2534,
+      "step": 113205
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.244480071105243e-06,
+      "loss": 0.258,
+      "step": 113210
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.242734439603306e-06,
+      "loss": 0.2804,
+      "step": 113215
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.240989064636451e-06,
+      "loss": 0.246,
+      "step": 113220
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.2392439462273315e-06,
+      "loss": 0.2575,
+      "step": 113225
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.2374990843985975e-06,
+      "loss": 0.2712,
+      "step": 113230
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.235754479172925e-06,
+      "loss": 0.2583,
+      "step": 113235
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.23401013057295e-06,
+      "loss": 0.2454,
+      "step": 113240
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.2322660386213306e-06,
+      "loss": 0.2527,
+      "step": 113245
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.2305222033407015e-06,
+      "loss": 0.2681,
+      "step": 113250
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.2287786247537215e-06,
+      "loss": 0.2444,
+      "step": 113255
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.227035302883018e-06,
+      "loss": 0.2593,
+      "step": 113260
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.225292237751225e-06,
+      "loss": 0.2631,
+      "step": 113265
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.223549429380983e-06,
+      "loss": 0.2598,
+      "step": 113270
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.221806877794922e-06,
+      "loss": 0.2705,
+      "step": 113275
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.220064583015666e-06,
+      "loss": 0.2551,
+      "step": 113280
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.218322545065834e-06,
+      "loss": 0.2478,
+      "step": 113285
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.216580763968043e-06,
+      "loss": 0.2449,
+      "step": 113290
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.214839239744917e-06,
+      "loss": 0.2426,
+      "step": 113295
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.2130979724190644e-06,
+      "loss": 0.2532,
+      "step": 113300
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.211356962013098e-06,
+      "loss": 0.2717,
+      "step": 113305
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.209616208549611e-06,
+      "loss": 0.259,
+      "step": 113310
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.2078757120512215e-06,
+      "loss": 0.2383,
+      "step": 113315
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.206135472540519e-06,
+      "loss": 0.2458,
+      "step": 113320
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.204395490040106e-06,
+      "loss": 0.2585,
+      "step": 113325
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.202655764572567e-06,
+      "loss": 0.2655,
+      "step": 113330
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.200916296160502e-06,
+      "loss": 0.2825,
+      "step": 113335
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.1991770848264894e-06,
+      "loss": 0.2488,
+      "step": 113340
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.197438130593107e-06,
+      "loss": 0.2558,
+      "step": 113345
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.195699433482948e-06,
+      "loss": 0.2687,
+      "step": 113350
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.193960993518579e-06,
+      "loss": 0.2517,
+      "step": 113355
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.1922228107225746e-06,
+      "loss": 0.2554,
+      "step": 113360
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.190484885117497e-06,
+      "loss": 0.2562,
+      "step": 113365
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.188747216725925e-06,
+      "loss": 0.2593,
+      "step": 113370
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.187009805570406e-06,
+      "loss": 0.2627,
+      "step": 113375
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.1852726516735175e-06,
+      "loss": 0.2581,
+      "step": 113380
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.1835357550577945e-06,
+      "loss": 0.2622,
+      "step": 113385
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.181799115745809e-06,
+      "loss": 0.2618,
+      "step": 113390
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.180062733760102e-06,
+      "loss": 0.2673,
+      "step": 113395
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.178326609123216e-06,
+      "loss": 0.2467,
+      "step": 113400
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.176590741857689e-06,
+      "loss": 0.28,
+      "step": 113405
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.174855131986079e-06,
+      "loss": 0.2534,
+      "step": 113410
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.173119779530905e-06,
+      "loss": 0.2566,
+      "step": 113415
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.171384684514705e-06,
+      "loss": 0.2594,
+      "step": 113420
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.169649846960001e-06,
+      "loss": 0.2377,
+      "step": 113425
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.1679152668893255e-06,
+      "loss": 0.2462,
+      "step": 113430
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.166180944325208e-06,
+      "loss": 0.2307,
+      "step": 113435
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.164446879290158e-06,
+      "loss": 0.2457,
+      "step": 113440
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.162713071806688e-06,
+      "loss": 0.2578,
+      "step": 113445
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.160979521897322e-06,
+      "loss": 0.2747,
+      "step": 113450
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.159246229584561e-06,
+      "loss": 0.2453,
+      "step": 113455
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.157513194890906e-06,
+      "loss": 0.2429,
+      "step": 113460
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.155780417838874e-06,
+      "loss": 0.2773,
+      "step": 113465
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.154047898450954e-06,
+      "loss": 0.2238,
+      "step": 113470
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.152315636749641e-06,
+      "loss": 0.2749,
+      "step": 113475
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.150583632757427e-06,
+      "loss": 0.2671,
+      "step": 113480
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.148851886496809e-06,
+      "loss": 0.2504,
+      "step": 113485
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.147120397990257e-06,
+      "loss": 0.2435,
+      "step": 113490
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.145389167260273e-06,
+      "loss": 0.2397,
+      "step": 113495
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.143658194329318e-06,
+      "loss": 0.2781,
+      "step": 113500
+    },
+    {
+      "epoch": 3.99,
+      "eval_loss": 0.2501179575920105,
+      "eval_runtime": 10.5566,
+      "eval_samples_per_second": 9.473,
+      "eval_steps_per_second": 9.473,
+      "step": 113500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.14192747921988e-06,
+      "loss": 0.2406,
+      "step": 113505
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.1401970219544296e-06,
+      "loss": 0.2449,
+      "step": 113510
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.138466822555432e-06,
+      "loss": 0.2581,
+      "step": 113515
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.136736881045348e-06,
+      "loss": 0.2583,
+      "step": 113520
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.13500719744665e-06,
+      "loss": 0.2618,
+      "step": 113525
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.133277771781794e-06,
+      "loss": 0.2484,
+      "step": 113530
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.131548604073225e-06,
+      "loss": 0.2693,
+      "step": 113535
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.12981969434341e-06,
+      "loss": 0.247,
+      "step": 113540
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.128091042614785e-06,
+      "loss": 0.2388,
+      "step": 113545
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.1263626489098106e-06,
+      "loss": 0.2597,
+      "step": 113550
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.124634513250917e-06,
+      "loss": 0.2461,
+      "step": 113555
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.12290663566054e-06,
+      "loss": 0.2549,
+      "step": 113560
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.121179016161129e-06,
+      "loss": 0.2586,
+      "step": 113565
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.119451654775106e-06,
+      "loss": 0.2343,
+      "step": 113570
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.117724551524897e-06,
+      "loss": 0.2777,
+      "step": 113575
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.115997706432937e-06,
+      "loss": 0.2766,
+      "step": 113580
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.114271119521644e-06,
+      "loss": 0.2655,
+      "step": 113585
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.1125447908134335e-06,
+      "loss": 0.2774,
+      "step": 113590
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.110818720330715e-06,
+      "loss": 0.2562,
+      "step": 113595
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.109092908095911e-06,
+      "loss": 0.2677,
+      "step": 113600
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.107367354131432e-06,
+      "loss": 0.2523,
+      "step": 113605
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.105642058459678e-06,
+      "loss": 0.2611,
+      "step": 113610
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.103917021103044e-06,
+      "loss": 0.2413,
+      "step": 113615
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.102192242083945e-06,
+      "loss": 0.2559,
+      "step": 113620
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.100467721424765e-06,
+      "loss": 0.2795,
+      "step": 113625
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.098743459147895e-06,
+      "loss": 0.2676,
+      "step": 113630
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.09701945527572e-06,
+      "loss": 0.2559,
+      "step": 113635
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.095295709830638e-06,
+      "loss": 0.2421,
+      "step": 113640
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.093572222835025e-06,
+      "loss": 0.2599,
+      "step": 113645
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.091848994311252e-06,
+      "loss": 0.253,
+      "step": 113650
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.090126024281697e-06,
+      "loss": 0.2594,
+      "step": 113655
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.088403312768741e-06,
+      "loss": 0.275,
+      "step": 113660
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.086680859794748e-06,
+      "loss": 0.2643,
+      "step": 113665
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.084958665382078e-06,
+      "loss": 0.2585,
+      "step": 113670
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.08323672955309e-06,
+      "loss": 0.2616,
+      "step": 113675
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.081515052330154e-06,
+      "loss": 0.2442,
+      "step": 113680
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.079793633735619e-06,
+      "loss": 0.2518,
+      "step": 113685
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.0780724737918285e-06,
+      "loss": 0.2479,
+      "step": 113690
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.076351572521146e-06,
+      "loss": 0.2412,
+      "step": 113695
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.074630929945906e-06,
+      "loss": 0.2516,
+      "step": 113700
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.0729105460884464e-06,
+      "loss": 0.2561,
+      "step": 113705
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.071190420971112e-06,
+      "loss": 0.2434,
+      "step": 113710
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.06947055461624e-06,
+      "loss": 0.2354,
+      "step": 113715
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.06775094704616e-06,
+      "loss": 0.2413,
+      "step": 113720
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.066031598283197e-06,
+      "loss": 0.2387,
+      "step": 113725
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.064312508349672e-06,
+      "loss": 0.2448,
+      "step": 113730
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.062593677267916e-06,
+      "loss": 0.2661,
+      "step": 113735
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.060875105060242e-06,
+      "loss": 0.2472,
+      "step": 113740
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.059156791748965e-06,
+      "loss": 0.2515,
+      "step": 113745
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.05743873735639e-06,
+      "loss": 0.258,
+      "step": 113750
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.055720941904835e-06,
+      "loss": 0.2551,
+      "step": 113755
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.0540034054165934e-06,
+      "loss": 0.2431,
+      "step": 113760
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.05228612791398e-06,
+      "loss": 0.2497,
+      "step": 113765
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.050569109419279e-06,
+      "loss": 0.2363,
+      "step": 113770
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.048852349954797e-06,
+      "loss": 0.2531,
+      "step": 113775
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.04713584954282e-06,
+      "loss": 0.2519,
+      "step": 113780
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.045419608205632e-06,
+      "loss": 0.2519,
+      "step": 113785
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.0437036259655155e-06,
+      "loss": 0.2466,
+      "step": 113790
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.041987902844761e-06,
+      "loss": 0.2551,
+      "step": 113795
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.040272438865642e-06,
+      "loss": 0.2258,
+      "step": 113800
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.038557234050423e-06,
+      "loss": 0.2406,
+      "step": 113805
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.036842288421392e-06,
+      "loss": 0.2453,
+      "step": 113810
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.0351276020008e-06,
+      "loss": 0.2551,
+      "step": 113815
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.0334131748109245e-06,
+      "loss": 0.254,
+      "step": 113820
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.0316990068740165e-06,
+      "loss": 0.2592,
+      "step": 113825
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.029985098212342e-06,
+      "loss": 0.2589,
+      "step": 113830
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.028271448848149e-06,
+      "loss": 0.257,
+      "step": 113835
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.026558058803691e-06,
+      "loss": 0.2377,
+      "step": 113840
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.024844928101208e-06,
+      "loss": 0.2496,
+      "step": 113845
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.023132056762953e-06,
+      "loss": 0.2402,
+      "step": 113850
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.021419444811165e-06,
+      "loss": 0.2561,
+      "step": 113855
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.01970709226808e-06,
+      "loss": 0.259,
+      "step": 113860
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.017994999155923e-06,
+      "loss": 0.2348,
+      "step": 113865
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.016283165496933e-06,
+      "loss": 0.2411,
+      "step": 113870
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.014571591313341e-06,
+      "loss": 0.2645,
+      "step": 113875
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.012860276627368e-06,
+      "loss": 0.2558,
+      "step": 113880
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.0111492214612254e-06,
+      "loss": 0.2518,
+      "step": 113885
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.009438425837143e-06,
+      "loss": 0.2665,
+      "step": 113890
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.007727889777327e-06,
+      "loss": 0.2439,
+      "step": 113895
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.006017613303993e-06,
+      "loss": 0.2468,
+      "step": 113900
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.004307596439334e-06,
+      "loss": 0.2549,
+      "step": 113905
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.002597839205572e-06,
+      "loss": 0.2707,
+      "step": 113910
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 5.000888341624896e-06,
+      "loss": 0.2407,
+      "step": 113915
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.999179103719501e-06,
+      "loss": 0.2382,
+      "step": 113920
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.997470125511588e-06,
+      "loss": 0.2536,
+      "step": 113925
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.995761407023339e-06,
+      "loss": 0.2392,
+      "step": 113930
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.994052948276948e-06,
+      "loss": 0.2406,
+      "step": 113935
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.992344749294592e-06,
+      "loss": 0.2357,
+      "step": 113940
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.990636810098459e-06,
+      "loss": 0.2519,
+      "step": 113945
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.98892913071072e-06,
+      "loss": 0.2271,
+      "step": 113950
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.98722171115355e-06,
+      "loss": 0.2694,
+      "step": 113955
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.985514551449108e-06,
+      "loss": 0.2582,
+      "step": 113960
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9838076516195796e-06,
+      "loss": 0.2609,
+      "step": 113965
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.982101011687115e-06,
+      "loss": 0.2533,
+      "step": 113970
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.980394631673871e-06,
+      "loss": 0.2564,
+      "step": 113975
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.978688511602014e-06,
+      "loss": 0.2653,
+      "step": 113980
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.976982651493689e-06,
+      "loss": 0.2505,
+      "step": 113985
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9752770513710524e-06,
+      "loss": 0.2538,
+      "step": 113990
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9735717112562475e-06,
+      "loss": 0.2483,
+      "step": 113995
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.971866631171409e-06,
+      "loss": 0.2482,
+      "step": 114000
+    },
+    {
+      "epoch": 4.01,
+      "eval_loss": 0.25011008977890015,
+      "eval_runtime": 10.5643,
+      "eval_samples_per_second": 9.466,
+      "eval_steps_per_second": 9.466,
+      "step": 114000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9701618111386906e-06,
+      "loss": 0.2451,
+      "step": 114005
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.968457251180222e-06,
+      "loss": 0.2401,
+      "step": 114010
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.966752951318132e-06,
+      "loss": 0.2673,
+      "step": 114015
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.965048911574547e-06,
+      "loss": 0.2462,
+      "step": 114020
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.963345131971606e-06,
+      "loss": 0.2481,
+      "step": 114025
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.961641612531423e-06,
+      "loss": 0.2547,
+      "step": 114030
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.959938353276111e-06,
+      "loss": 0.257,
+      "step": 114035
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.958235354227791e-06,
+      "loss": 0.2402,
+      "step": 114040
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.956532615408588e-06,
+      "loss": 0.2731,
+      "step": 114045
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.954830136840597e-06,
+      "loss": 0.2485,
+      "step": 114050
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9531279185459186e-06,
+      "loss": 0.2538,
+      "step": 114055
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9514259605466715e-06,
+      "loss": 0.2468,
+      "step": 114060
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.949724262864944e-06,
+      "loss": 0.2409,
+      "step": 114065
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.948022825522835e-06,
+      "loss": 0.2396,
+      "step": 114070
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.946321648542429e-06,
+      "loss": 0.2473,
+      "step": 114075
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.944620731945826e-06,
+      "loss": 0.2257,
+      "step": 114080
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.942920075755103e-06,
+      "loss": 0.2723,
+      "step": 114085
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9412196799923424e-06,
+      "loss": 0.2209,
+      "step": 114090
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.939519544679622e-06,
+      "loss": 0.2604,
+      "step": 114095
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.937819669839028e-06,
+      "loss": 0.2597,
+      "step": 114100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.936120055492624e-06,
+      "loss": 0.2312,
+      "step": 114105
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.934420701662479e-06,
+      "loss": 0.2435,
+      "step": 114110
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.93272160837065e-06,
+      "loss": 0.2504,
+      "step": 114115
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.931022775639213e-06,
+      "loss": 0.2324,
+      "step": 114120
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.929324203490221e-06,
+      "loss": 0.2535,
+      "step": 114125
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.927625891945725e-06,
+      "loss": 0.2404,
+      "step": 114130
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.925927841027772e-06,
+      "loss": 0.2518,
+      "step": 114135
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.924230050758425e-06,
+      "loss": 0.2626,
+      "step": 114140
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.922532521159712e-06,
+      "loss": 0.267,
+      "step": 114145
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.920835252253683e-06,
+      "loss": 0.2593,
+      "step": 114150
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.9191382440623826e-06,
+      "loss": 0.2333,
+      "step": 114155
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.917441496607836e-06,
+      "loss": 0.2776,
+      "step": 114160
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.915745009912079e-06,
+      "loss": 0.2543,
+      "step": 114165
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.91404878399713e-06,
+      "loss": 0.255,
+      "step": 114170
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.912352818885027e-06,
+      "loss": 0.2524,
+      "step": 114175
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.910657114597786e-06,
+      "loss": 0.2635,
+      "step": 114180
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.908961671157422e-06,
+      "loss": 0.2681,
+      "step": 114185
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.9072664885859435e-06,
+      "loss": 0.2215,
+      "step": 114190
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.905571566905373e-06,
+      "loss": 0.229,
+      "step": 114195
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.9038769061377085e-06,
+      "loss": 0.2318,
+      "step": 114200
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.902182506304964e-06,
+      "loss": 0.2536,
+      "step": 114205
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.900488367429129e-06,
+      "loss": 0.2424,
+      "step": 114210
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.898794489532213e-06,
+      "loss": 0.248,
+      "step": 114215
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.897100872636201e-06,
+      "loss": 0.2413,
+      "step": 114220
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.895407516763087e-06,
+      "loss": 0.2437,
+      "step": 114225
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.893714421934853e-06,
+      "loss": 0.2403,
+      "step": 114230
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.89202158817349e-06,
+      "loss": 0.2628,
+      "step": 114235
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.890329015500977e-06,
+      "loss": 0.2512,
+      "step": 114240
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.888636703939289e-06,
+      "loss": 0.2466,
+      "step": 114245
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.886944653510392e-06,
+      "loss": 0.2495,
+      "step": 114250
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.885252864236264e-06,
+      "loss": 0.2409,
+      "step": 114255
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.883561336138878e-06,
+      "loss": 0.2546,
+      "step": 114260
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.881870069240191e-06,
+      "loss": 0.2439,
+      "step": 114265
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.880179063562154e-06,
+      "loss": 0.2482,
+      "step": 114270
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.878826447113366e-06,
+      "loss": 0.2398,
+      "step": 114275
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.8771359116878464e-06,
+      "loss": 0.2486,
+      "step": 114280
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.875445637544462e-06,
+      "loss": 0.2419,
+      "step": 114285
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.873755624705145e-06,
+      "loss": 0.2332,
+      "step": 114290
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.872065873191855e-06,
+      "loss": 0.2522,
+      "step": 114295
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.870376383026518e-06,
+      "loss": 0.2662,
+      "step": 114300
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.868687154231083e-06,
+      "loss": 0.2625,
+      "step": 114305
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.866998186827482e-06,
+      "loss": 0.2336,
+      "step": 114310
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.865309480837638e-06,
+      "loss": 0.2593,
+      "step": 114315
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.863621036283475e-06,
+      "loss": 0.2451,
+      "step": 114320
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.861932853186929e-06,
+      "loss": 0.2312,
+      "step": 114325
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.860244931569915e-06,
+      "loss": 0.2612,
+      "step": 114330
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.858557271454348e-06,
+      "loss": 0.2539,
+      "step": 114335
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.856869872862132e-06,
+      "loss": 0.2428,
+      "step": 114340
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.855182735815189e-06,
+      "loss": 0.2413,
+      "step": 114345
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.853495860335428e-06,
+      "loss": 0.2506,
+      "step": 114350
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.851809246444739e-06,
+      "loss": 0.2711,
+      "step": 114355
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.850122894165035e-06,
+      "loss": 0.2475,
+      "step": 114360
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.848436803518205e-06,
+      "loss": 0.2499,
+      "step": 114365
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.846750974526146e-06,
+      "loss": 0.2573,
+      "step": 114370
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.845065407210733e-06,
+      "loss": 0.23,
+      "step": 114375
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.843380101593872e-06,
+      "loss": 0.2409,
+      "step": 114380
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.8416950576974375e-06,
+      "loss": 0.2442,
+      "step": 114385
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.840010275543308e-06,
+      "loss": 0.2409,
+      "step": 114390
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.838325755153353e-06,
+      "loss": 0.2386,
+      "step": 114395
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.836641496549449e-06,
+      "loss": 0.2406,
+      "step": 114400
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.834957499753473e-06,
+      "loss": 0.2305,
+      "step": 114405
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.833273764787286e-06,
+      "loss": 0.2556,
+      "step": 114410
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.831590291672744e-06,
+      "loss": 0.2356,
+      "step": 114415
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.8299070804317145e-06,
+      "loss": 0.2411,
+      "step": 114420
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.828224131086051e-06,
+      "loss": 0.2539,
+      "step": 114425
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.826541443657601e-06,
+      "loss": 0.2573,
+      "step": 114430
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.824859018168209e-06,
+      "loss": 0.2431,
+      "step": 114435
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.823176854639735e-06,
+      "loss": 0.2348,
+      "step": 114440
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.821494953094013e-06,
+      "loss": 0.2431,
+      "step": 114445
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.819813313552873e-06,
+      "loss": 0.2549,
+      "step": 114450
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.818131936038159e-06,
+      "loss": 0.2597,
+      "step": 114455
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.816450820571708e-06,
+      "loss": 0.2243,
+      "step": 114460
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.814769967175339e-06,
+      "loss": 0.2375,
+      "step": 114465
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.813089375870878e-06,
+      "loss": 0.2516,
+      "step": 114470
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.811409046680149e-06,
+      "loss": 0.258,
+      "step": 114475
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.809728979624972e-06,
+      "loss": 0.2288,
+      "step": 114480
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.808049174727155e-06,
+      "loss": 0.2605,
+      "step": 114485
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.806369632008509e-06,
+      "loss": 0.2512,
+      "step": 114490
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.804690351490851e-06,
+      "loss": 0.2438,
+      "step": 114495
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.8030113331959805e-06,
+      "loss": 0.2472,
+      "step": 114500
+    },
+    {
+      "epoch": 4.03,
+      "eval_loss": 0.24984058737754822,
+      "eval_runtime": 10.5589,
+      "eval_samples_per_second": 9.471,
+      "eval_steps_per_second": 9.471,
+      "step": 114500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.80133257714569e-06,
+      "loss": 0.2481,
+      "step": 114505
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.79965408336179e-06,
+      "loss": 0.2397,
+      "step": 114510
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.797975851866063e-06,
+      "loss": 0.2519,
+      "step": 114515
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.796297882680315e-06,
+      "loss": 0.229,
+      "step": 114520
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.794620175826323e-06,
+      "loss": 0.2386,
+      "step": 114525
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.792942731325864e-06,
+      "loss": 0.2488,
+      "step": 114530
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.791265549200735e-06,
+      "loss": 0.2647,
+      "step": 114535
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.789588629472702e-06,
+      "loss": 0.2424,
+      "step": 114540
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.7879119721635444e-06,
+      "loss": 0.2584,
+      "step": 114545
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.786235577295023e-06,
+      "loss": 0.2679,
+      "step": 114550
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.784559444888917e-06,
+      "loss": 0.2353,
+      "step": 114555
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.782883574966979e-06,
+      "loss": 0.2396,
+      "step": 114560
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.781207967550977e-06,
+      "loss": 0.2575,
+      "step": 114565
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.779532622662661e-06,
+      "loss": 0.2447,
+      "step": 114570
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.777857540323794e-06,
+      "loss": 0.2481,
+      "step": 114575
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.776182720556121e-06,
+      "loss": 0.2639,
+      "step": 114580
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.774508163381381e-06,
+      "loss": 0.2617,
+      "step": 114585
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.772833868821328e-06,
+      "loss": 0.2352,
+      "step": 114590
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.771159836897699e-06,
+      "loss": 0.2407,
+      "step": 114595
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.769486067632226e-06,
+      "loss": 0.2519,
+      "step": 114600
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.767812561046639e-06,
+      "loss": 0.2444,
+      "step": 114605
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.76613931716268e-06,
+      "loss": 0.2578,
+      "step": 114610
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.764466336002066e-06,
+      "loss": 0.2422,
+      "step": 114615
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.762793617586514e-06,
+      "loss": 0.2292,
+      "step": 114620
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.761121161937751e-06,
+      "loss": 0.2377,
+      "step": 114625
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.759448969077496e-06,
+      "loss": 0.2634,
+      "step": 114630
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.757777039027458e-06,
+      "loss": 0.2383,
+      "step": 114635
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.756105371809347e-06,
+      "loss": 0.2616,
+      "step": 114640
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.75443396744486e-06,
+      "loss": 0.2624,
+      "step": 114645
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.752762825955711e-06,
+      "loss": 0.2428,
+      "step": 114650
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.751091947363592e-06,
+      "loss": 0.2472,
+      "step": 114655
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.749421331690202e-06,
+      "loss": 0.2706,
+      "step": 114660
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.747750978957221e-06,
+      "loss": 0.2544,
+      "step": 114665
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.746080889186355e-06,
+      "loss": 0.2538,
+      "step": 114670
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.744411062399276e-06,
+      "loss": 0.2644,
+      "step": 114675
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.742741498617676e-06,
+      "loss": 0.2504,
+      "step": 114680
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.74107219786322e-06,
+      "loss": 0.2516,
+      "step": 114685
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.739403160157596e-06,
+      "loss": 0.2469,
+      "step": 114690
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.737734385522471e-06,
+      "loss": 0.2598,
+      "step": 114695
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.736065873979506e-06,
+      "loss": 0.2376,
+      "step": 114700
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.734397625550377e-06,
+      "loss": 0.2479,
+      "step": 114705
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.732729640256739e-06,
+      "loss": 0.2631,
+      "step": 114710
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7310619181202495e-06,
+      "loss": 0.2706,
+      "step": 114715
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.729394459162556e-06,
+      "loss": 0.2574,
+      "step": 114720
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.727727263405326e-06,
+      "loss": 0.2489,
+      "step": 114725
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7260603308701885e-06,
+      "loss": 0.2596,
+      "step": 114730
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7243936615788045e-06,
+      "loss": 0.2392,
+      "step": 114735
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7227272555528e-06,
+      "loss": 0.2669,
+      "step": 114740
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.721061112813827e-06,
+      "loss": 0.2326,
+      "step": 114745
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.719395233383509e-06,
+      "loss": 0.2521,
+      "step": 114750
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.717729617283478e-06,
+      "loss": 0.2751,
+      "step": 114755
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.716064264535356e-06,
+      "loss": 0.2576,
+      "step": 114760
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.714399175160777e-06,
+      "loss": 0.2684,
+      "step": 114765
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.712734349181358e-06,
+      "loss": 0.2575,
+      "step": 114770
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.7110697866187135e-06,
+      "loss": 0.248,
+      "step": 114775
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.709405487494447e-06,
+      "loss": 0.2375,
+      "step": 114780
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.707741451830183e-06,
+      "loss": 0.2677,
+      "step": 114785
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.706077679647528e-06,
+      "loss": 0.2777,
+      "step": 114790
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.704414170968077e-06,
+      "loss": 0.2592,
+      "step": 114795
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.702750925813432e-06,
+      "loss": 0.252,
+      "step": 114800
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.70108794420519e-06,
+      "loss": 0.2279,
+      "step": 114805
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.69942522616495e-06,
+      "loss": 0.2586,
+      "step": 114810
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.697762771714284e-06,
+      "loss": 0.2462,
+      "step": 114815
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.696100580874796e-06,
+      "loss": 0.2519,
+      "step": 114820
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.694438653668062e-06,
+      "loss": 0.2567,
+      "step": 114825
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.692776990115661e-06,
+      "loss": 0.2444,
+      "step": 114830
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.691115590239159e-06,
+      "loss": 0.2809,
+      "step": 114835
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.689454454060138e-06,
+      "loss": 0.2277,
+      "step": 114840
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.6877935816001725e-06,
+      "loss": 0.2642,
+      "step": 114845
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.686132972880819e-06,
+      "loss": 0.2511,
+      "step": 114850
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.684472627923639e-06,
+      "loss": 0.236,
+      "step": 114855
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.682812546750198e-06,
+      "loss": 0.2635,
+      "step": 114860
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.681152729382046e-06,
+      "loss": 0.2551,
+      "step": 114865
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.6794931758407354e-06,
+      "loss": 0.2644,
+      "step": 114870
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.677833886147809e-06,
+      "loss": 0.2385,
+      "step": 114875
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.676174860324822e-06,
+      "loss": 0.2558,
+      "step": 114880
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.674516098393311e-06,
+      "loss": 0.2599,
+      "step": 114885
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.672857600374805e-06,
+      "loss": 0.2463,
+      "step": 114890
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.671199366290849e-06,
+      "loss": 0.2445,
+      "step": 114895
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.66954139616298e-06,
+      "loss": 0.2477,
+      "step": 114900
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.6678836900127145e-06,
+      "loss": 0.2558,
+      "step": 114905
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.666226247861583e-06,
+      "loss": 0.2577,
+      "step": 114910
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.6645690697310925e-06,
+      "loss": 0.2541,
+      "step": 114915
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.662912155642782e-06,
+      "loss": 0.2404,
+      "step": 114920
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.661255505618153e-06,
+      "loss": 0.2522,
+      "step": 114925
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.659599119678709e-06,
+      "loss": 0.2446,
+      "step": 114930
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.657942997845974e-06,
+      "loss": 0.2606,
+      "step": 114935
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.656287140141444e-06,
+      "loss": 0.248,
+      "step": 114940
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.65463154658661e-06,
+      "loss": 0.2571,
+      "step": 114945
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.652976217202987e-06,
+      "loss": 0.2576,
+      "step": 114950
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.651321152012048e-06,
+      "loss": 0.2447,
+      "step": 114955
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.649666351035304e-06,
+      "loss": 0.2749,
+      "step": 114960
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.648011814294229e-06,
+      "loss": 0.2494,
+      "step": 114965
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.646357541810301e-06,
+      "loss": 0.2668,
+      "step": 114970
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.644703533605013e-06,
+      "loss": 0.2353,
+      "step": 114975
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.643049789699835e-06,
+      "loss": 0.2493,
+      "step": 114980
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.641396310116241e-06,
+      "loss": 0.2338,
+      "step": 114985
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.639743094875693e-06,
+      "loss": 0.2765,
+      "step": 114990
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.638090143999668e-06,
+      "loss": 0.2424,
+      "step": 114995
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.636437457509618e-06,
+      "loss": 0.2276,
+      "step": 115000
+    },
+    {
+      "epoch": 4.05,
+      "eval_loss": 0.24977658689022064,
+      "eval_runtime": 10.5663,
+      "eval_samples_per_second": 9.464,
+      "eval_steps_per_second": 9.464,
+      "step": 115000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.634785035427014e-06,
+      "loss": 0.2404,
+      "step": 115005
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.6331328777733e-06,
+      "loss": 0.2405,
+      "step": 115010
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.6314809845699355e-06,
+      "loss": 0.2592,
+      "step": 115015
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.629829355838372e-06,
+      "loss": 0.2474,
+      "step": 115020
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.628177991600047e-06,
+      "loss": 0.2443,
+      "step": 115025
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.626526891876398e-06,
+      "loss": 0.2654,
+      "step": 115030
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.624876056688879e-06,
+      "loss": 0.2398,
+      "step": 115035
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.623225486058918e-06,
+      "loss": 0.2607,
+      "step": 115040
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.621575180007937e-06,
+      "loss": 0.2391,
+      "step": 115045
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.61992513855738e-06,
+      "loss": 0.2658,
+      "step": 115050
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.61827536172866e-06,
+      "loss": 0.2454,
+      "step": 115055
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.616625849543199e-06,
+      "loss": 0.271,
+      "step": 115060
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.61497660202242e-06,
+      "loss": 0.2539,
+      "step": 115065
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.6133276191877405e-06,
+      "loss": 0.2412,
+      "step": 115070
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.611678901060565e-06,
+      "loss": 0.2474,
+      "step": 115075
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.610030447662303e-06,
+      "loss": 0.2609,
+      "step": 115080
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.608382259014355e-06,
+      "loss": 0.2499,
+      "step": 115085
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.606734335138127e-06,
+      "loss": 0.2305,
+      "step": 115090
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.6050866760550145e-06,
+      "loss": 0.2512,
+      "step": 115095
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.6034392817864145e-06,
+      "loss": 0.2668,
+      "step": 115100
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.601792152353704e-06,
+      "loss": 0.2516,
+      "step": 115105
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.600145287778285e-06,
+      "loss": 0.248,
+      "step": 115110
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.598498688081532e-06,
+      "loss": 0.2313,
+      "step": 115115
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.596852353284834e-06,
+      "loss": 0.2647,
+      "step": 115120
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.595206283409554e-06,
+      "loss": 0.2435,
+      "step": 115125
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.59356047847708e-06,
+      "loss": 0.2797,
+      "step": 115130
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.591914938508776e-06,
+      "loss": 0.2365,
+      "step": 115135
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.590269663526006e-06,
+      "loss": 0.2377,
+      "step": 115140
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.5886246535501275e-06,
+      "loss": 0.2622,
+      "step": 115145
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.586979908602515e-06,
+      "loss": 0.2367,
+      "step": 115150
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.585335428704512e-06,
+      "loss": 0.254,
+      "step": 115155
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.583691213877472e-06,
+      "loss": 0.2522,
+      "step": 115160
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.5820472641427495e-06,
+      "loss": 0.249,
+      "step": 115165
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.580403579521683e-06,
+      "loss": 0.2377,
+      "step": 115170
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.578760160035625e-06,
+      "loss": 0.2376,
+      "step": 115175
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.577117005705902e-06,
+      "loss": 0.2611,
+      "step": 115180
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.575474116553863e-06,
+      "loss": 0.2428,
+      "step": 115185
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.573831492600831e-06,
+      "loss": 0.283,
+      "step": 115190
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.5721891338681345e-06,
+      "loss": 0.2481,
+      "step": 115195
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.570547040377096e-06,
+      "loss": 0.2675,
+      "step": 115200
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.568905212149047e-06,
+      "loss": 0.2363,
+      "step": 115205
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.567263649205297e-06,
+      "loss": 0.2414,
+      "step": 115210
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.565622351567164e-06,
+      "loss": 0.2436,
+      "step": 115215
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.56398131925595e-06,
+      "loss": 0.2647,
+      "step": 115220
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.5623405522929735e-06,
+      "loss": 0.2445,
+      "step": 115225
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.56070005069954e-06,
+      "loss": 0.2502,
+      "step": 115230
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.559059814496947e-06,
+      "loss": 0.2407,
+      "step": 115235
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.557419843706484e-06,
+      "loss": 0.2386,
+      "step": 115240
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.555780138349461e-06,
+      "loss": 0.2654,
+      "step": 115245
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.55414069844716e-06,
+      "loss": 0.2552,
+      "step": 115250
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.552501524020863e-06,
+      "loss": 0.2426,
+      "step": 115255
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.550862615091855e-06,
+      "loss": 0.2663,
+      "step": 115260
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.549223971681426e-06,
+      "loss": 0.2714,
+      "step": 115265
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.547585593810846e-06,
+      "loss": 0.2595,
+      "step": 115270
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.545947481501384e-06,
+      "loss": 0.2411,
+      "step": 115275
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.544309634774313e-06,
+      "loss": 0.2261,
+      "step": 115280
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.542672053650904e-06,
+      "loss": 0.2438,
+      "step": 115285
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.541034738152419e-06,
+      "loss": 0.2659,
+      "step": 115290
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.539397688300109e-06,
+      "loss": 0.2445,
+      "step": 115295
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.537760904115243e-06,
+      "loss": 0.2389,
+      "step": 115300
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.5361243856190645e-06,
+      "loss": 0.2672,
+      "step": 115305
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.534488132832826e-06,
+      "loss": 0.2355,
+      "step": 115310
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.532852145777763e-06,
+      "loss": 0.2406,
+      "step": 115315
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.531216424475135e-06,
+      "loss": 0.2558,
+      "step": 115320
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.529580968946173e-06,
+      "loss": 0.235,
+      "step": 115325
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.5279457792121034e-06,
+      "loss": 0.2618,
+      "step": 115330
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.526310855294166e-06,
+      "loss": 0.2475,
+      "step": 115335
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.524676197213598e-06,
+      "loss": 0.2574,
+      "step": 115340
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.523041804991615e-06,
+      "loss": 0.235,
+      "step": 115345
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.521407678649439e-06,
+      "loss": 0.2392,
+      "step": 115350
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.519773818208284e-06,
+      "loss": 0.2594,
+      "step": 115355
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.518140223689371e-06,
+      "loss": 0.249,
+      "step": 115360
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.5165068951139115e-06,
+      "loss": 0.2587,
+      "step": 115365
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.514873832503111e-06,
+      "loss": 0.2529,
+      "step": 115370
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.513241035878168e-06,
+      "loss": 0.2408,
+      "step": 115375
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.511608505260295e-06,
+      "loss": 0.2692,
+      "step": 115380
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.509976240670677e-06,
+      "loss": 0.2393,
+      "step": 115385
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.508344242130519e-06,
+      "loss": 0.2522,
+      "step": 115390
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.506712509661002e-06,
+      "loss": 0.2416,
+      "step": 115395
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.505081043283321e-06,
+      "loss": 0.2572,
+      "step": 115400
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.503449843018659e-06,
+      "loss": 0.2597,
+      "step": 115405
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.501818908888184e-06,
+      "loss": 0.2414,
+      "step": 115410
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.500188240913089e-06,
+      "loss": 0.2503,
+      "step": 115415
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.498557839114542e-06,
+      "loss": 0.2564,
+      "step": 115420
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.496927703513709e-06,
+      "loss": 0.2322,
+      "step": 115425
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.495297834131751e-06,
+      "loss": 0.2533,
+      "step": 115430
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.493668230989842e-06,
+      "loss": 0.2587,
+      "step": 115435
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.492038894109133e-06,
+      "loss": 0.2619,
+      "step": 115440
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.490409823510791e-06,
+      "loss": 0.2575,
+      "step": 115445
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.488781019215951e-06,
+      "loss": 0.2596,
+      "step": 115450
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.487152481245782e-06,
+      "loss": 0.2451,
+      "step": 115455
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.4855242096214195e-06,
+      "loss": 0.2547,
+      "step": 115460
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.483896204364002e-06,
+      "loss": 0.248,
+      "step": 115465
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.482268465494669e-06,
+      "loss": 0.2582,
+      "step": 115470
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.480640993034563e-06,
+      "loss": 0.2328,
+      "step": 115475
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.479013787004813e-06,
+      "loss": 0.2704,
+      "step": 115480
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.477386847426543e-06,
+      "loss": 0.2622,
+      "step": 115485
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.475760174320872e-06,
+      "loss": 0.2616,
+      "step": 115490
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.474133767708938e-06,
+      "loss": 0.2608,
+      "step": 115495
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.472507627611844e-06,
+      "loss": 0.2762,
+      "step": 115500
+    },
+    {
+      "epoch": 4.06,
+      "eval_loss": 0.2497061789035797,
+      "eval_runtime": 10.5326,
+      "eval_samples_per_second": 9.494,
+      "eval_steps_per_second": 9.494,
+      "step": 115500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.470881754050716e-06,
+      "loss": 0.254,
+      "step": 115505
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.469256147046652e-06,
+      "loss": 0.2564,
+      "step": 115510
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.467630806620773e-06,
+      "loss": 0.2523,
+      "step": 115515
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.4660057327941765e-06,
+      "loss": 0.2488,
+      "step": 115520
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.464380925587958e-06,
+      "loss": 0.2506,
+      "step": 115525
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.462756385023226e-06,
+      "loss": 0.2627,
+      "step": 115530
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.461132111121066e-06,
+      "loss": 0.2383,
+      "step": 115535
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.459508103902573e-06,
+      "loss": 0.2519,
+      "step": 115540
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.457884363388823e-06,
+      "loss": 0.2319,
+      "step": 115545
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.456260889600913e-06,
+      "loss": 0.2517,
+      "step": 115550
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.454637682559909e-06,
+      "loss": 0.2411,
+      "step": 115555
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.453014742286901e-06,
+      "loss": 0.2421,
+      "step": 115560
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.451392068802951e-06,
+      "loss": 0.2497,
+      "step": 115565
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.449769662129138e-06,
+      "loss": 0.2512,
+      "step": 115570
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.448147522286525e-06,
+      "loss": 0.256,
+      "step": 115575
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.446525649296168e-06,
+      "loss": 0.2621,
+      "step": 115580
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.444904043179127e-06,
+      "loss": 0.2608,
+      "step": 115585
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.443282703956467e-06,
+      "loss": 0.2618,
+      "step": 115590
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.4416616316492326e-06,
+      "loss": 0.2595,
+      "step": 115595
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.440040826278468e-06,
+      "loss": 0.2439,
+      "step": 115600
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.438420287865231e-06,
+      "loss": 0.2538,
+      "step": 115605
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.436800016430548e-06,
+      "loss": 0.2538,
+      "step": 115610
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.435180011995471e-06,
+      "loss": 0.2486,
+      "step": 115615
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.433560274581031e-06,
+      "loss": 0.2631,
+      "step": 115620
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.431940804208248e-06,
+      "loss": 0.227,
+      "step": 115625
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.4303216008981685e-06,
+      "loss": 0.2532,
+      "step": 115630
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.428702664671805e-06,
+      "loss": 0.2379,
+      "step": 115635
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.427083995550174e-06,
+      "loss": 0.2311,
+      "step": 115640
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.425465593554304e-06,
+      "loss": 0.2583,
+      "step": 115645
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.423847458705207e-06,
+      "loss": 0.2367,
+      "step": 115650
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.422229591023891e-06,
+      "loss": 0.2599,
+      "step": 115655
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.420611990531353e-06,
+      "loss": 0.2391,
+      "step": 115660
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.418994657248607e-06,
+      "loss": 0.2353,
+      "step": 115665
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.417377591196659e-06,
+      "loss": 0.2539,
+      "step": 115670
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.4157607923965006e-06,
+      "loss": 0.2533,
+      "step": 115675
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.414144260869113e-06,
+      "loss": 0.2263,
+      "step": 115680
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.412527996635504e-06,
+      "loss": 0.251,
+      "step": 115685
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.41091199971665e-06,
+      "loss": 0.2435,
+      "step": 115690
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.409296270133537e-06,
+      "loss": 0.2552,
+      "step": 115695
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.407680807907133e-06,
+      "loss": 0.2533,
+      "step": 115700
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.40606561305843e-06,
+      "loss": 0.2538,
+      "step": 115705
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.404450685608394e-06,
+      "loss": 0.2393,
+      "step": 115710
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.402836025577986e-06,
+      "loss": 0.2787,
+      "step": 115715
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.401221632988178e-06,
+      "loss": 0.2305,
+      "step": 115720
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3996075078599405e-06,
+      "loss": 0.2592,
+      "step": 115725
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3979936502142186e-06,
+      "loss": 0.2396,
+      "step": 115730
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.396380060071975e-06,
+      "loss": 0.2475,
+      "step": 115735
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3947667374541534e-06,
+      "loss": 0.2531,
+      "step": 115740
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.3931536823817095e-06,
+      "loss": 0.2214,
+      "step": 115745
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.391540894875587e-06,
+      "loss": 0.2422,
+      "step": 115750
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.389928374956717e-06,
+      "loss": 0.2586,
+      "step": 115755
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.38831612264605e-06,
+      "loss": 0.2234,
+      "step": 115760
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.386704137964517e-06,
+      "loss": 0.261,
+      "step": 115765
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.385092420933037e-06,
+      "loss": 0.2488,
+      "step": 115770
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.383480971572549e-06,
+      "loss": 0.2711,
+      "step": 115775
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.381869789903978e-06,
+      "loss": 0.2592,
+      "step": 115780
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.38025887594824e-06,
+      "loss": 0.2521,
+      "step": 115785
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.378648229726251e-06,
+      "loss": 0.2311,
+      "step": 115790
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.377037851258919e-06,
+      "loss": 0.2464,
+      "step": 115795
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.375427740567167e-06,
+      "loss": 0.2263,
+      "step": 115800
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.373817897671895e-06,
+      "loss": 0.2361,
+      "step": 115805
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.372208322594001e-06,
+      "loss": 0.2518,
+      "step": 115810
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.370599015354382e-06,
+      "loss": 0.2433,
+      "step": 115815
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.368989975973947e-06,
+      "loss": 0.2632,
+      "step": 115820
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.367381204473575e-06,
+      "loss": 0.2546,
+      "step": 115825
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.365772700874165e-06,
+      "loss": 0.2687,
+      "step": 115830
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.364164465196594e-06,
+      "loss": 0.241,
+      "step": 115835
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.36255649746175e-06,
+      "loss": 0.2409,
+      "step": 115840
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.360948797690515e-06,
+      "loss": 0.2332,
+      "step": 115845
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3593413659037544e-06,
+      "loss": 0.2577,
+      "step": 115850
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.357734202122338e-06,
+      "loss": 0.2678,
+      "step": 115855
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.356127306367147e-06,
+      "loss": 0.257,
+      "step": 115860
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.354520678659035e-06,
+      "loss": 0.2372,
+      "step": 115865
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.352914319018864e-06,
+      "loss": 0.2447,
+      "step": 115870
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.351308227467496e-06,
+      "loss": 0.2486,
+      "step": 115875
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.349702404025777e-06,
+      "loss": 0.2786,
+      "step": 115880
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.348096848714573e-06,
+      "loss": 0.2499,
+      "step": 115885
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.346491561554711e-06,
+      "loss": 0.252,
+      "step": 115890
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.344886542567053e-06,
+      "loss": 0.264,
+      "step": 115895
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.34328179177243e-06,
+      "loss": 0.2619,
+      "step": 115900
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.34167730919168e-06,
+      "loss": 0.2441,
+      "step": 115905
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.340073094845631e-06,
+      "loss": 0.2411,
+      "step": 115910
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.338469148755123e-06,
+      "loss": 0.2495,
+      "step": 115915
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.336865470940976e-06,
+      "loss": 0.2487,
+      "step": 115920
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.335262061424011e-06,
+      "loss": 0.2509,
+      "step": 115925
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.333658920225048e-06,
+      "loss": 0.2359,
+      "step": 115930
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.332056047364908e-06,
+      "loss": 0.263,
+      "step": 115935
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3304534428643935e-06,
+      "loss": 0.2865,
+      "step": 115940
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.328851106744325e-06,
+      "loss": 0.2596,
+      "step": 115945
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.327249039025497e-06,
+      "loss": 0.2376,
+      "step": 115950
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3256472397287224e-06,
+      "loss": 0.247,
+      "step": 115955
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.324045708874793e-06,
+      "loss": 0.252,
+      "step": 115960
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.322444446484505e-06,
+      "loss": 0.2502,
+      "step": 115965
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.32084345257864e-06,
+      "loss": 0.2715,
+      "step": 115970
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.319242727178005e-06,
+      "loss": 0.2224,
+      "step": 115975
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.317642270303371e-06,
+      "loss": 0.2457,
+      "step": 115980
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.316042081975519e-06,
+      "loss": 0.2356,
+      "step": 115985
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.314442162215232e-06,
+      "loss": 0.2499,
+      "step": 115990
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.312842511043274e-06,
+      "loss": 0.2539,
+      "step": 115995
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.311243128480433e-06,
+      "loss": 0.2663,
+      "step": 116000
+    },
+    {
+      "epoch": 4.08,
+      "eval_loss": 0.24984586238861084,
+      "eval_runtime": 10.5713,
+      "eval_samples_per_second": 9.46,
+      "eval_steps_per_second": 9.46,
+      "step": 116000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.309644014547457e-06,
+      "loss": 0.2392,
+      "step": 116005
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.308045169265126e-06,
+      "loss": 0.2409,
+      "step": 116010
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.306446592654189e-06,
+      "loss": 0.2452,
+      "step": 116015
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.304848284735408e-06,
+      "loss": 0.2598,
+      "step": 116020
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3032502455295275e-06,
+      "loss": 0.2368,
+      "step": 116025
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.301652475057308e-06,
+      "loss": 0.246,
+      "step": 116030
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.3000549733394905e-06,
+      "loss": 0.2407,
+      "step": 116035
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.29845774039681e-06,
+      "loss": 0.2708,
+      "step": 116040
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.29686077625002e-06,
+      "loss": 0.2366,
+      "step": 116045
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.295264080919842e-06,
+      "loss": 0.2562,
+      "step": 116050
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.293667654427022e-06,
+      "loss": 0.2507,
+      "step": 116055
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.292071496792282e-06,
+      "loss": 0.2261,
+      "step": 116060
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.29047560803634e-06,
+      "loss": 0.2649,
+      "step": 116065
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.2888799881799305e-06,
+      "loss": 0.2358,
+      "step": 116070
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.287284637243763e-06,
+      "loss": 0.2658,
+      "step": 116075
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.2856895552485565e-06,
+      "loss": 0.2595,
+      "step": 116080
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.284094742215014e-06,
+      "loss": 0.2434,
+      "step": 116085
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.282500198163855e-06,
+      "loss": 0.2826,
+      "step": 116090
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.280905923115777e-06,
+      "loss": 0.2464,
+      "step": 116095
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.279311917091475e-06,
+      "loss": 0.2642,
+      "step": 116100
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.277718180111653e-06,
+      "loss": 0.2515,
+      "step": 116105
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.276124712197011e-06,
+      "loss": 0.2432,
+      "step": 116110
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.2745315133682295e-06,
+      "loss": 0.2789,
+      "step": 116115
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.2729385836459915e-06,
+      "loss": 0.2398,
+      "step": 116120
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.2713459230509955e-06,
+      "loss": 0.2717,
+      "step": 116125
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.269753531603909e-06,
+      "loss": 0.255,
+      "step": 116130
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.268161409325413e-06,
+      "loss": 0.2511,
+      "step": 116135
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.2665695562361715e-06,
+      "loss": 0.2584,
+      "step": 116140
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.264977972356868e-06,
+      "loss": 0.249,
+      "step": 116145
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.2633866577081576e-06,
+      "loss": 0.2773,
+      "step": 116150
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.261795612310701e-06,
+      "loss": 0.2374,
+      "step": 116155
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.260204836185161e-06,
+      "loss": 0.2514,
+      "step": 116160
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.2586143293522005e-06,
+      "loss": 0.2469,
+      "step": 116165
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.257024091832463e-06,
+      "loss": 0.2366,
+      "step": 116170
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.255434123646598e-06,
+      "loss": 0.2403,
+      "step": 116175
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.253844424815243e-06,
+      "loss": 0.2289,
+      "step": 116180
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.252254995359051e-06,
+      "loss": 0.2692,
+      "step": 116185
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.250665835298656e-06,
+      "loss": 0.2502,
+      "step": 116190
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.249076944654692e-06,
+      "loss": 0.246,
+      "step": 116195
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.247488323447779e-06,
+      "loss": 0.2582,
+      "step": 116200
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.245899971698564e-06,
+      "loss": 0.2559,
+      "step": 116205
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.244311889427652e-06,
+      "loss": 0.2409,
+      "step": 116210
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.242724076655671e-06,
+      "loss": 0.2466,
+      "step": 116215
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.241136533403245e-06,
+      "loss": 0.2795,
+      "step": 116220
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.2395492596909806e-06,
+      "loss": 0.2432,
+      "step": 116225
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.237962255539488e-06,
+      "loss": 0.2446,
+      "step": 116230
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.236375520969366e-06,
+      "loss": 0.2807,
+      "step": 116235
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.234789056001231e-06,
+      "loss": 0.2676,
+      "step": 116240
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.233202860655675e-06,
+      "loss": 0.2475,
+      "step": 116245
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.231616934953292e-06,
+      "loss": 0.2703,
+      "step": 116250
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.230031278914673e-06,
+      "loss": 0.2574,
+      "step": 116255
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.228445892560415e-06,
+      "loss": 0.2597,
+      "step": 116260
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.226860775911093e-06,
+      "loss": 0.2352,
+      "step": 116265
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.225592876793025e-06,
+      "loss": 0.2428,
+      "step": 116270
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.224008245664465e-06,
+      "loss": 0.2457,
+      "step": 116275
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.222423884298468e-06,
+      "loss": 0.2538,
+      "step": 116280
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.220839792715597e-06,
+      "loss": 0.2322,
+      "step": 116285
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.219255970936431e-06,
+      "loss": 0.2299,
+      "step": 116290
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.217672418981533e-06,
+      "loss": 0.2667,
+      "step": 116295
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.216089136871457e-06,
+      "loss": 0.2434,
+      "step": 116300
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.214506124626766e-06,
+      "loss": 0.2448,
+      "step": 116305
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.212923382268022e-06,
+      "loss": 0.2586,
+      "step": 116310
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.211340909815767e-06,
+      "loss": 0.2356,
+      "step": 116315
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.209758707290551e-06,
+      "loss": 0.2419,
+      "step": 116320
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.20817677471291e-06,
+      "loss": 0.232,
+      "step": 116325
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.206595112103398e-06,
+      "loss": 0.2493,
+      "step": 116330
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.205013719482545e-06,
+      "loss": 0.262,
+      "step": 116335
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.203432596870885e-06,
+      "loss": 0.2578,
+      "step": 116340
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.20185174428894e-06,
+      "loss": 0.2718,
+      "step": 116345
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.200271161757249e-06,
+      "loss": 0.2511,
+      "step": 116350
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.198690849296325e-06,
+      "loss": 0.2662,
+      "step": 116355
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.197110806926696e-06,
+      "loss": 0.2592,
+      "step": 116360
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.19553103466887e-06,
+      "loss": 0.2554,
+      "step": 116365
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.1939515325433665e-06,
+      "loss": 0.2606,
+      "step": 116370
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.192372300570693e-06,
+      "loss": 0.2593,
+      "step": 116375
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.190793338771351e-06,
+      "loss": 0.2644,
+      "step": 116380
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.189214647165837e-06,
+      "loss": 0.2494,
+      "step": 116385
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.187636225774663e-06,
+      "loss": 0.2551,
+      "step": 116390
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.186058074618318e-06,
+      "loss": 0.2457,
+      "step": 116395
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.184480193717285e-06,
+      "loss": 0.2555,
+      "step": 116400
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.182902583092063e-06,
+      "loss": 0.2545,
+      "step": 116405
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.181325242763129e-06,
+      "loss": 0.2385,
+      "step": 116410
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.179748172750972e-06,
+      "loss": 0.229,
+      "step": 116415
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.178171373076056e-06,
+      "loss": 0.2596,
+      "step": 116420
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.176594843758869e-06,
+      "loss": 0.2502,
+      "step": 116425
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.175018584819876e-06,
+      "loss": 0.2371,
+      "step": 116430
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.173442596279542e-06,
+      "loss": 0.234,
+      "step": 116435
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.171866878158326e-06,
+      "loss": 0.2441,
+      "step": 116440
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.170291430476697e-06,
+      "loss": 0.2283,
+      "step": 116445
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.168716253255106e-06,
+      "loss": 0.2577,
+      "step": 116450
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.167141346514006e-06,
+      "loss": 0.2455,
+      "step": 116455
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.165566710273841e-06,
+      "loss": 0.2543,
+      "step": 116460
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.163992344555062e-06,
+      "loss": 0.2596,
+      "step": 116465
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.162418249378117e-06,
+      "loss": 0.2453,
+      "step": 116470
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.160844424763438e-06,
+      "loss": 0.2464,
+      "step": 116475
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.159270870731455e-06,
+      "loss": 0.2531,
+      "step": 116480
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.157697587302612e-06,
+      "loss": 0.2168,
+      "step": 116485
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.156124574497327e-06,
+      "loss": 0.2791,
+      "step": 116490
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.154551832336031e-06,
+      "loss": 0.2586,
+      "step": 116495
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.152979360839135e-06,
+      "loss": 0.2233,
+      "step": 116500
+    },
+    {
+      "epoch": 4.1,
+      "eval_loss": 0.2499181181192398,
+      "eval_runtime": 10.5421,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 116500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.151407160027068e-06,
+      "loss": 0.2409,
+      "step": 116505
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.149835229920241e-06,
+      "loss": 0.2562,
+      "step": 116510
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.148263570539057e-06,
+      "loss": 0.2317,
+      "step": 116515
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.146692181903933e-06,
+      "loss": 0.2491,
+      "step": 116520
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.1451210640352625e-06,
+      "loss": 0.2983,
+      "step": 116525
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.1435502169534595e-06,
+      "loss": 0.251,
+      "step": 116530
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.141979640678903e-06,
+      "loss": 0.2506,
+      "step": 116535
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.140409335232004e-06,
+      "loss": 0.2547,
+      "step": 116540
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.13883930063314e-06,
+      "loss": 0.242,
+      "step": 116545
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.137269536902699e-06,
+      "loss": 0.2716,
+      "step": 116550
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.135700044061061e-06,
+      "loss": 0.2277,
+      "step": 116555
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.134130822128612e-06,
+      "loss": 0.2366,
+      "step": 116560
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.132561871125723e-06,
+      "loss": 0.2422,
+      "step": 116565
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.1309931910727604e-06,
+      "loss": 0.2761,
+      "step": 116570
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.129424781990104e-06,
+      "loss": 0.243,
+      "step": 116575
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.127856643898109e-06,
+      "loss": 0.2461,
+      "step": 116580
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.126288776817142e-06,
+      "loss": 0.2552,
+      "step": 116585
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.1247211807675625e-06,
+      "loss": 0.2589,
+      "step": 116590
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.123153855769712e-06,
+      "loss": 0.2355,
+      "step": 116595
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.121586801843958e-06,
+      "loss": 0.267,
+      "step": 116600
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.12002001901064e-06,
+      "loss": 0.2511,
+      "step": 116605
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.118453507290102e-06,
+      "loss": 0.2624,
+      "step": 116610
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.116887266702676e-06,
+      "loss": 0.2692,
+      "step": 116615
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.115321297268715e-06,
+      "loss": 0.2561,
+      "step": 116620
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.113755599008537e-06,
+      "loss": 0.2559,
+      "step": 116625
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.112190171942482e-06,
+      "loss": 0.2299,
+      "step": 116630
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.110625016090866e-06,
+      "loss": 0.2546,
+      "step": 116635
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.109060131474026e-06,
+      "loss": 0.2338,
+      "step": 116640
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.10749551811227e-06,
+      "loss": 0.2297,
+      "step": 116645
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.105931176025912e-06,
+      "loss": 0.2499,
+      "step": 116650
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.104367105235274e-06,
+      "loss": 0.268,
+      "step": 116655
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.102803305760658e-06,
+      "loss": 0.2344,
+      "step": 116660
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.101239777622368e-06,
+      "loss": 0.2388,
+      "step": 116665
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.099676520840701e-06,
+      "loss": 0.2223,
+      "step": 116670
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.098113535435968e-06,
+      "loss": 0.2801,
+      "step": 116675
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.096550821428455e-06,
+      "loss": 0.259,
+      "step": 116680
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.094988378838446e-06,
+      "loss": 0.2475,
+      "step": 116685
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.093426207686238e-06,
+      "loss": 0.233,
+      "step": 116690
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.091864307992119e-06,
+      "loss": 0.2482,
+      "step": 116695
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.090302679776361e-06,
+      "loss": 0.2608,
+      "step": 116700
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.0887413230592444e-06,
+      "loss": 0.2399,
+      "step": 116705
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.087180237861035e-06,
+      "loss": 0.2476,
+      "step": 116710
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.0856194242020144e-06,
+      "loss": 0.2735,
+      "step": 116715
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.08405888210244e-06,
+      "loss": 0.2621,
+      "step": 116720
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.08249861158258e-06,
+      "loss": 0.244,
+      "step": 116725
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.080938612662682e-06,
+      "loss": 0.2618,
+      "step": 116730
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.079378885363017e-06,
+      "loss": 0.2448,
+      "step": 116735
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.0778194297038245e-06,
+      "loss": 0.2634,
+      "step": 116740
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.076260245705362e-06,
+      "loss": 0.2495,
+      "step": 116745
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.074701333387867e-06,
+      "loss": 0.2487,
+      "step": 116750
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.073142692771592e-06,
+      "loss": 0.2425,
+      "step": 116755
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.071584323876765e-06,
+      "loss": 0.2397,
+      "step": 116760
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.0700262267236195e-06,
+      "loss": 0.267,
+      "step": 116765
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.068468401332395e-06,
+      "loss": 0.2278,
+      "step": 116770
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.066910847723315e-06,
+      "loss": 0.2298,
+      "step": 116775
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.065353565916599e-06,
+      "loss": 0.2571,
+      "step": 116780
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.063796555932467e-06,
+      "loss": 0.2576,
+      "step": 116785
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.062239817791147e-06,
+      "loss": 0.2294,
+      "step": 116790
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.060683351512837e-06,
+      "loss": 0.2432,
+      "step": 116795
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.0591271571177596e-06,
+      "loss": 0.2427,
+      "step": 116800
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.057571234626109e-06,
+      "loss": 0.2379,
+      "step": 116805
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.056015584058101e-06,
+      "loss": 0.2645,
+      "step": 116810
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.054460205433927e-06,
+      "loss": 0.2201,
+      "step": 116815
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.052905098773785e-06,
+      "loss": 0.2621,
+      "step": 116820
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.0513502640978584e-06,
+      "loss": 0.2348,
+      "step": 116825
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.049795701426351e-06,
+      "loss": 0.2692,
+      "step": 116830
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.048241410779438e-06,
+      "loss": 0.2655,
+      "step": 116835
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.046687392177303e-06,
+      "loss": 0.2686,
+      "step": 116840
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.045133645640118e-06,
+      "loss": 0.2371,
+      "step": 116845
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.043580171188061e-06,
+      "loss": 0.2566,
+      "step": 116850
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.042026968841311e-06,
+      "loss": 0.2576,
+      "step": 116855
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.040474038620032e-06,
+      "loss": 0.2379,
+      "step": 116860
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.0389213805443775e-06,
+      "loss": 0.2348,
+      "step": 116865
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.037368994634519e-06,
+      "loss": 0.2544,
+      "step": 116870
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.035816880910612e-06,
+      "loss": 0.2621,
+      "step": 116875
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.034265039392798e-06,
+      "loss": 0.2743,
+      "step": 116880
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.032713470101243e-06,
+      "loss": 0.2546,
+      "step": 116885
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.031162173056086e-06,
+      "loss": 0.2264,
+      "step": 116890
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.029611148277468e-06,
+      "loss": 0.2662,
+      "step": 116895
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.028060395785524e-06,
+      "loss": 0.2493,
+      "step": 116900
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.026509915600393e-06,
+      "loss": 0.2831,
+      "step": 116905
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.024959707742218e-06,
+      "loss": 0.2659,
+      "step": 116910
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.023409772231115e-06,
+      "loss": 0.246,
+      "step": 116915
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.021860109087208e-06,
+      "loss": 0.2348,
+      "step": 116920
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.020310718330628e-06,
+      "loss": 0.2468,
+      "step": 116925
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.018761599981488e-06,
+      "loss": 0.2415,
+      "step": 116930
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.017212754059901e-06,
+      "loss": 0.2619,
+      "step": 116935
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.015664180585971e-06,
+      "loss": 0.2512,
+      "step": 116940
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.0141158795798185e-06,
+      "loss": 0.2675,
+      "step": 116945
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.012567851061544e-06,
+      "loss": 0.243,
+      "step": 116950
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.011020095051235e-06,
+      "loss": 0.2618,
+      "step": 116955
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.009472611569006e-06,
+      "loss": 0.2237,
+      "step": 116960
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.007925400634935e-06,
+      "loss": 0.2513,
+      "step": 116965
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.0063784622691255e-06,
+      "loss": 0.2646,
+      "step": 116970
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.004831796491653e-06,
+      "loss": 0.2411,
+      "step": 116975
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.003285403322602e-06,
+      "loss": 0.244,
+      "step": 116980
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.001739282782055e-06,
+      "loss": 0.2428,
+      "step": 116985
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.000193434890087e-06,
+      "loss": 0.2518,
+      "step": 116990
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.99864785966676e-06,
+      "loss": 0.2488,
+      "step": 116995
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.997102557132157e-06,
+      "loss": 0.25,
+      "step": 117000
+    },
+    {
+      "epoch": 4.12,
+      "eval_loss": 0.24963203072547913,
+      "eval_runtime": 10.5439,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 117000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.995557527306337e-06,
+      "loss": 0.2382,
+      "step": 117005
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.994012770209354e-06,
+      "loss": 0.2457,
+      "step": 117010
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.992468285861275e-06,
+      "loss": 0.2589,
+      "step": 117015
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9909240742821484e-06,
+      "loss": 0.2411,
+      "step": 117020
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.989380135492032e-06,
+      "loss": 0.2512,
+      "step": 117025
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.987836469510967e-06,
+      "loss": 0.2572,
+      "step": 117030
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.986293076358993e-06,
+      "loss": 0.2433,
+      "step": 117035
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9847499560561605e-06,
+      "loss": 0.2575,
+      "step": 117040
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9832071086225e-06,
+      "loss": 0.2379,
+      "step": 117045
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.981664534078045e-06,
+      "loss": 0.2512,
+      "step": 117050
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.980122232442817e-06,
+      "loss": 0.2441,
+      "step": 117055
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.978580203736857e-06,
+      "loss": 0.2501,
+      "step": 117060
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.977038447980172e-06,
+      "loss": 0.2599,
+      "step": 117065
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.975496965192796e-06,
+      "loss": 0.2381,
+      "step": 117070
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9739557553947285e-06,
+      "loss": 0.2545,
+      "step": 117075
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.972414818605996e-06,
+      "loss": 0.2488,
+      "step": 117080
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.970874154846599e-06,
+      "loss": 0.2561,
+      "step": 117085
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.969333764136543e-06,
+      "loss": 0.2487,
+      "step": 117090
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.967793646495821e-06,
+      "loss": 0.2425,
+      "step": 117095
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.966253801944444e-06,
+      "loss": 0.2653,
+      "step": 117100
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9647142305024e-06,
+      "loss": 0.2509,
+      "step": 117105
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.963174932189672e-06,
+      "loss": 0.241,
+      "step": 117110
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9616359070262625e-06,
+      "loss": 0.2357,
+      "step": 117115
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9600971550321444e-06,
+      "loss": 0.2503,
+      "step": 117120
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.958558676227292e-06,
+      "loss": 0.2639,
+      "step": 117125
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9570204706316895e-06,
+      "loss": 0.2484,
+      "step": 117130
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.955482538265312e-06,
+      "loss": 0.2749,
+      "step": 117135
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.953944879148128e-06,
+      "loss": 0.2528,
+      "step": 117140
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9524074933000974e-06,
+      "loss": 0.2484,
+      "step": 117145
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.950870380741181e-06,
+      "loss": 0.245,
+      "step": 117150
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.949333541491346e-06,
+      "loss": 0.2538,
+      "step": 117155
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.947796975570542e-06,
+      "loss": 0.258,
+      "step": 117160
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.946260682998717e-06,
+      "loss": 0.2483,
+      "step": 117165
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.94472466379582e-06,
+      "loss": 0.2473,
+      "step": 117170
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.943188917981802e-06,
+      "loss": 0.2507,
+      "step": 117175
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.941653445576596e-06,
+      "loss": 0.2639,
+      "step": 117180
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.940118246600144e-06,
+      "loss": 0.2472,
+      "step": 117185
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.938583321072373e-06,
+      "loss": 0.2471,
+      "step": 117190
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.937048669013224e-06,
+      "loss": 0.247,
+      "step": 117195
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.935514290442616e-06,
+      "loss": 0.2579,
+      "step": 117200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.933980185380473e-06,
+      "loss": 0.261,
+      "step": 117205
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.932446353846708e-06,
+      "loss": 0.2598,
+      "step": 117210
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.930912795861249e-06,
+      "loss": 0.262,
+      "step": 117215
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.929379511444006e-06,
+      "loss": 0.2617,
+      "step": 117220
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9278465006148735e-06,
+      "loss": 0.2531,
+      "step": 117225
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.926313763393774e-06,
+      "loss": 0.2529,
+      "step": 117230
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.924781299800598e-06,
+      "loss": 0.2453,
+      "step": 117235
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.9232491098552535e-06,
+      "loss": 0.2564,
+      "step": 117240
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.921717193577621e-06,
+      "loss": 0.2345,
+      "step": 117245
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.920185550987607e-06,
+      "loss": 0.2375,
+      "step": 117250
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.918654182105092e-06,
+      "loss": 0.2377,
+      "step": 117255
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.917123086949959e-06,
+      "loss": 0.2494,
+      "step": 117260
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.915592265542081e-06,
+      "loss": 0.2634,
+      "step": 117265
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.914061717901352e-06,
+      "loss": 0.2385,
+      "step": 117270
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.9125314440476316e-06,
+      "loss": 0.2662,
+      "step": 117275
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.911001444000795e-06,
+      "loss": 0.2587,
+      "step": 117280
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.9094717177806995e-06,
+      "loss": 0.2776,
+      "step": 117285
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.907942265407216e-06,
+      "loss": 0.2637,
+      "step": 117290
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.906413086900207e-06,
+      "loss": 0.2112,
+      "step": 117295
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.904884182279523e-06,
+      "loss": 0.259,
+      "step": 117300
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.903355551565011e-06,
+      "loss": 0.2503,
+      "step": 117305
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.9018271947765274e-06,
+      "loss": 0.2463,
+      "step": 117310
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.900299111933917e-06,
+      "loss": 0.2535,
+      "step": 117315
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.898771303057014e-06,
+      "loss": 0.2524,
+      "step": 117320
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.8972437681656535e-06,
+      "loss": 0.258,
+      "step": 117325
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.895716507279684e-06,
+      "loss": 0.2496,
+      "step": 117330
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.8941895204189256e-06,
+      "loss": 0.2401,
+      "step": 117335
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.8926628076032e-06,
+      "loss": 0.2656,
+      "step": 117340
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.891136368852341e-06,
+      "loss": 0.2565,
+      "step": 117345
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.8896102041861695e-06,
+      "loss": 0.2567,
+      "step": 117350
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.888084313624496e-06,
+      "loss": 0.2358,
+      "step": 117355
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.886558697187129e-06,
+      "loss": 0.2781,
+      "step": 117360
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.885033354893888e-06,
+      "loss": 0.2804,
+      "step": 117365
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.883508286764576e-06,
+      "loss": 0.257,
+      "step": 117370
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.881983492818991e-06,
+      "loss": 0.2531,
+      "step": 117375
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.880458973076928e-06,
+      "loss": 0.2463,
+      "step": 117380
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.878934727558192e-06,
+      "loss": 0.2458,
+      "step": 117385
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.877410756282571e-06,
+      "loss": 0.2487,
+      "step": 117390
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.875887059269842e-06,
+      "loss": 0.2371,
+      "step": 117395
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.874363636539808e-06,
+      "loss": 0.2452,
+      "step": 117400
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.872840488112231e-06,
+      "loss": 0.2494,
+      "step": 117405
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.871317614006903e-06,
+      "loss": 0.2555,
+      "step": 117410
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.869795014243591e-06,
+      "loss": 0.2455,
+      "step": 117415
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.868272688842062e-06,
+      "loss": 0.2585,
+      "step": 117420
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.866750637822089e-06,
+      "loss": 0.2314,
+      "step": 117425
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.865228861203435e-06,
+      "loss": 0.2348,
+      "step": 117430
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.863707359005852e-06,
+      "loss": 0.2387,
+      "step": 117435
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.862186131249096e-06,
+      "loss": 0.2661,
+      "step": 117440
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.860665177952927e-06,
+      "loss": 0.2644,
+      "step": 117445
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.859144499137085e-06,
+      "loss": 0.2466,
+      "step": 117450
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.857624094821324e-06,
+      "loss": 0.2712,
+      "step": 117455
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.856103965025376e-06,
+      "loss": 0.2314,
+      "step": 117460
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.854584109768991e-06,
+      "loss": 0.2546,
+      "step": 117465
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.853064529071895e-06,
+      "loss": 0.2517,
+      "step": 117470
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.851545222953812e-06,
+      "loss": 0.2374,
+      "step": 117475
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.850026191434483e-06,
+      "loss": 0.2398,
+      "step": 117480
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.848507434533627e-06,
+      "loss": 0.2402,
+      "step": 117485
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.846988952270964e-06,
+      "loss": 0.2551,
+      "step": 117490
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.8454707446661995e-06,
+      "loss": 0.2419,
+      "step": 117495
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.8439528117390625e-06,
+      "loss": 0.2405,
+      "step": 117500
+    },
+    {
+      "epoch": 4.13,
+      "eval_loss": 0.2497539073228836,
+      "eval_runtime": 10.5514,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 117500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.842435153509252e-06,
+      "loss": 0.2553,
+      "step": 117505
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.840917769996483e-06,
+      "loss": 0.231,
+      "step": 117510
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.839400661220446e-06,
+      "loss": 0.2546,
+      "step": 117515
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.837883827200853e-06,
+      "loss": 0.2302,
+      "step": 117520
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.836367267957391e-06,
+      "loss": 0.2676,
+      "step": 117525
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.834850983509752e-06,
+      "loss": 0.2365,
+      "step": 117530
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.833334973877623e-06,
+      "loss": 0.26,
+      "step": 117535
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.831819239080692e-06,
+      "loss": 0.2557,
+      "step": 117540
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.830303779138639e-06,
+      "loss": 0.2491,
+      "step": 117545
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.828788594071137e-06,
+      "loss": 0.2355,
+      "step": 117550
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.827273683897861e-06,
+      "loss": 0.2416,
+      "step": 117555
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.825759048638486e-06,
+      "loss": 0.2717,
+      "step": 117560
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.824244688312673e-06,
+      "loss": 0.2813,
+      "step": 117565
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.8227306029400834e-06,
+      "loss": 0.2594,
+      "step": 117570
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.821216792540388e-06,
+      "loss": 0.2486,
+      "step": 117575
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.819703257133236e-06,
+      "loss": 0.251,
+      "step": 117580
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.818189996738278e-06,
+      "loss": 0.2698,
+      "step": 117585
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.8166770113751585e-06,
+      "loss": 0.2715,
+      "step": 117590
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.815164301063531e-06,
+      "loss": 0.2503,
+      "step": 117595
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.8136518658230367e-06,
+      "loss": 0.2364,
+      "step": 117600
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.8121397056733093e-06,
+      "loss": 0.2451,
+      "step": 117605
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.81062782063398e-06,
+      "loss": 0.2568,
+      "step": 117610
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.809116210724689e-06,
+      "loss": 0.2539,
+      "step": 117615
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.807604875965051e-06,
+      "loss": 0.2518,
+      "step": 117620
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.806093816374706e-06,
+      "loss": 0.2304,
+      "step": 117625
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.8045830319732577e-06,
+      "loss": 0.2379,
+      "step": 117630
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.8030725227803377e-06,
+      "loss": 0.2314,
+      "step": 117635
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.8015622888155523e-06,
+      "loss": 0.2765,
+      "step": 117640
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.800052330098508e-06,
+      "loss": 0.2572,
+      "step": 117645
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.798542646648809e-06,
+      "loss": 0.2456,
+      "step": 117650
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7970332384860642e-06,
+      "loss": 0.2484,
+      "step": 117655
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.795524105629872e-06,
+      "loss": 0.2325,
+      "step": 117660
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.794015248099825e-06,
+      "loss": 0.249,
+      "step": 117665
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.79250666591551e-06,
+      "loss": 0.2444,
+      "step": 117670
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7909983590965176e-06,
+      "loss": 0.2437,
+      "step": 117675
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.78949032766244e-06,
+      "loss": 0.2645,
+      "step": 117680
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7879825716328475e-06,
+      "loss": 0.2362,
+      "step": 117685
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.786475091027328e-06,
+      "loss": 0.2244,
+      "step": 117690
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.784967885865451e-06,
+      "loss": 0.2653,
+      "step": 117695
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7834609561667817e-06,
+      "loss": 0.2634,
+      "step": 117700
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7819543019508852e-06,
+      "loss": 0.2359,
+      "step": 117705
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7804479232373345e-06,
+      "loss": 0.2654,
+      "step": 117710
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.778941820045681e-06,
+      "loss": 0.2537,
+      "step": 117715
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7774359923954866e-06,
+      "loss": 0.2491,
+      "step": 117720
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7759304403062914e-06,
+      "loss": 0.2456,
+      "step": 117725
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.774425163797654e-06,
+      "loss": 0.2637,
+      "step": 117730
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7729201628891207e-06,
+      "loss": 0.2687,
+      "step": 117735
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7714154376002313e-06,
+      "loss": 0.2358,
+      "step": 117740
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7699109879505173e-06,
+      "loss": 0.268,
+      "step": 117745
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7684068139595243e-06,
+      "loss": 0.2769,
+      "step": 117750
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.766902915646775e-06,
+      "loss": 0.2521,
+      "step": 117755
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7653992930317987e-06,
+      "loss": 0.2455,
+      "step": 117760
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7638959461341133e-06,
+      "loss": 0.2522,
+      "step": 117765
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.76239287497325e-06,
+      "loss": 0.2544,
+      "step": 117770
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7608900795687186e-06,
+      "loss": 0.2571,
+      "step": 117775
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.759387559940025e-06,
+      "loss": 0.2311,
+      "step": 117780
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7578853161066878e-06,
+      "loss": 0.2418,
+      "step": 117785
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.756383348088216e-06,
+      "loss": 0.2437,
+      "step": 117790
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.754881655904105e-06,
+      "loss": 0.2508,
+      "step": 117795
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7533802395738503e-06,
+      "loss": 0.2473,
+      "step": 117800
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7518790991169563e-06,
+      "loss": 0.2565,
+      "step": 117805
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.7503782345529094e-06,
+      "loss": 0.2237,
+      "step": 117810
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7488776459011972e-06,
+      "loss": 0.2534,
+      "step": 117815
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7473773331812982e-06,
+      "loss": 0.249,
+      "step": 117820
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.745877296412706e-06,
+      "loss": 0.2373,
+      "step": 117825
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7443775356148875e-06,
+      "loss": 0.2657,
+      "step": 117830
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7428780508073134e-06,
+      "loss": 0.287,
+      "step": 117835
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.741378842009466e-06,
+      "loss": 0.2629,
+      "step": 117840
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7398799092407956e-06,
+      "loss": 0.2411,
+      "step": 117845
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.738381252520781e-06,
+      "loss": 0.2443,
+      "step": 117850
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.736882871868877e-06,
+      "loss": 0.2538,
+      "step": 117855
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7353847673045255e-06,
+      "loss": 0.2542,
+      "step": 117860
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7338869388471975e-06,
+      "loss": 0.2587,
+      "step": 117865
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.732389386516333e-06,
+      "loss": 0.2654,
+      "step": 117870
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7308921103313742e-06,
+      "loss": 0.253,
+      "step": 117875
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.729395110311759e-06,
+      "loss": 0.2443,
+      "step": 117880
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7278983864769386e-06,
+      "loss": 0.2693,
+      "step": 117885
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.72640193884633e-06,
+      "loss": 0.2623,
+      "step": 117890
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7249057674393797e-06,
+      "loss": 0.251,
+      "step": 117895
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7234098722755017e-06,
+      "loss": 0.2495,
+      "step": 117900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7219142533741288e-06,
+      "loss": 0.2352,
+      "step": 117905
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7204189107546777e-06,
+      "loss": 0.2399,
+      "step": 117910
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.718923844436559e-06,
+      "loss": 0.2328,
+      "step": 117915
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.717429054439195e-06,
+      "loss": 0.247,
+      "step": 117920
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7159345407819874e-06,
+      "loss": 0.2389,
+      "step": 117925
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.714440303484343e-06,
+      "loss": 0.2324,
+      "step": 117930
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.712946342565657e-06,
+      "loss": 0.2551,
+      "step": 117935
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7114526580453416e-06,
+      "loss": 0.245,
+      "step": 117940
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7099592499427764e-06,
+      "loss": 0.2531,
+      "step": 117945
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.708466118277368e-06,
+      "loss": 0.2384,
+      "step": 117950
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.706973263068489e-06,
+      "loss": 0.2494,
+      "step": 117955
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.7054806843355328e-06,
+      "loss": 0.2568,
+      "step": 117960
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.703988382097878e-06,
+      "loss": 0.2533,
+      "step": 117965
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.702496356374899e-06,
+      "loss": 0.2504,
+      "step": 117970
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.701004607185965e-06,
+      "loss": 0.2526,
+      "step": 117975
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.699513134550453e-06,
+      "loss": 0.2547,
+      "step": 117980
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.698021938487728e-06,
+      "loss": 0.2365,
+      "step": 117985
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.696531019017149e-06,
+      "loss": 0.2438,
+      "step": 117990
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6950403761580683e-06,
+      "loss": 0.2765,
+      "step": 117995
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6935500099298524e-06,
+      "loss": 0.2527,
+      "step": 118000
+    },
+    {
+      "epoch": 4.15,
+      "eval_loss": 0.24940206110477448,
+      "eval_runtime": 10.5669,
+      "eval_samples_per_second": 9.464,
+      "eval_steps_per_second": 9.464,
+      "step": 118000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6920599203518453e-06,
+      "loss": 0.2677,
+      "step": 118005
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.690570107443403e-06,
+      "loss": 0.2606,
+      "step": 118010
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6890805712238576e-06,
+      "loss": 0.2524,
+      "step": 118015
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6875913117125626e-06,
+      "loss": 0.2526,
+      "step": 118020
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6861023289288476e-06,
+      "loss": 0.2534,
+      "step": 118025
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6846136228920497e-06,
+      "loss": 0.2687,
+      "step": 118030
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6831251936214894e-06,
+      "loss": 0.2521,
+      "step": 118035
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.681637041136507e-06,
+      "loss": 0.2366,
+      "step": 118040
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6801491654564174e-06,
+      "loss": 0.2429,
+      "step": 118045
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.678661566600533e-06,
+      "loss": 0.2454,
+      "step": 118050
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.677174244588186e-06,
+      "loss": 0.2515,
+      "step": 118055
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.675687199438671e-06,
+      "loss": 0.2501,
+      "step": 118060
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6742004311713123e-06,
+      "loss": 0.2654,
+      "step": 118065
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6727139398054e-06,
+      "loss": 0.2568,
+      "step": 118070
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.671227725360249e-06,
+      "loss": 0.2438,
+      "step": 118075
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6697417878551465e-06,
+      "loss": 0.2377,
+      "step": 118080
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6682561273093914e-06,
+      "loss": 0.2439,
+      "step": 118085
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6667707437422677e-06,
+      "loss": 0.2687,
+      "step": 118090
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.6652856371730715e-06,
+      "loss": 0.2454,
+      "step": 118095
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.663800807621082e-06,
+      "loss": 0.2303,
+      "step": 118100
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6623162551055774e-06,
+      "loss": 0.2616,
+      "step": 118105
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.660831979645829e-06,
+      "loss": 0.2408,
+      "step": 118110
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6593479812611133e-06,
+      "loss": 0.2452,
+      "step": 118115
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.657864259970706e-06,
+      "loss": 0.2708,
+      "step": 118120
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6563808157938666e-06,
+      "loss": 0.2449,
+      "step": 118125
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6548976487498498e-06,
+      "loss": 0.2404,
+      "step": 118130
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6534147588579258e-06,
+      "loss": 0.257,
+      "step": 118135
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.651932146137346e-06,
+      "loss": 0.2464,
+      "step": 118140
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6504498106073564e-06,
+      "loss": 0.2475,
+      "step": 118145
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6489677522871997e-06,
+      "loss": 0.259,
+      "step": 118150
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6474859711961324e-06,
+      "loss": 0.2459,
+      "step": 118155
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6460044673533865e-06,
+      "loss": 0.2547,
+      "step": 118160
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.644523240778197e-06,
+      "loss": 0.2544,
+      "step": 118165
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6430422914897984e-06,
+      "loss": 0.2793,
+      "step": 118170
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6415616195074246e-06,
+      "loss": 0.2845,
+      "step": 118175
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6400812248502976e-06,
+      "loss": 0.2606,
+      "step": 118180
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.638601107537637e-06,
+      "loss": 0.2469,
+      "step": 118185
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6371212675886668e-06,
+      "loss": 0.2576,
+      "step": 118190
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6356417050225967e-06,
+      "loss": 0.2511,
+      "step": 118195
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6341624198586416e-06,
+      "loss": 0.264,
+      "step": 118200
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6326834121160002e-06,
+      "loss": 0.2621,
+      "step": 118205
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6312046818138877e-06,
+      "loss": 0.2457,
+      "step": 118210
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6297262289715028e-06,
+      "loss": 0.2514,
+      "step": 118215
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.62824805360803e-06,
+      "loss": 0.269,
+      "step": 118220
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.626770155742676e-06,
+      "loss": 0.2475,
+      "step": 118225
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6252925353946284e-06,
+      "loss": 0.2417,
+      "step": 118230
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6238151925830716e-06,
+      "loss": 0.2555,
+      "step": 118235
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6223381273271876e-06,
+      "loss": 0.2482,
+      "step": 118240
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6208613396461476e-06,
+      "loss": 0.2454,
+      "step": 118245
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6193848295591416e-06,
+      "loss": 0.2589,
+      "step": 118250
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6179085970853317e-06,
+      "loss": 0.2389,
+      "step": 118255
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6164326422438886e-06,
+      "loss": 0.2575,
+      "step": 118260
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.614956965053967e-06,
+      "loss": 0.2208,
+      "step": 118265
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.613481565534743e-06,
+      "loss": 0.2806,
+      "step": 118270
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6120064437053625e-06,
+      "loss": 0.2387,
+      "step": 118275
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6105315995849876e-06,
+      "loss": 0.248,
+      "step": 118280
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6090570331927586e-06,
+      "loss": 0.2652,
+      "step": 118285
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.60758274454783e-06,
+      "loss": 0.2443,
+      "step": 118290
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6061087336693444e-06,
+      "loss": 0.2452,
+      "step": 118295
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6046350005764314e-06,
+      "loss": 0.2472,
+      "step": 118300
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6031615452882388e-06,
+      "loss": 0.2673,
+      "step": 118305
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.601688367823891e-06,
+      "loss": 0.2678,
+      "step": 118310
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.6002154682025167e-06,
+      "loss": 0.2513,
+      "step": 118315
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.59874284644324e-06,
+      "loss": 0.2497,
+      "step": 118320
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5972705025651863e-06,
+      "loss": 0.2727,
+      "step": 118325
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5957984365874637e-06,
+      "loss": 0.2371,
+      "step": 118330
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5943266485292005e-06,
+      "loss": 0.2343,
+      "step": 118335
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5928551384094926e-06,
+      "loss": 0.2563,
+      "step": 118340
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5913839062474584e-06,
+      "loss": 0.2377,
+      "step": 118345
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.589912952062194e-06,
+      "loss": 0.2378,
+      "step": 118350
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5884422758728026e-06,
+      "loss": 0.28,
+      "step": 118355
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5869718776983694e-06,
+      "loss": 0.2559,
+      "step": 118360
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5855017575580017e-06,
+      "loss": 0.252,
+      "step": 118365
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5840319154707815e-06,
+      "loss": 0.2521,
+      "step": 118370
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.582562351455793e-06,
+      "loss": 0.2526,
+      "step": 118375
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.5810930655321125e-06,
+      "loss": 0.2539,
+      "step": 118380
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5796240577188196e-06,
+      "loss": 0.2334,
+      "step": 118385
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5781553280350016e-06,
+      "loss": 0.272,
+      "step": 118390
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5766868764997103e-06,
+      "loss": 0.2575,
+      "step": 118395
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.57521870313203e-06,
+      "loss": 0.2603,
+      "step": 118400
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5737508079510125e-06,
+      "loss": 0.2526,
+      "step": 118405
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5722831909757225e-06,
+      "loss": 0.267,
+      "step": 118410
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5708158522252067e-06,
+      "loss": 0.237,
+      "step": 118415
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5693487917185325e-06,
+      "loss": 0.2494,
+      "step": 118420
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.567882009474738e-06,
+      "loss": 0.2606,
+      "step": 118425
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.566415505512874e-06,
+      "loss": 0.24,
+      "step": 118430
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.564949279851973e-06,
+      "loss": 0.2432,
+      "step": 118435
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5634833325110835e-06,
+      "loss": 0.2706,
+      "step": 118440
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5620176635092317e-06,
+      "loss": 0.2564,
+      "step": 118445
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5605522728654584e-06,
+      "loss": 0.2298,
+      "step": 118450
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5590871605987756e-06,
+      "loss": 0.2499,
+      "step": 118455
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5576223267282242e-06,
+      "loss": 0.2324,
+      "step": 118460
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5561577712728168e-06,
+      "loss": 0.2419,
+      "step": 118465
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.554693494251568e-06,
+      "loss": 0.2381,
+      "step": 118470
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5532294956834854e-06,
+      "loss": 0.2537,
+      "step": 118475
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.55176577558759e-06,
+      "loss": 0.2592,
+      "step": 118480
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.55030233398288e-06,
+      "loss": 0.245,
+      "step": 118485
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.548839170888352e-06,
+      "loss": 0.2571,
+      "step": 118490
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5473762863230176e-06,
+      "loss": 0.2119,
+      "step": 118495
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5459136803058597e-06,
+      "loss": 0.2487,
+      "step": 118500
+    },
+    {
+      "epoch": 4.17,
+      "eval_loss": 0.2493087649345398,
+      "eval_runtime": 10.5519,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 118500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5444513528558766e-06,
+      "loss": 0.2779,
+      "step": 118505
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.54298930399205e-06,
+      "loss": 0.2735,
+      "step": 118510
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5415275337333714e-06,
+      "loss": 0.2558,
+      "step": 118515
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5400660420988136e-06,
+      "loss": 0.2297,
+      "step": 118520
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.538604829107359e-06,
+      "loss": 0.2475,
+      "step": 118525
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5371438947779694e-06,
+      "loss": 0.2564,
+      "step": 118530
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.535683239129628e-06,
+      "loss": 0.2665,
+      "step": 118535
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.534222862181297e-06,
+      "loss": 0.2458,
+      "step": 118540
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.532762763951933e-06,
+      "loss": 0.2273,
+      "step": 118545
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.531302944460491e-06,
+      "loss": 0.2514,
+      "step": 118550
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5298434037259327e-06,
+      "loss": 0.2274,
+      "step": 118555
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.528384141767213e-06,
+      "loss": 0.2526,
+      "step": 118560
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.526925158603275e-06,
+      "loss": 0.2429,
+      "step": 118565
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5254664542530596e-06,
+      "loss": 0.2546,
+      "step": 118570
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.524008028735512e-06,
+      "loss": 0.2668,
+      "step": 118575
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5225498820695703e-06,
+      "loss": 0.2632,
+      "step": 118580
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5210920142741605e-06,
+      "loss": 0.2648,
+      "step": 118585
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5196344253682124e-06,
+      "loss": 0.2559,
+      "step": 118590
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5181771153706605e-06,
+      "loss": 0.2444,
+      "step": 118595
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5167200843004233e-06,
+      "loss": 0.2436,
+      "step": 118600
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5152633321764107e-06,
+      "loss": 0.259,
+      "step": 118605
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5138068590175435e-06,
+      "loss": 0.2611,
+      "step": 118610
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5123506648427425e-06,
+      "loss": 0.2542,
+      "step": 118615
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5108947496709065e-06,
+      "loss": 0.242,
+      "step": 118620
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.509439113520932e-06,
+      "loss": 0.2552,
+      "step": 118625
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.507983756411737e-06,
+      "loss": 0.2484,
+      "step": 118630
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.506528678362206e-06,
+      "loss": 0.2656,
+      "step": 118635
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.505073879391238e-06,
+      "loss": 0.2556,
+      "step": 118640
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.5036193595177123e-06,
+      "loss": 0.2626,
+      "step": 118645
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.502165118760531e-06,
+      "loss": 0.2427,
+      "step": 118650
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.500711157138567e-06,
+      "loss": 0.2403,
+      "step": 118655
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.499257474670692e-06,
+      "loss": 0.2379,
+      "step": 118660
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.4978040713757902e-06,
+      "loss": 0.2363,
+      "step": 118665
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4963509472727392e-06,
+      "loss": 0.2586,
+      "step": 118670
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4948981023804007e-06,
+      "loss": 0.2377,
+      "step": 118675
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.493445536717635e-06,
+      "loss": 0.2561,
+      "step": 118680
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4919932503033022e-06,
+      "loss": 0.2447,
+      "step": 118685
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4905412431562675e-06,
+      "loss": 0.2475,
+      "step": 118690
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4890895152953794e-06,
+      "loss": 0.258,
+      "step": 118695
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4876380667394874e-06,
+      "loss": 0.2361,
+      "step": 118700
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.486186897507432e-06,
+      "loss": 0.2671,
+      "step": 118705
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.484736007618067e-06,
+      "loss": 0.2538,
+      "step": 118710
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4832853970902218e-06,
+      "loss": 0.2454,
+      "step": 118715
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.48183506594274e-06,
+      "loss": 0.2589,
+      "step": 118720
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4803850141944426e-06,
+      "loss": 0.2587,
+      "step": 118725
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4789352418641728e-06,
+      "loss": 0.2427,
+      "step": 118730
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4774857489707435e-06,
+      "loss": 0.2502,
+      "step": 118735
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4760365355329725e-06,
+      "loss": 0.2455,
+      "step": 118740
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4745876015696872e-06,
+      "loss": 0.2785,
+      "step": 118745
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.473138947099697e-06,
+      "loss": 0.2643,
+      "step": 118750
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4716905721418114e-06,
+      "loss": 0.2686,
+      "step": 118755
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.47024247671483e-06,
+      "loss": 0.255,
+      "step": 118760
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4687946608375655e-06,
+      "loss": 0.2389,
+      "step": 118765
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4673471245288083e-06,
+      "loss": 0.2675,
+      "step": 118770
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.465899867807365e-06,
+      "loss": 0.2451,
+      "step": 118775
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4644528906920132e-06,
+      "loss": 0.233,
+      "step": 118780
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.463006193201554e-06,
+      "loss": 0.2499,
+      "step": 118785
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.461559775354767e-06,
+      "loss": 0.2304,
+      "step": 118790
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4601136371704314e-06,
+      "loss": 0.2727,
+      "step": 118795
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.458667778667318e-06,
+      "loss": 0.26,
+      "step": 118800
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4572221998642128e-06,
+      "loss": 0.2551,
+      "step": 118805
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4557769007798807e-06,
+      "loss": 0.2581,
+      "step": 118810
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.45433188143309e-06,
+      "loss": 0.2619,
+      "step": 118815
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4528871418425922e-06,
+      "loss": 0.2278,
+      "step": 118820
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.451442682027156e-06,
+      "loss": 0.2565,
+      "step": 118825
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4499985020055386e-06,
+      "loss": 0.2589,
+      "step": 118830
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4485546017964916e-06,
+      "loss": 0.2446,
+      "step": 118835
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4471109814187552e-06,
+      "loss": 0.2836,
+      "step": 118840
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4456676408910814e-06,
+      "loss": 0.2464,
+      "step": 118845
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.444224580232211e-06,
+      "loss": 0.2539,
+      "step": 118850
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4427817994608702e-06,
+      "loss": 0.2403,
+      "step": 118855
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.441339298595808e-06,
+      "loss": 0.2511,
+      "step": 118860
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.439897077655749e-06,
+      "loss": 0.2489,
+      "step": 118865
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.438455136659416e-06,
+      "loss": 0.2584,
+      "step": 118870
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.437013475625528e-06,
+      "loss": 0.2549,
+      "step": 118875
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.435572094572817e-06,
+      "loss": 0.2346,
+      "step": 118880
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.434130993519982e-06,
+      "loss": 0.269,
+      "step": 118885
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.43269017248575e-06,
+      "loss": 0.2394,
+      "step": 118890
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4312496314888198e-06,
+      "loss": 0.2548,
+      "step": 118895
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4298093705479012e-06,
+      "loss": 0.2422,
+      "step": 118900
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.428369389681696e-06,
+      "loss": 0.24,
+      "step": 118905
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4269296889088946e-06,
+      "loss": 0.2584,
+      "step": 118910
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.425490268248191e-06,
+      "loss": 0.2567,
+      "step": 118915
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4240511277182808e-06,
+      "loss": 0.2364,
+      "step": 118920
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.422612267337852e-06,
+      "loss": 0.25,
+      "step": 118925
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.421173687125573e-06,
+      "loss": 0.2475,
+      "step": 118930
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.41973538710014e-06,
+      "loss": 0.27,
+      "step": 118935
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4182973672802153e-06,
+      "loss": 0.2454,
+      "step": 118940
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.4168596276844823e-06,
+      "loss": 0.2377,
+      "step": 118945
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.415709637781853e-06,
+      "loss": 0.241,
+      "step": 118950
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.4142724026366953e-06,
+      "loss": 0.2411,
+      "step": 118955
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.412835447767987e-06,
+      "loss": 0.2634,
+      "step": 118960
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.4113987731943743e-06,
+      "loss": 0.2555,
+      "step": 118965
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.409962378934525e-06,
+      "loss": 0.2322,
+      "step": 118970
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.408526265007089e-06,
+      "loss": 0.2418,
+      "step": 118975
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.407090431430712e-06,
+      "loss": 0.2518,
+      "step": 118980
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.4056548782240265e-06,
+      "loss": 0.2366,
+      "step": 118985
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.4042196054056896e-06,
+      "loss": 0.2806,
+      "step": 118990
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.402784612994328e-06,
+      "loss": 0.2627,
+      "step": 118995
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.401349901008577e-06,
+      "loss": 0.2632,
+      "step": 119000
+    },
+    {
+      "epoch": 4.19,
+      "eval_loss": 0.2492852360010147,
+      "eval_runtime": 10.5487,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 119000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3999154694670548e-06,
+      "loss": 0.254,
+      "step": 119005
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3984813183884022e-06,
+      "loss": 0.2624,
+      "step": 119010
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.397047447791235e-06,
+      "loss": 0.2382,
+      "step": 119015
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3956138576941633e-06,
+      "loss": 0.258,
+      "step": 119020
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3941805481158133e-06,
+      "loss": 0.2312,
+      "step": 119025
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.392747519074785e-06,
+      "loss": 0.245,
+      "step": 119030
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.391314770589696e-06,
+      "loss": 0.2552,
+      "step": 119035
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.389882302679137e-06,
+      "loss": 0.2511,
+      "step": 119040
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3884501153617214e-06,
+      "loss": 0.2263,
+      "step": 119045
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3870182086560396e-06,
+      "loss": 0.268,
+      "step": 119050
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3855865825806797e-06,
+      "loss": 0.2514,
+      "step": 119055
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3841552371542297e-06,
+      "loss": 0.2475,
+      "step": 119060
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.382724172395285e-06,
+      "loss": 0.2654,
+      "step": 119065
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.381293388322421e-06,
+      "loss": 0.2376,
+      "step": 119070
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3798628849542072e-06,
+      "loss": 0.2617,
+      "step": 119075
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3784326623092333e-06,
+      "loss": 0.2533,
+      "step": 119080
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3770027204060555e-06,
+      "loss": 0.2464,
+      "step": 119085
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.375573059263254e-06,
+      "loss": 0.2422,
+      "step": 119090
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.374143678899383e-06,
+      "loss": 0.2526,
+      "step": 119095
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.372714579333003e-06,
+      "loss": 0.2593,
+      "step": 119100
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3712857605826738e-06,
+      "loss": 0.255,
+      "step": 119105
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3698572226669446e-06,
+      "loss": 0.2451,
+      "step": 119110
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3684289656043645e-06,
+      "loss": 0.2499,
+      "step": 119115
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3670009894134743e-06,
+      "loss": 0.2308,
+      "step": 119120
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3655732941128255e-06,
+      "loss": 0.2485,
+      "step": 119125
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.364145879720948e-06,
+      "loss": 0.2657,
+      "step": 119130
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.362718746256374e-06,
+      "loss": 0.2777,
+      "step": 119135
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3612918937376366e-06,
+      "loss": 0.2472,
+      "step": 119140
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.35986532218327e-06,
+      "loss": 0.2415,
+      "step": 119145
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3584390316117904e-06,
+      "loss": 0.2415,
+      "step": 119150
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3570130220417107e-06,
+      "loss": 0.2461,
+      "step": 119155
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3555872934915604e-06,
+      "loss": 0.2749,
+      "step": 119160
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3541618459798447e-06,
+      "loss": 0.2483,
+      "step": 119165
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.352736679525073e-06,
+      "loss": 0.2565,
+      "step": 119170
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3513117941457424e-06,
+      "loss": 0.2477,
+      "step": 119175
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3498871898603677e-06,
+      "loss": 0.2491,
+      "step": 119180
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3484628666874405e-06,
+      "loss": 0.254,
+      "step": 119185
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.347038824645446e-06,
+      "loss": 0.2663,
+      "step": 119190
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3456150637528856e-06,
+      "loss": 0.2688,
+      "step": 119195
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3441915840282473e-06,
+      "loss": 0.274,
+      "step": 119200
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.342768385490008e-06,
+      "loss": 0.2508,
+      "step": 119205
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.341345468156651e-06,
+      "loss": 0.2563,
+      "step": 119210
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.339922832046641e-06,
+      "loss": 0.2641,
+      "step": 119215
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3385004771784663e-06,
+      "loss": 0.263,
+      "step": 119220
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.337078403570587e-06,
+      "loss": 0.2462,
+      "step": 119225
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.3356566112414695e-06,
+      "loss": 0.2361,
+      "step": 119230
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.334235100209565e-06,
+      "loss": 0.2614,
+      "step": 119235
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.332813870493348e-06,
+      "loss": 0.2713,
+      "step": 119240
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3313929221112564e-06,
+      "loss": 0.2877,
+      "step": 119245
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3299722550817536e-06,
+      "loss": 0.2484,
+      "step": 119250
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.328551869423274e-06,
+      "loss": 0.2591,
+      "step": 119255
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3271317651542733e-06,
+      "loss": 0.2664,
+      "step": 119260
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.325711942293183e-06,
+      "loss": 0.2699,
+      "step": 119265
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.324292400858439e-06,
+      "loss": 0.2607,
+      "step": 119270
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3228731408684654e-06,
+      "loss": 0.2423,
+      "step": 119275
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.321454162341708e-06,
+      "loss": 0.2607,
+      "step": 119280
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.320035465296578e-06,
+      "loss": 0.2525,
+      "step": 119285
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3186170497514963e-06,
+      "loss": 0.2772,
+      "step": 119290
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3171989157248897e-06,
+      "loss": 0.2663,
+      "step": 119295
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3157810632351597e-06,
+      "loss": 0.2536,
+      "step": 119300
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.314363492300726e-06,
+      "loss": 0.2519,
+      "step": 119305
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.312946202939987e-06,
+      "loss": 0.2738,
+      "step": 119310
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.311529195171356e-06,
+      "loss": 0.2578,
+      "step": 119315
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.310112469013224e-06,
+      "loss": 0.2636,
+      "step": 119320
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.308696024483987e-06,
+      "loss": 0.2374,
+      "step": 119325
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3072798616020306e-06,
+      "loss": 0.2472,
+      "step": 119330
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3058639803857568e-06,
+      "loss": 0.2488,
+      "step": 119335
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.304448380853542e-06,
+      "loss": 0.2489,
+      "step": 119340
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3030330630237666e-06,
+      "loss": 0.2465,
+      "step": 119345
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3016180269148018e-06,
+      "loss": 0.2632,
+      "step": 119350
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.3002032725450274e-06,
+      "loss": 0.2499,
+      "step": 119355
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2987887999328173e-06,
+      "loss": 0.2446,
+      "step": 119360
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2973746090965347e-06,
+      "loss": 0.2313,
+      "step": 119365
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2959607000545346e-06,
+      "loss": 0.2592,
+      "step": 119370
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2945470728251855e-06,
+      "loss": 0.2602,
+      "step": 119375
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2931337274268393e-06,
+      "loss": 0.2268,
+      "step": 119380
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2917206638778453e-06,
+      "loss": 0.2442,
+      "step": 119385
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.29030788219655e-06,
+      "loss": 0.2654,
+      "step": 119390
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2888953824013026e-06,
+      "loss": 0.2611,
+      "step": 119395
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2874831645104383e-06,
+      "loss": 0.2603,
+      "step": 119400
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.286071228542295e-06,
+      "loss": 0.2546,
+      "step": 119405
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2846595745152033e-06,
+      "loss": 0.2498,
+      "step": 119410
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.283248202447503e-06,
+      "loss": 0.277,
+      "step": 119415
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2818371123575136e-06,
+      "loss": 0.2537,
+      "step": 119420
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.280426304263548e-06,
+      "loss": 0.2423,
+      "step": 119425
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2790157781839414e-06,
+      "loss": 0.2577,
+      "step": 119430
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2776055341370014e-06,
+      "loss": 0.2648,
+      "step": 119435
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2761955721410357e-06,
+      "loss": 0.225,
+      "step": 119440
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2747858922143493e-06,
+      "loss": 0.2659,
+      "step": 119445
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.273376494375255e-06,
+      "loss": 0.28,
+      "step": 119450
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.271967378642049e-06,
+      "loss": 0.2796,
+      "step": 119455
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.270558545033023e-06,
+      "loss": 0.2658,
+      "step": 119460
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2691499935664784e-06,
+      "loss": 0.2435,
+      "step": 119465
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.267741724260695e-06,
+      "loss": 0.2611,
+      "step": 119470
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.266333737133967e-06,
+      "loss": 0.2511,
+      "step": 119475
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.264926032204574e-06,
+      "loss": 0.2607,
+      "step": 119480
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2635186094907872e-06,
+      "loss": 0.2444,
+      "step": 119485
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.262111469010892e-06,
+      "loss": 0.2625,
+      "step": 119490
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2607046107831522e-06,
+      "loss": 0.2604,
+      "step": 119495
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.259298034825836e-06,
+      "loss": 0.2672,
+      "step": 119500
+    },
+    {
+      "epoch": 4.2,
+      "eval_loss": 0.24933363497257233,
+      "eval_runtime": 10.5337,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 119500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.257891741157204e-06,
+      "loss": 0.2517,
+      "step": 119505
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2564857297955247e-06,
+      "loss": 0.265,
+      "step": 119510
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.255080000759042e-06,
+      "loss": 0.2615,
+      "step": 119515
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.253674554066019e-06,
+      "loss": 0.2522,
+      "step": 119520
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.252269389734697e-06,
+      "loss": 0.2523,
+      "step": 119525
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.25086450778333e-06,
+      "loss": 0.2653,
+      "step": 119530
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.249459908230154e-06,
+      "loss": 0.2465,
+      "step": 119535
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2480555910934017e-06,
+      "loss": 0.2678,
+      "step": 119540
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2466515563913167e-06,
+      "loss": 0.2462,
+      "step": 119545
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2452478041421263e-06,
+      "loss": 0.2608,
+      "step": 119550
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.243844334364057e-06,
+      "loss": 0.247,
+      "step": 119555
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.242441147075323e-06,
+      "loss": 0.2504,
+      "step": 119560
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.241038242294159e-06,
+      "loss": 0.2384,
+      "step": 119565
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.239635620038775e-06,
+      "loss": 0.2452,
+      "step": 119570
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.238233280327374e-06,
+      "loss": 0.2408,
+      "step": 119575
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2368312231781746e-06,
+      "loss": 0.2511,
+      "step": 119580
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.235429448609384e-06,
+      "loss": 0.2409,
+      "step": 119585
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.234027956639199e-06,
+      "loss": 0.2468,
+      "step": 119590
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2326267472858164e-06,
+      "loss": 0.2512,
+      "step": 119595
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.231225820567424e-06,
+      "loss": 0.2381,
+      "step": 119600
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2298251765022243e-06,
+      "loss": 0.2532,
+      "step": 119605
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2284248151083967e-06,
+      "loss": 0.2436,
+      "step": 119610
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2270247364041246e-06,
+      "loss": 0.2452,
+      "step": 119615
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2256249404075846e-06,
+      "loss": 0.2474,
+      "step": 119620
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2242254271369593e-06,
+      "loss": 0.2411,
+      "step": 119625
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2228261966104097e-06,
+      "loss": 0.2612,
+      "step": 119630
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2214272488461096e-06,
+      "loss": 0.2593,
+      "step": 119635
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2200285838622305e-06,
+      "loss": 0.2482,
+      "step": 119640
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2186302016769256e-06,
+      "loss": 0.2464,
+      "step": 119645
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.217232102308354e-06,
+      "loss": 0.253,
+      "step": 119650
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2158342857746632e-06,
+      "loss": 0.2417,
+      "step": 119655
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2144367520940135e-06,
+      "loss": 0.2467,
+      "step": 119660
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2130395012845455e-06,
+      "loss": 0.2488,
+      "step": 119665
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2116425333644007e-06,
+      "loss": 0.2447,
+      "step": 119670
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.210245848351712e-06,
+      "loss": 0.2906,
+      "step": 119675
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.208849446264628e-06,
+      "loss": 0.2576,
+      "step": 119680
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.207453327121268e-06,
+      "loss": 0.2508,
+      "step": 119685
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2060574909397677e-06,
+      "loss": 0.2648,
+      "step": 119690
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.204661937738246e-06,
+      "loss": 0.2435,
+      "step": 119695
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2032666675348266e-06,
+      "loss": 0.2678,
+      "step": 119700
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.201871680347629e-06,
+      "loss": 0.2452,
+      "step": 119705
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.2004769761947604e-06,
+      "loss": 0.2479,
+      "step": 119710
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1990825550943264e-06,
+      "loss": 0.2468,
+      "step": 119715
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.197688417064443e-06,
+      "loss": 0.2438,
+      "step": 119720
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1962945621232087e-06,
+      "loss": 0.229,
+      "step": 119725
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1949009902887185e-06,
+      "loss": 0.248,
+      "step": 119730
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1935077015790655e-06,
+      "loss": 0.2514,
+      "step": 119735
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1921146960123437e-06,
+      "loss": 0.2503,
+      "step": 119740
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1907219736066448e-06,
+      "loss": 0.242,
+      "step": 119745
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1893295343800424e-06,
+      "loss": 0.2572,
+      "step": 119750
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.187937378350628e-06,
+      "loss": 0.2471,
+      "step": 119755
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.186545505536473e-06,
+      "loss": 0.2571,
+      "step": 119760
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.185153915955649e-06,
+      "loss": 0.262,
+      "step": 119765
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1837626096262175e-06,
+      "loss": 0.2488,
+      "step": 119770
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.18237158656626e-06,
+      "loss": 0.2683,
+      "step": 119775
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.180980846793827e-06,
+      "loss": 0.2407,
+      "step": 119780
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.179590390326978e-06,
+      "loss": 0.2662,
+      "step": 119785
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.178200217183763e-06,
+      "loss": 0.2616,
+      "step": 119790
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1768103273822377e-06,
+      "loss": 0.2381,
+      "step": 119795
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.17542072094045e-06,
+      "loss": 0.2518,
+      "step": 119800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.174031397876445e-06,
+      "loss": 0.238,
+      "step": 119805
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1726423582082524e-06,
+      "loss": 0.2465,
+      "step": 119810
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.171253601953919e-06,
+      "loss": 0.2735,
+      "step": 119815
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1698651291314717e-06,
+      "loss": 0.2575,
+      "step": 119820
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.168476939758941e-06,
+      "loss": 0.2336,
+      "step": 119825
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1670890338543425e-06,
+      "loss": 0.2427,
+      "step": 119830
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1657014114357093e-06,
+      "loss": 0.2528,
+      "step": 119835
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1643140725210546e-06,
+      "loss": 0.241,
+      "step": 119840
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1629270171283864e-06,
+      "loss": 0.2535,
+      "step": 119845
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.161540245275721e-06,
+      "loss": 0.2415,
+      "step": 119850
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1601537569810655e-06,
+      "loss": 0.2592,
+      "step": 119855
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1587675522624233e-06,
+      "loss": 0.2557,
+      "step": 119860
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1573816311377847e-06,
+      "loss": 0.2379,
+      "step": 119865
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.155995993625155e-06,
+      "loss": 0.2406,
+      "step": 119870
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.154610639742522e-06,
+      "loss": 0.2616,
+      "step": 119875
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1532255695078725e-06,
+      "loss": 0.2408,
+      "step": 119880
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.151840782939186e-06,
+      "loss": 0.2526,
+      "step": 119885
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.150456280054456e-06,
+      "loss": 0.2352,
+      "step": 119890
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1490720608716493e-06,
+      "loss": 0.2672,
+      "step": 119895
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1476881254087374e-06,
+      "loss": 0.2511,
+      "step": 119900
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1463044736837007e-06,
+      "loss": 0.2399,
+      "step": 119905
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1449211057144905e-06,
+      "loss": 0.248,
+      "step": 119910
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1435380215190853e-06,
+      "loss": 0.2494,
+      "step": 119915
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.142155221115431e-06,
+      "loss": 0.2433,
+      "step": 119920
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.140772704521483e-06,
+      "loss": 0.2539,
+      "step": 119925
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.139390471755202e-06,
+      "loss": 0.251,
+      "step": 119930
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.138008522834526e-06,
+      "loss": 0.2298,
+      "step": 119935
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1366268577774057e-06,
+      "loss": 0.2414,
+      "step": 119940
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1352454766017698e-06,
+      "loss": 0.2607,
+      "step": 119945
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.133864379325566e-06,
+      "loss": 0.232,
+      "step": 119950
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.132483565966718e-06,
+      "loss": 0.2561,
+      "step": 119955
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1311030365431675e-06,
+      "loss": 0.2499,
+      "step": 119960
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.129722791072823e-06,
+      "loss": 0.2691,
+      "step": 119965
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1283428295736243e-06,
+      "loss": 0.2471,
+      "step": 119970
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.126963152063478e-06,
+      "loss": 0.2565,
+      "step": 119975
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1255837585602936e-06,
+      "loss": 0.2482,
+      "step": 119980
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1242046490819935e-06,
+      "loss": 0.2418,
+      "step": 119985
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.122825823646483e-06,
+      "loss": 0.2433,
+      "step": 119990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1214472822716584e-06,
+      "loss": 0.2518,
+      "step": 119995
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1200690249754193e-06,
+      "loss": 0.2385,
+      "step": 120000
+    },
+    {
+      "epoch": 4.22,
+      "eval_loss": 0.2491784542798996,
+      "eval_runtime": 10.5428,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 120000
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1186910517756684e-06,
+      "loss": 0.2656,
+      "step": 120005
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.117313362690294e-06,
+      "loss": 0.2626,
+      "step": 120010
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1159359577371787e-06,
+      "loss": 0.2739,
+      "step": 120015
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1145588369342117e-06,
+      "loss": 0.2564,
+      "step": 120020
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1131820002992835e-06,
+      "loss": 0.2396,
+      "step": 120025
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1118054478502605e-06,
+      "loss": 0.2415,
+      "step": 120030
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1104291796050204e-06,
+      "loss": 0.2329,
+      "step": 120035
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.109053195581424e-06,
+      "loss": 0.2654,
+      "step": 120040
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1076774957973537e-06,
+      "loss": 0.2385,
+      "step": 120045
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1063020802706623e-06,
+      "loss": 0.2725,
+      "step": 120050
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1049269490192102e-06,
+      "loss": 0.2649,
+      "step": 120055
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1035521020608444e-06,
+      "loss": 0.2447,
+      "step": 120060
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1021775394134314e-06,
+      "loss": 0.2564,
+      "step": 120065
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.100803261094809e-06,
+      "loss": 0.252,
+      "step": 120070
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.0994292671228247e-06,
+      "loss": 0.2462,
+      "step": 120075
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.098055557515317e-06,
+      "loss": 0.2526,
+      "step": 120080
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.096682132290127e-06,
+      "loss": 0.2599,
+      "step": 120085
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0953089914650875e-06,
+      "loss": 0.2409,
+      "step": 120090
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.093936135058015e-06,
+      "loss": 0.2726,
+      "step": 120095
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0925635630867534e-06,
+      "loss": 0.2622,
+      "step": 120100
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0911912755691163e-06,
+      "loss": 0.237,
+      "step": 120105
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.089819272522923e-06,
+      "loss": 0.2613,
+      "step": 120110
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0884475539659813e-06,
+      "loss": 0.2473,
+      "step": 120115
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.08707611991611e-06,
+      "loss": 0.2497,
+      "step": 120120
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0857049703911124e-06,
+      "loss": 0.2457,
+      "step": 120125
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0843341054087982e-06,
+      "loss": 0.2464,
+      "step": 120130
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0829635249869566e-06,
+      "loss": 0.2472,
+      "step": 120135
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0815932291433952e-06,
+      "loss": 0.256,
+      "step": 120140
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.080223217895903e-06,
+      "loss": 0.2195,
+      "step": 120145
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0788534912622654e-06,
+      "loss": 0.2621,
+      "step": 120150
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0774840492602625e-06,
+      "loss": 0.2544,
+      "step": 120155
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.076114891907686e-06,
+      "loss": 0.2692,
+      "step": 120160
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0747460192223103e-06,
+      "loss": 0.2575,
+      "step": 120165
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0733774312219104e-06,
+      "loss": 0.265,
+      "step": 120170
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.072009127924247e-06,
+      "loss": 0.2286,
+      "step": 120175
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0706411093470944e-06,
+      "loss": 0.2429,
+      "step": 120180
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0692733755082194e-06,
+      "loss": 0.2679,
+      "step": 120185
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0679059264253795e-06,
+      "loss": 0.2519,
+      "step": 120190
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0665387621163192e-06,
+      "loss": 0.2533,
+      "step": 120195
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.065171882598805e-06,
+      "loss": 0.259,
+      "step": 120200
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.063805287890578e-06,
+      "loss": 0.2365,
+      "step": 120205
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.062438978009377e-06,
+      "loss": 0.2513,
+      "step": 120210
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.061072952972954e-06,
+      "loss": 0.2644,
+      "step": 120215
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.05970721279904e-06,
+      "loss": 0.2378,
+      "step": 120220
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.05834175750537e-06,
+      "loss": 0.2433,
+      "step": 120225
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0569765871096666e-06,
+      "loss": 0.2603,
+      "step": 120230
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0556117016296597e-06,
+      "loss": 0.2841,
+      "step": 120235
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.05424710108308e-06,
+      "loss": 0.2475,
+      "step": 120240
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0528827854876374e-06,
+      "loss": 0.2428,
+      "step": 120245
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0515187548610437e-06,
+      "loss": 0.2155,
+      "step": 120250
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0501550092210203e-06,
+      "loss": 0.2604,
+      "step": 120255
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.048791548585267e-06,
+      "loss": 0.2346,
+      "step": 120260
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.047428372971492e-06,
+      "loss": 0.2381,
+      "step": 120265
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0460654823973837e-06,
+      "loss": 0.2488,
+      "step": 120270
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0447028768806535e-06,
+      "loss": 0.2498,
+      "step": 120275
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0433405564389866e-06,
+      "loss": 0.2293,
+      "step": 120280
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.041978521090069e-06,
+      "loss": 0.251,
+      "step": 120285
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.040616770851587e-06,
+      "loss": 0.2576,
+      "step": 120290
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0392553057412316e-06,
+      "loss": 0.2502,
+      "step": 120295
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0378941257766745e-06,
+      "loss": 0.2585,
+      "step": 120300
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0365332309755856e-06,
+      "loss": 0.2348,
+      "step": 120305
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0351726213556364e-06,
+      "loss": 0.2497,
+      "step": 120310
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.033812296934499e-06,
+      "loss": 0.2333,
+      "step": 120315
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0324522577298312e-06,
+      "loss": 0.2393,
+      "step": 120320
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.031092503759289e-06,
+      "loss": 0.2626,
+      "step": 120325
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0297330350405382e-06,
+      "loss": 0.226,
+      "step": 120330
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0283738515912262e-06,
+      "loss": 0.2447,
+      "step": 120335
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.027014953428994e-06,
+      "loss": 0.25,
+      "step": 120340
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.025656340571495e-06,
+      "loss": 0.2542,
+      "step": 120345
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.024298013036364e-06,
+      "loss": 0.2484,
+      "step": 120350
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.022939970841246e-06,
+      "loss": 0.2477,
+      "step": 120355
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.021582214003768e-06,
+      "loss": 0.2303,
+      "step": 120360
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0202247425415553e-06,
+      "loss": 0.2471,
+      "step": 120365
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0188675564722463e-06,
+      "loss": 0.2564,
+      "step": 120370
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0175106558134546e-06,
+      "loss": 0.2469,
+      "step": 120375
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0161540405828e-06,
+      "loss": 0.257,
+      "step": 120380
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0147977107978926e-06,
+      "loss": 0.2453,
+      "step": 120385
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0134416664763522e-06,
+      "loss": 0.2795,
+      "step": 120390
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0120859076357755e-06,
+      "loss": 0.2609,
+      "step": 120395
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.010730434293782e-06,
+      "loss": 0.2433,
+      "step": 120400
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.009375246467955e-06,
+      "loss": 0.26,
+      "step": 120405
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.008020344175902e-06,
+      "loss": 0.2411,
+      "step": 120410
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.006665727435212e-06,
+      "loss": 0.2562,
+      "step": 120415
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0053113962634744e-06,
+      "loss": 0.2415,
+      "step": 120420
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.003957350678269e-06,
+      "loss": 0.2475,
+      "step": 120425
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.002603590697184e-06,
+      "loss": 0.256,
+      "step": 120430
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0012501163377974e-06,
+      "loss": 0.2479,
+      "step": 120435
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.999896927617679e-06,
+      "loss": 0.2468,
+      "step": 120440
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.998544024554395e-06,
+      "loss": 0.242,
+      "step": 120445
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9971914071655227e-06,
+      "loss": 0.2531,
+      "step": 120450
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9958390754686148e-06,
+      "loss": 0.2293,
+      "step": 120455
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9944870294812326e-06,
+      "loss": 0.2333,
+      "step": 120460
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9931352692209423e-06,
+      "loss": 0.2495,
+      "step": 120465
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9917837947052885e-06,
+      "loss": 0.2532,
+      "step": 120470
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.990432605951815e-06,
+      "loss": 0.2596,
+      "step": 120475
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9890817029780666e-06,
+      "loss": 0.2376,
+      "step": 120480
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.98773108580159e-06,
+      "loss": 0.2323,
+      "step": 120485
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9863807544399186e-06,
+      "loss": 0.2409,
+      "step": 120490
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9850307089105856e-06,
+      "loss": 0.24,
+      "step": 120495
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9836809492311156e-06,
+      "loss": 0.2306,
+      "step": 120500
+    },
+    {
+      "epoch": 4.24,
+      "eval_loss": 0.24912191927433014,
+      "eval_runtime": 10.5509,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 120500
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.982331475419045e-06,
+      "loss": 0.2357,
+      "step": 120505
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.980982287491885e-06,
+      "loss": 0.2507,
+      "step": 120510
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.979633385467165e-06,
+      "loss": 0.2242,
+      "step": 120515
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9782847693623857e-06,
+      "loss": 0.2441,
+      "step": 120520
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9769364391950717e-06,
+      "loss": 0.2661,
+      "step": 120525
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9755883949827255e-06,
+      "loss": 0.2413,
+      "step": 120530
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9742406367428477e-06,
+      "loss": 0.25,
+      "step": 120535
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9728931644929348e-06,
+      "loss": 0.2511,
+      "step": 120540
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.971545978250495e-06,
+      "loss": 0.2275,
+      "step": 120545
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.970199078033012e-06,
+      "loss": 0.2579,
+      "step": 120550
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9688524638579712e-06,
+      "loss": 0.232,
+      "step": 120555
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9675061357428647e-06,
+      "loss": 0.2694,
+      "step": 120560
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.96616009370517e-06,
+      "loss": 0.2594,
+      "step": 120565
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.964814337762367e-06,
+      "loss": 0.2464,
+      "step": 120570
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9634688679319263e-06,
+      "loss": 0.2456,
+      "step": 120575
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9621236842313217e-06,
+      "loss": 0.2684,
+      "step": 120580
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9607787866780202e-06,
+      "loss": 0.2601,
+      "step": 120585
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.95943417528948e-06,
+      "loss": 0.2679,
+      "step": 120590
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9580898500831565e-06,
+      "loss": 0.2576,
+      "step": 120595
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9567458110765166e-06,
+      "loss": 0.2452,
+      "step": 120600
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9554020582870046e-06,
+      "loss": 0.2327,
+      "step": 120605
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9540585917320673e-06,
+      "loss": 0.2696,
+      "step": 120610
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9527154114291495e-06,
+      "loss": 0.2686,
+      "step": 120615
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9513725173956897e-06,
+      "loss": 0.2216,
+      "step": 120620
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9500299096491323e-06,
+      "loss": 0.2457,
+      "step": 120625
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9486875882069027e-06,
+      "loss": 0.2603,
+      "step": 120630
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.947345553086431e-06,
+      "loss": 0.2277,
+      "step": 120635
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9460038043051475e-06,
+      "loss": 0.265,
+      "step": 120640
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.944662341880472e-06,
+      "loss": 0.2402,
+      "step": 120645
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.943321165829818e-06,
+      "loss": 0.2363,
+      "step": 120650
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.9419802761706e-06,
+      "loss": 0.2539,
+      "step": 120655
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9406396729202367e-06,
+      "loss": 0.248,
+      "step": 120660
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9392993560961287e-06,
+      "loss": 0.2392,
+      "step": 120665
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9379593257156753e-06,
+      "loss": 0.2712,
+      "step": 120670
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9366195817962793e-06,
+      "loss": 0.2678,
+      "step": 120675
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.935280124355344e-06,
+      "loss": 0.2323,
+      "step": 120680
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.933940953410255e-06,
+      "loss": 0.2584,
+      "step": 120685
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9326020689783963e-06,
+      "loss": 0.247,
+      "step": 120690
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9312634710771613e-06,
+      "loss": 0.2425,
+      "step": 120695
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9299251597239256e-06,
+      "loss": 0.2444,
+      "step": 120700
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.928587134936067e-06,
+      "loss": 0.2666,
+      "step": 120705
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.927249396730955e-06,
+      "loss": 0.2563,
+      "step": 120710
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.925911945125967e-06,
+      "loss": 0.2358,
+      "step": 120715
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9245747801384644e-06,
+      "loss": 0.2709,
+      "step": 120720
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.923237901785808e-06,
+      "loss": 0.2481,
+      "step": 120725
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9219013100853536e-06,
+      "loss": 0.2693,
+      "step": 120730
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.920565005054471e-06,
+      "loss": 0.2394,
+      "step": 120735
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9192289867104983e-06,
+      "loss": 0.2328,
+      "step": 120740
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9178932550707834e-06,
+      "loss": 0.2582,
+      "step": 120745
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9165578101526703e-06,
+      "loss": 0.2498,
+      "step": 120750
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9152226519735036e-06,
+      "loss": 0.2736,
+      "step": 120755
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.913887780550617e-06,
+      "loss": 0.2409,
+      "step": 120760
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9125531959013434e-06,
+      "loss": 0.243,
+      "step": 120765
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.911218898043003e-06,
+      "loss": 0.2517,
+      "step": 120770
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9098848869929334e-06,
+      "loss": 0.2564,
+      "step": 120775
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.908551162768447e-06,
+      "loss": 0.2724,
+      "step": 120780
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.907217725386871e-06,
+      "loss": 0.2543,
+      "step": 120785
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9058845748655056e-06,
+      "loss": 0.2508,
+      "step": 120790
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9045517112216757e-06,
+      "loss": 0.2684,
+      "step": 120795
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.903219134472679e-06,
+      "loss": 0.2584,
+      "step": 120800
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.901886844635812e-06,
+      "loss": 0.2492,
+      "step": 120805
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9005548417283897e-06,
+      "loss": 0.2458,
+      "step": 120810
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8992231257676974e-06,
+      "loss": 0.2482,
+      "step": 120815
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8978916967710297e-06,
+      "loss": 0.2641,
+      "step": 120820
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8965605547556646e-06,
+      "loss": 0.2626,
+      "step": 120825
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.895229699738902e-06,
+      "loss": 0.26,
+      "step": 120830
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8938991317380065e-06,
+      "loss": 0.2568,
+      "step": 120835
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8925688507702692e-06,
+      "loss": 0.2586,
+      "step": 120840
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8912388568529485e-06,
+      "loss": 0.2451,
+      "step": 120845
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.889909150003328e-06,
+      "loss": 0.2543,
+      "step": 120850
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8885797302386665e-06,
+      "loss": 0.2514,
+      "step": 120855
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8872505975762244e-06,
+      "loss": 0.2537,
+      "step": 120860
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.885921752033255e-06,
+      "loss": 0.248,
+      "step": 120865
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8845931936270247e-06,
+      "loss": 0.2462,
+      "step": 120870
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.883264922374776e-06,
+      "loss": 0.2396,
+      "step": 120875
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8819369382937555e-06,
+      "loss": 0.2467,
+      "step": 120880
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.880609241401205e-06,
+      "loss": 0.2532,
+      "step": 120885
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8792818317143698e-06,
+      "loss": 0.253,
+      "step": 120890
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.877954709250477e-06,
+      "loss": 0.2502,
+      "step": 120895
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.876627874026769e-06,
+      "loss": 0.2448,
+      "step": 120900
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.875301326060462e-06,
+      "loss": 0.2594,
+      "step": 120905
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.873975065368792e-06,
+      "loss": 0.2519,
+      "step": 120910
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.872649091968976e-06,
+      "loss": 0.2296,
+      "step": 120915
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8713234058782207e-06,
+      "loss": 0.2255,
+      "step": 120920
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8699980071137554e-06,
+      "loss": 0.26,
+      "step": 120925
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8686728956927813e-06,
+      "loss": 0.2547,
+      "step": 120930
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.867348071632503e-06,
+      "loss": 0.2509,
+      "step": 120935
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8660235349501213e-06,
+      "loss": 0.2581,
+      "step": 120940
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.864699285662839e-06,
+      "loss": 0.2587,
+      "step": 120945
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.863375323787848e-06,
+      "loss": 0.2342,
+      "step": 120950
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8620516493423427e-06,
+      "loss": 0.2549,
+      "step": 120955
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8607282623435013e-06,
+      "loss": 0.2453,
+      "step": 120960
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.859405162808518e-06,
+      "loss": 0.2539,
+      "step": 120965
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.858082350754568e-06,
+      "loss": 0.2283,
+      "step": 120970
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8567598261988267e-06,
+      "loss": 0.2295,
+      "step": 120975
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8554375891584634e-06,
+      "loss": 0.2649,
+      "step": 120980
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.854115639650651e-06,
+      "loss": 0.2505,
+      "step": 120985
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.852793977692553e-06,
+      "loss": 0.2589,
+      "step": 120990
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.851472603301325e-06,
+      "loss": 0.2311,
+      "step": 120995
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8501515164941346e-06,
+      "loss": 0.2492,
+      "step": 121000
+    },
+    {
+      "epoch": 4.26,
+      "eval_loss": 0.2490115761756897,
+      "eval_runtime": 10.5538,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 9.475,
+      "step": 121000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.848830717288123e-06,
+      "loss": 0.247,
+      "step": 121005
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8475102057004542e-06,
+      "loss": 0.2259,
+      "step": 121010
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8461899817482625e-06,
+      "loss": 0.2403,
+      "step": 121015
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8448700454486917e-06,
+      "loss": 0.2519,
+      "step": 121020
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8435503968188863e-06,
+      "loss": 0.2418,
+      "step": 121025
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.842231035875978e-06,
+      "loss": 0.2397,
+      "step": 121030
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8409119626370913e-06,
+      "loss": 0.2589,
+      "step": 121035
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.839593177119365e-06,
+      "loss": 0.2376,
+      "step": 121040
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8382746793399166e-06,
+      "loss": 0.259,
+      "step": 121045
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8369564693158684e-06,
+      "loss": 0.2618,
+      "step": 121050
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.835638547064329e-06,
+      "loss": 0.2712,
+      "step": 121055
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8343209126024147e-06,
+      "loss": 0.2358,
+      "step": 121060
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.83300356594724e-06,
+      "loss": 0.2489,
+      "step": 121065
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8316865071159078e-06,
+      "loss": 0.2154,
+      "step": 121070
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.83036973612551e-06,
+      "loss": 0.2405,
+      "step": 121075
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.829053252993155e-06,
+      "loss": 0.2571,
+      "step": 121080
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8277370577359354e-06,
+      "loss": 0.2402,
+      "step": 121085
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8264211503709335e-06,
+      "loss": 0.2652,
+      "step": 121090
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8251055309152396e-06,
+      "loss": 0.2483,
+      "step": 121095
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8237901993859396e-06,
+      "loss": 0.2549,
+      "step": 121100
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8224751558001084e-06,
+      "loss": 0.2425,
+      "step": 121105
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.821160400174819e-06,
+      "loss": 0.2424,
+      "step": 121110
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.819845932527143e-06,
+      "loss": 0.2504,
+      "step": 121115
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.818531752874157e-06,
+      "loss": 0.2361,
+      "step": 121120
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.817217861232918e-06,
+      "loss": 0.2291,
+      "step": 121125
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.815904257620483e-06,
+      "loss": 0.2484,
+      "step": 121130
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8145909420539103e-06,
+      "loss": 0.2387,
+      "step": 121135
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.813277914550258e-06,
+      "loss": 0.2293,
+      "step": 121140
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8119651751265687e-06,
+      "loss": 0.2186,
+      "step": 121145
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8106527237998835e-06,
+      "loss": 0.2495,
+      "step": 121150
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8093405605872565e-06,
+      "loss": 0.2678,
+      "step": 121155
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8080286855057174e-06,
+      "loss": 0.2364,
+      "step": 121160
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.806717098572295e-06,
+      "loss": 0.2825,
+      "step": 121165
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8054057998040255e-06,
+      "loss": 0.2593,
+      "step": 121170
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8040947892179424e-06,
+      "loss": 0.2576,
+      "step": 121175
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.802784066831057e-06,
+      "loss": 0.2451,
+      "step": 121180
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.801473632660395e-06,
+      "loss": 0.2642,
+      "step": 121185
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.800163486722962e-06,
+      "loss": 0.2485,
+      "step": 121190
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.79885362903578e-06,
+      "loss": 0.2436,
+      "step": 121195
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.7975440596158533e-06,
+      "loss": 0.2518,
+      "step": 121200
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.796234778480186e-06,
+      "loss": 0.2564,
+      "step": 121205
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.7949257856457715e-06,
+      "loss": 0.281,
+      "step": 121210
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.793617081129618e-06,
+      "loss": 0.2407,
+      "step": 121215
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.792308664948706e-06,
+      "loss": 0.2698,
+      "step": 121220
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7910005371200337e-06,
+      "loss": 0.2251,
+      "step": 121225
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7896926976605813e-06,
+      "loss": 0.2399,
+      "step": 121230
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7883851465873356e-06,
+      "loss": 0.2554,
+      "step": 121235
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7870778839172718e-06,
+      "loss": 0.2535,
+      "step": 121240
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7857709096673623e-06,
+      "loss": 0.261,
+      "step": 121245
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7844642238545717e-06,
+      "loss": 0.2599,
+      "step": 121250
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.78315782649588e-06,
+      "loss": 0.2503,
+      "step": 121255
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7818517176082388e-06,
+      "loss": 0.2507,
+      "step": 121260
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7805458972086084e-06,
+      "loss": 0.2684,
+      "step": 121265
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.779240365313951e-06,
+      "loss": 0.246,
+      "step": 121270
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.777935121941208e-06,
+      "loss": 0.2402,
+      "step": 121275
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7766301671073386e-06,
+      "loss": 0.2546,
+      "step": 121280
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7753255008292763e-06,
+      "loss": 0.2616,
+      "step": 121285
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.774021123123968e-06,
+      "loss": 0.2429,
+      "step": 121290
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7727170340083487e-06,
+      "loss": 0.2257,
+      "step": 121295
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.771413233499351e-06,
+      "loss": 0.2453,
+      "step": 121300
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7701097216139006e-06,
+      "loss": 0.2324,
+      "step": 121305
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7688064983689253e-06,
+      "loss": 0.247,
+      "step": 121310
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.767503563781351e-06,
+      "loss": 0.2538,
+      "step": 121315
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.766200917868089e-06,
+      "loss": 0.2301,
+      "step": 121320
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.764898560646051e-06,
+      "loss": 0.2277,
+      "step": 121325
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.763596492132159e-06,
+      "loss": 0.2596,
+      "step": 121330
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.762294712343305e-06,
+      "loss": 0.2412,
+      "step": 121335
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.760993221296404e-06,
+      "loss": 0.2399,
+      "step": 121340
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7596920190083445e-06,
+      "loss": 0.2316,
+      "step": 121345
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.758391105496033e-06,
+      "loss": 0.2384,
+      "step": 121350
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.757090480776353e-06,
+      "loss": 0.2528,
+      "step": 121355
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7557901448661964e-06,
+      "loss": 0.2692,
+      "step": 121360
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7544900977824416e-06,
+      "loss": 0.251,
+      "step": 121365
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.753190339541978e-06,
+      "loss": 0.2458,
+      "step": 121370
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.751890870161675e-06,
+      "loss": 0.2533,
+      "step": 121375
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7505916896583993e-06,
+      "loss": 0.2551,
+      "step": 121380
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7492927980490356e-06,
+      "loss": 0.2596,
+      "step": 121385
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7479941953504366e-06,
+      "loss": 0.2748,
+      "step": 121390
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7466958815794723e-06,
+      "loss": 0.2447,
+      "step": 121395
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.745397856752993e-06,
+      "loss": 0.2598,
+      "step": 121400
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7441001208878575e-06,
+      "loss": 0.2256,
+      "step": 121405
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.742802674000919e-06,
+      "loss": 0.2501,
+      "step": 121410
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7415055161090166e-06,
+      "loss": 0.2437,
+      "step": 121415
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.740208647228992e-06,
+      "loss": 0.247,
+      "step": 121420
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.738912067377694e-06,
+      "loss": 0.2348,
+      "step": 121425
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.737615776571953e-06,
+      "loss": 0.2418,
+      "step": 121430
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7363197748285942e-06,
+      "loss": 0.2489,
+      "step": 121435
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.735024062164454e-06,
+      "loss": 0.2592,
+      "step": 121440
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7337286385963503e-06,
+      "loss": 0.2512,
+      "step": 121445
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7324335041411105e-06,
+      "loss": 0.2565,
+      "step": 121450
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7311386588155464e-06,
+      "loss": 0.2642,
+      "step": 121455
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7298441026364667e-06,
+      "loss": 0.2381,
+      "step": 121460
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.728549835620689e-06,
+      "loss": 0.2478,
+      "step": 121465
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.727255857785016e-06,
+      "loss": 0.255,
+      "step": 121470
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7259621691462454e-06,
+      "loss": 0.2477,
+      "step": 121475
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7246687697211728e-06,
+      "loss": 0.2686,
+      "step": 121480
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7233756595266003e-06,
+      "loss": 0.2574,
+      "step": 121485
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.722082838579315e-06,
+      "loss": 0.2669,
+      "step": 121490
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7207903068960982e-06,
+      "loss": 0.2532,
+      "step": 121495
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.7194980644937335e-06,
+      "loss": 0.2605,
+      "step": 121500
+    },
+    {
+      "epoch": 4.27,
+      "eval_loss": 0.2489774525165558,
+      "eval_runtime": 10.5356,
+      "eval_samples_per_second": 9.492,
+      "eval_steps_per_second": 9.492,
+      "step": 121500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.71820611138901e-06,
+      "loss": 0.2496,
+      "step": 121505
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.716914447598698e-06,
+      "loss": 0.2578,
+      "step": 121510
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.715623073139559e-06,
+      "loss": 0.2563,
+      "step": 121515
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.7143319880283735e-06,
+      "loss": 0.2564,
+      "step": 121520
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.713041192281901e-06,
+      "loss": 0.2646,
+      "step": 121525
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.7117506859169028e-06,
+      "loss": 0.254,
+      "step": 121530
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.710460468950127e-06,
+      "loss": 0.2599,
+      "step": 121535
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.709170541398337e-06,
+      "loss": 0.2422,
+      "step": 121540
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.707880903278276e-06,
+      "loss": 0.2631,
+      "step": 121545
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.7065915546066883e-06,
+      "loss": 0.2415,
+      "step": 121550
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.7053024954003163e-06,
+      "loss": 0.2481,
+      "step": 121555
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.7040137256759022e-06,
+      "loss": 0.2766,
+      "step": 121560
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.7027252454501772e-06,
+      "loss": 0.2354,
+      "step": 121565
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.701437054739872e-06,
+      "loss": 0.2468,
+      "step": 121570
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.700149153561704e-06,
+      "loss": 0.2653,
+      "step": 121575
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.69886154193241e-06,
+      "loss": 0.2803,
+      "step": 121580
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.697574219868701e-06,
+      "loss": 0.256,
+      "step": 121585
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.696287187387292e-06,
+      "loss": 0.2505,
+      "step": 121590
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6950004445048917e-06,
+      "loss": 0.2512,
+      "step": 121595
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6937139912382142e-06,
+      "loss": 0.255,
+      "step": 121600
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.692427827603958e-06,
+      "loss": 0.2581,
+      "step": 121605
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6911419536188228e-06,
+      "loss": 0.2419,
+      "step": 121610
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.689856369299512e-06,
+      "loss": 0.2618,
+      "step": 121615
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.688571074662716e-06,
+      "loss": 0.2373,
+      "step": 121620
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.68728606972512e-06,
+      "loss": 0.2525,
+      "step": 121625
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6860013545034034e-06,
+      "loss": 0.2449,
+      "step": 121630
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6847169290142583e-06,
+      "loss": 0.2477,
+      "step": 121635
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6834327932743598e-06,
+      "loss": 0.2501,
+      "step": 121640
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.682148947300378e-06,
+      "loss": 0.2325,
+      "step": 121645
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6808653911089775e-06,
+      "loss": 0.2497,
+      "step": 121650
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6795821247168394e-06,
+      "loss": 0.2622,
+      "step": 121655
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.678299148140609e-06,
+      "loss": 0.2621,
+      "step": 121660
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.677016461396961e-06,
+      "loss": 0.2589,
+      "step": 121665
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6757340645025387e-06,
+      "loss": 0.2481,
+      "step": 121670
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.674451957474e-06,
+      "loss": 0.2638,
+      "step": 121675
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.673170140327991e-06,
+      "loss": 0.268,
+      "step": 121680
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.671888613081153e-06,
+      "loss": 0.2838,
+      "step": 121685
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6706073757501227e-06,
+      "loss": 0.2576,
+      "step": 121690
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.669326428351543e-06,
+      "loss": 0.2445,
+      "step": 121695
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6680457709020447e-06,
+      "loss": 0.2423,
+      "step": 121700
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.666765403418256e-06,
+      "loss": 0.2709,
+      "step": 121705
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6654853259167943e-06,
+      "loss": 0.2508,
+      "step": 121710
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6642055384142857e-06,
+      "loss": 0.2431,
+      "step": 121715
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.662926040927352e-06,
+      "loss": 0.2804,
+      "step": 121720
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.661646833472606e-06,
+      "loss": 0.2608,
+      "step": 121725
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.660367916066647e-06,
+      "loss": 0.2494,
+      "step": 121730
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6590892887260963e-06,
+      "loss": 0.232,
+      "step": 121735
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.657810951467546e-06,
+      "loss": 0.2577,
+      "step": 121740
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6565329043075905e-06,
+      "loss": 0.2576,
+      "step": 121745
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6552551472628363e-06,
+      "loss": 0.253,
+      "step": 121750
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.65397768034987e-06,
+      "loss": 0.2486,
+      "step": 121755
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.652700503585276e-06,
+      "loss": 0.2248,
+      "step": 121760
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.651423616985632e-06,
+      "loss": 0.2486,
+      "step": 121765
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6501470205675283e-06,
+      "loss": 0.2499,
+      "step": 121770
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6488707143475343e-06,
+      "loss": 0.2508,
+      "step": 121775
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.647594698342229e-06,
+      "loss": 0.2659,
+      "step": 121780
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.6463189725681683e-06,
+      "loss": 0.2229,
+      "step": 121785
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.64504353704193e-06,
+      "loss": 0.2552,
+      "step": 121790
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.643768391780069e-06,
+      "loss": 0.2331,
+      "step": 121795
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6424935367991425e-06,
+      "loss": 0.2623,
+      "step": 121800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6412189721157e-06,
+      "loss": 0.2548,
+      "step": 121805
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6399446977462968e-06,
+      "loss": 0.2427,
+      "step": 121810
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.638670713707475e-06,
+      "loss": 0.2621,
+      "step": 121815
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.637397020015775e-06,
+      "loss": 0.2476,
+      "step": 121820
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6361236166877414e-06,
+      "loss": 0.2595,
+      "step": 121825
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6348505037398995e-06,
+      "loss": 0.2507,
+      "step": 121830
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.633577681188787e-06,
+      "loss": 0.2412,
+      "step": 121835
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6323051490509314e-06,
+      "loss": 0.2364,
+      "step": 121840
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6310329073428454e-06,
+      "loss": 0.2592,
+      "step": 121845
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.629760956081062e-06,
+      "loss": 0.2489,
+      "step": 121850
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.628489295282091e-06,
+      "loss": 0.2743,
+      "step": 121855
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6272179249624355e-06,
+      "loss": 0.2614,
+      "step": 121860
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.625946845138616e-06,
+      "loss": 0.2532,
+      "step": 121865
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.624676055827133e-06,
+      "loss": 0.2724,
+      "step": 121870
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.623405557044481e-06,
+      "loss": 0.2745,
+      "step": 121875
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.622135348807164e-06,
+      "loss": 0.2399,
+      "step": 121880
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6208654311316694e-06,
+      "loss": 0.2491,
+      "step": 121885
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.619595804034494e-06,
+      "loss": 0.2312,
+      "step": 121890
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6183264675321145e-06,
+      "loss": 0.2636,
+      "step": 121895
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.617057421641014e-06,
+      "loss": 0.2386,
+      "step": 121900
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.615788666377675e-06,
+      "loss": 0.2337,
+      "step": 121905
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.614520201758569e-06,
+      "loss": 0.2488,
+      "step": 121910
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6132520278001672e-06,
+      "loss": 0.2382,
+      "step": 121915
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.611984144518928e-06,
+      "loss": 0.2425,
+      "step": 121920
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.610716551931325e-06,
+      "loss": 0.2491,
+      "step": 121925
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.609449250053814e-06,
+      "loss": 0.2611,
+      "step": 121930
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6081822389028427e-06,
+      "loss": 0.2375,
+      "step": 121935
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6069155184948677e-06,
+      "loss": 0.2616,
+      "step": 121940
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.605649088846343e-06,
+      "loss": 0.2383,
+      "step": 121945
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.604382949973705e-06,
+      "loss": 0.2471,
+      "step": 121950
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.603117101893396e-06,
+      "loss": 0.2555,
+      "step": 121955
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.601851544621847e-06,
+      "loss": 0.2565,
+      "step": 121960
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.6005862781754986e-06,
+      "loss": 0.242,
+      "step": 121965
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.599321302570776e-06,
+      "loss": 0.2707,
+      "step": 121970
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.598056617824099e-06,
+      "loss": 0.2544,
+      "step": 121975
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5967922239518966e-06,
+      "loss": 0.2463,
+      "step": 121980
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.595528120970586e-06,
+      "loss": 0.2752,
+      "step": 121985
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5942643088965712e-06,
+      "loss": 0.2545,
+      "step": 121990
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.593000787746269e-06,
+      "loss": 0.2346,
+      "step": 121995
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.591737557536089e-06,
+      "loss": 0.2512,
+      "step": 122000
+    },
+    {
+      "epoch": 4.29,
+      "eval_loss": 0.24890007078647614,
+      "eval_runtime": 10.5474,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 122000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5904746182824292e-06,
+      "loss": 0.2254,
+      "step": 122005
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.589211970001687e-06,
+      "loss": 0.2532,
+      "step": 122010
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5879496127102556e-06,
+      "loss": 0.2646,
+      "step": 122015
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.586687546424532e-06,
+      "loss": 0.248,
+      "step": 122020
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.585425771160899e-06,
+      "loss": 0.2387,
+      "step": 122025
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.584164286935742e-06,
+      "loss": 0.2512,
+      "step": 122030
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.582903093765432e-06,
+      "loss": 0.2743,
+      "step": 122035
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5816421916663585e-06,
+      "loss": 0.2504,
+      "step": 122040
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5803815806548835e-06,
+      "loss": 0.2552,
+      "step": 122045
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5791212607473804e-06,
+      "loss": 0.2443,
+      "step": 122050
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.57786123196021e-06,
+      "loss": 0.2674,
+      "step": 122055
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5766014943097385e-06,
+      "loss": 0.2529,
+      "step": 122060
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.575342047812318e-06,
+      "loss": 0.2512,
+      "step": 122065
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.574082892484303e-06,
+      "loss": 0.2352,
+      "step": 122070
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.5728240283420383e-06,
+      "loss": 0.2629,
+      "step": 122075
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.571565455401878e-06,
+      "loss": 0.2581,
+      "step": 122080
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.570307173680159e-06,
+      "loss": 0.274,
+      "step": 122085
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5690491831932147e-06,
+      "loss": 0.2403,
+      "step": 122090
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.567791483957388e-06,
+      "loss": 0.2349,
+      "step": 122095
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5665340759890024e-06,
+      "loss": 0.2499,
+      "step": 122100
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5652769593043912e-06,
+      "loss": 0.261,
+      "step": 122105
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5640201339198697e-06,
+      "loss": 0.2513,
+      "step": 122110
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5627635998517666e-06,
+      "loss": 0.2698,
+      "step": 122115
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.561507357116391e-06,
+      "loss": 0.2702,
+      "step": 122120
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.560251405730055e-06,
+      "loss": 0.242,
+      "step": 122125
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5589957457090586e-06,
+      "loss": 0.2438,
+      "step": 122130
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.55774037706972e-06,
+      "loss": 0.2389,
+      "step": 122135
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5564852998283344e-06,
+      "loss": 0.2274,
+      "step": 122140
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5552305140011944e-06,
+      "loss": 0.2233,
+      "step": 122145
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5539760196045896e-06,
+      "loss": 0.2624,
+      "step": 122150
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5527218166548154e-06,
+      "loss": 0.2347,
+      "step": 122155
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.551467905168159e-06,
+      "loss": 0.2507,
+      "step": 122160
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.550214285160896e-06,
+      "loss": 0.2562,
+      "step": 122165
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5489609566493015e-06,
+      "loss": 0.2428,
+      "step": 122170
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5477079196496584e-06,
+      "loss": 0.282,
+      "step": 122175
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5464551741782334e-06,
+      "loss": 0.2664,
+      "step": 122180
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5452027202512883e-06,
+      "loss": 0.2606,
+      "step": 122185
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.543950557885083e-06,
+      "loss": 0.2485,
+      "step": 122190
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.542698687095885e-06,
+      "loss": 0.2526,
+      "step": 122195
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.541447107899947e-06,
+      "loss": 0.2412,
+      "step": 122200
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5401958203135097e-06,
+      "loss": 0.2406,
+      "step": 122205
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5389448243528345e-06,
+      "loss": 0.2505,
+      "step": 122210
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5376941200341538e-06,
+      "loss": 0.261,
+      "step": 122215
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5364437073737146e-06,
+      "loss": 0.2548,
+      "step": 122220
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5351935863877464e-06,
+      "loss": 0.241,
+      "step": 122225
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.533943757092491e-06,
+      "loss": 0.2481,
+      "step": 122230
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.532694219504167e-06,
+      "loss": 0.2549,
+      "step": 122235
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.531444973639005e-06,
+      "loss": 0.2555,
+      "step": 122240
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5301960195132147e-06,
+      "loss": 0.2465,
+      "step": 122245
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.52894735714303e-06,
+      "loss": 0.2605,
+      "step": 122250
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5276989865446522e-06,
+      "loss": 0.2623,
+      "step": 122255
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5264509077342875e-06,
+      "loss": 0.2271,
+      "step": 122260
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5252031207281534e-06,
+      "loss": 0.2602,
+      "step": 122265
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.52395562554244e-06,
+      "loss": 0.261,
+      "step": 122270
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5227084221933567e-06,
+      "loss": 0.2442,
+      "step": 122275
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.52146151069709e-06,
+      "loss": 0.2489,
+      "step": 122280
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5202148910698276e-06,
+      "loss": 0.2632,
+      "step": 122285
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5189685633277644e-06,
+      "loss": 0.2268,
+      "step": 122290
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5177225274870792e-06,
+      "loss": 0.2557,
+      "step": 122295
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.516476783563951e-06,
+      "loss": 0.2423,
+      "step": 122300
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5152313315745496e-06,
+      "loss": 0.2396,
+      "step": 122305
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5139861715350545e-06,
+      "loss": 0.2371,
+      "step": 122310
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.512741303461627e-06,
+      "loss": 0.2349,
+      "step": 122315
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.511496727370438e-06,
+      "loss": 0.2572,
+      "step": 122320
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5102524432776386e-06,
+      "loss": 0.2773,
+      "step": 122325
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.509008451199396e-06,
+      "loss": 0.2501,
+      "step": 122330
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5077647511518557e-06,
+      "loss": 0.2553,
+      "step": 122335
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5065213431511637e-06,
+      "loss": 0.238,
+      "step": 122340
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.50527822721347e-06,
+      "loss": 0.2413,
+      "step": 122345
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.504035403354915e-06,
+      "loss": 0.2534,
+      "step": 122350
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.502792871591636e-06,
+      "loss": 0.2371,
+      "step": 122355
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.5015506319397615e-06,
+      "loss": 0.2468,
+      "step": 122360
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.5003086844154283e-06,
+      "loss": 0.2515,
+      "step": 122365
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.49906702903476e-06,
+      "loss": 0.2508,
+      "step": 122370
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4978256658138715e-06,
+      "loss": 0.2537,
+      "step": 122375
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.496584594768889e-06,
+      "loss": 0.2531,
+      "step": 122380
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4953438159159267e-06,
+      "loss": 0.244,
+      "step": 122385
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.494103329271097e-06,
+      "loss": 0.2435,
+      "step": 122390
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.492863134850501e-06,
+      "loss": 0.2725,
+      "step": 122395
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4916232326702428e-06,
+      "loss": 0.2411,
+      "step": 122400
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4903836227464258e-06,
+      "loss": 0.2417,
+      "step": 122405
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.489144305095145e-06,
+      "loss": 0.2568,
+      "step": 122410
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.487905279732489e-06,
+      "loss": 0.2252,
+      "step": 122415
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4866665466745404e-06,
+      "loss": 0.2488,
+      "step": 122420
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4854281059373984e-06,
+      "loss": 0.2501,
+      "step": 122425
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.484189957537128e-06,
+      "loss": 0.2605,
+      "step": 122430
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4829521014898103e-06,
+      "loss": 0.2593,
+      "step": 122435
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4817145378115274e-06,
+      "loss": 0.2526,
+      "step": 122440
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.480477266518341e-06,
+      "loss": 0.249,
+      "step": 122445
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4792402876263165e-06,
+      "loss": 0.2751,
+      "step": 122450
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.47800360115151e-06,
+      "loss": 0.2493,
+      "step": 122455
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4767672071099897e-06,
+      "loss": 0.2384,
+      "step": 122460
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.475531105517803e-06,
+      "loss": 0.257,
+      "step": 122465
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.474295296391002e-06,
+      "loss": 0.2513,
+      "step": 122470
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.473059779745626e-06,
+      "loss": 0.2359,
+      "step": 122475
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4718245555977266e-06,
+      "loss": 0.2462,
+      "step": 122480
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4705896239633347e-06,
+      "loss": 0.2384,
+      "step": 122485
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.469354984858496e-06,
+      "loss": 0.2534,
+      "step": 122490
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.468120638299226e-06,
+      "loss": 0.2218,
+      "step": 122495
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.466886584301567e-06,
+      "loss": 0.2444,
+      "step": 122500
+    },
+    {
+      "epoch": 4.31,
+      "eval_loss": 0.2489146739244461,
+      "eval_runtime": 10.5432,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 122500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.465652822881537e-06,
+      "loss": 0.2727,
+      "step": 122505
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.464419354055153e-06,
+      "loss": 0.262,
+      "step": 122510
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.463186177838428e-06,
+      "loss": 0.2561,
+      "step": 122515
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.461953294247382e-06,
+      "loss": 0.2547,
+      "step": 122520
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4607207032980195e-06,
+      "loss": 0.2372,
+      "step": 122525
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4594884050063465e-06,
+      "loss": 0.244,
+      "step": 122530
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4582563993883563e-06,
+      "loss": 0.265,
+      "step": 122535
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4570246864600493e-06,
+      "loss": 0.2328,
+      "step": 122540
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4557932662374273e-06,
+      "loss": 0.2509,
+      "step": 122545
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4545621387364658e-06,
+      "loss": 0.2652,
+      "step": 122550
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.453331303973164e-06,
+      "loss": 0.2345,
+      "step": 122555
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4521007619634965e-06,
+      "loss": 0.2505,
+      "step": 122560
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4508705127234384e-06,
+      "loss": 0.243,
+      "step": 122565
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4496405562689617e-06,
+      "loss": 0.2588,
+      "step": 122570
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4484108926160486e-06,
+      "loss": 0.2478,
+      "step": 122575
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.447181521780656e-06,
+      "loss": 0.2422,
+      "step": 122580
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4459524437787485e-06,
+      "loss": 0.249,
+      "step": 122585
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.44472365862628e-06,
+      "loss": 0.2487,
+      "step": 122590
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.443495166339213e-06,
+      "loss": 0.2348,
+      "step": 122595
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4422669669334986e-06,
+      "loss": 0.2635,
+      "step": 122600
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4410390604250795e-06,
+      "loss": 0.2381,
+      "step": 122605
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4398114468298985e-06,
+      "loss": 0.2357,
+      "step": 122610
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4385841261639035e-06,
+      "loss": 0.2647,
+      "step": 122615
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4373570984430242e-06,
+      "loss": 0.2329,
+      "step": 122620
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.436130363683195e-06,
+      "loss": 0.2431,
+      "step": 122625
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.434903921900336e-06,
+      "loss": 0.2418,
+      "step": 122630
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4336777731103847e-06,
+      "loss": 0.2545,
+      "step": 122635
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4324519173292565e-06,
+      "loss": 0.2493,
+      "step": 122640
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.4312263545728605e-06,
+      "loss": 0.2496,
+      "step": 122645
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4300010848571225e-06,
+      "loss": 0.2447,
+      "step": 122650
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4287761081979445e-06,
+      "loss": 0.2414,
+      "step": 122655
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4275514246112353e-06,
+      "loss": 0.2578,
+      "step": 122660
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.426327034112891e-06,
+      "loss": 0.266,
+      "step": 122665
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.425102936718818e-06,
+      "loss": 0.261,
+      "step": 122670
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.423879132444906e-06,
+      "loss": 0.2403,
+      "step": 122675
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.422655621307046e-06,
+      "loss": 0.2579,
+      "step": 122680
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4214324033211167e-06,
+      "loss": 0.2365,
+      "step": 122685
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.420209478503016e-06,
+      "loss": 0.2446,
+      "step": 122690
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4189868468686124e-06,
+      "loss": 0.2614,
+      "step": 122695
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4177645084337787e-06,
+      "loss": 0.2611,
+      "step": 122700
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.416542463214394e-06,
+      "loss": 0.2676,
+      "step": 122705
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.415320711226321e-06,
+      "loss": 0.251,
+      "step": 122710
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.414099252485427e-06,
+      "loss": 0.2726,
+      "step": 122715
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4128780870075695e-06,
+      "loss": 0.2336,
+      "step": 122720
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.411657214808602e-06,
+      "loss": 0.2599,
+      "step": 122725
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.410436635904381e-06,
+      "loss": 0.2276,
+      "step": 122730
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4092163503107556e-06,
+      "loss": 0.2533,
+      "step": 122735
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4079963580435656e-06,
+      "loss": 0.2604,
+      "step": 122740
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.406776659118651e-06,
+      "loss": 0.258,
+      "step": 122745
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4055572535518574e-06,
+      "loss": 0.2369,
+      "step": 122750
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4043381413590083e-06,
+      "loss": 0.2541,
+      "step": 122755
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.403119322555941e-06,
+      "loss": 0.2411,
+      "step": 122760
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.401900797158471e-06,
+      "loss": 0.2334,
+      "step": 122765
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.4006825651824326e-06,
+      "loss": 0.2497,
+      "step": 122770
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.399464626643638e-06,
+      "loss": 0.2425,
+      "step": 122775
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3982469815578972e-06,
+      "loss": 0.2526,
+      "step": 122780
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3970296299410277e-06,
+      "loss": 0.2486,
+      "step": 122785
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.395812571808831e-06,
+      "loss": 0.2415,
+      "step": 122790
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3945958071771134e-06,
+      "loss": 0.2286,
+      "step": 122795
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.393379336061666e-06,
+      "loss": 0.2465,
+      "step": 122800
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3921631584782942e-06,
+      "loss": 0.2524,
+      "step": 122805
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.390947274442787e-06,
+      "loss": 0.2628,
+      "step": 122810
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.389731683970922e-06,
+      "loss": 0.2426,
+      "step": 122815
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3885163870784907e-06,
+      "loss": 0.2562,
+      "step": 122820
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.387301383781279e-06,
+      "loss": 0.2338,
+      "step": 122825
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3860866740950562e-06,
+      "loss": 0.237,
+      "step": 122830
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.384872258035592e-06,
+      "loss": 0.2568,
+      "step": 122835
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.383658135618655e-06,
+      "loss": 0.2528,
+      "step": 122840
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3824443068600182e-06,
+      "loss": 0.2568,
+      "step": 122845
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3812307717754333e-06,
+      "loss": 0.2535,
+      "step": 122850
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3800175303806625e-06,
+      "loss": 0.2579,
+      "step": 122855
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3788045826914545e-06,
+      "loss": 0.2386,
+      "step": 122860
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3775919287235633e-06,
+      "loss": 0.2235,
+      "step": 122865
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3763795684927286e-06,
+      "loss": 0.2408,
+      "step": 122870
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3751675020146996e-06,
+      "loss": 0.261,
+      "step": 122875
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.373955729305205e-06,
+      "loss": 0.2728,
+      "step": 122880
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.372744250379988e-06,
+      "loss": 0.2683,
+      "step": 122885
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3715330652547774e-06,
+      "loss": 0.2324,
+      "step": 122890
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.370322173945291e-06,
+      "loss": 0.2572,
+      "step": 122895
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.369111576467262e-06,
+      "loss": 0.2803,
+      "step": 122900
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3679012728364063e-06,
+      "loss": 0.256,
+      "step": 122905
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.366691263068438e-06,
+      "loss": 0.2598,
+      "step": 122910
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3654815471790604e-06,
+      "loss": 0.2567,
+      "step": 122915
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3642721251839946e-06,
+      "loss": 0.2443,
+      "step": 122920
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.3630629970989332e-06,
+      "loss": 0.2458,
+      "step": 122925
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3618541629395833e-06,
+      "loss": 0.2581,
+      "step": 122930
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3606456227216357e-06,
+      "loss": 0.2582,
+      "step": 122935
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.359437376460788e-06,
+      "loss": 0.2519,
+      "step": 122940
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3582294241727284e-06,
+      "loss": 0.2416,
+      "step": 122945
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3570217658731352e-06,
+      "loss": 0.254,
+      "step": 122950
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3558144015776858e-06,
+      "loss": 0.2403,
+      "step": 122955
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.354607331302072e-06,
+      "loss": 0.2813,
+      "step": 122960
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3534005550619547e-06,
+      "loss": 0.2667,
+      "step": 122965
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3521940728730095e-06,
+      "loss": 0.2722,
+      "step": 122970
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.350987884750891e-06,
+      "loss": 0.263,
+      "step": 122975
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3497819907112722e-06,
+      "loss": 0.2437,
+      "step": 122980
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3485763907698076e-06,
+      "loss": 0.2371,
+      "step": 122985
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3473710849421542e-06,
+      "loss": 0.2622,
+      "step": 122990
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3461660732439524e-06,
+      "loss": 0.2443,
+      "step": 122995
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3449613556908613e-06,
+      "loss": 0.2414,
+      "step": 123000
+    },
+    {
+      "epoch": 4.33,
+      "eval_loss": 0.24904577434062958,
+      "eval_runtime": 10.5244,
+      "eval_samples_per_second": 9.502,
+      "eval_steps_per_second": 9.502,
+      "step": 123000
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.343756932298513e-06,
+      "loss": 0.2657,
+      "step": 123005
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3425528030825538e-06,
+      "loss": 0.2416,
+      "step": 123010
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.341348968058607e-06,
+      "loss": 0.2707,
+      "step": 123015
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3401454272423183e-06,
+      "loss": 0.2352,
+      "step": 123020
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.338942180649306e-06,
+      "loss": 0.2549,
+      "step": 123025
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.337739228295191e-06,
+      "loss": 0.276,
+      "step": 123030
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3365365701955993e-06,
+      "loss": 0.2473,
+      "step": 123035
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3353342063661466e-06,
+      "loss": 0.2493,
+      "step": 123040
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3341321368224422e-06,
+      "loss": 0.2441,
+      "step": 123045
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3329303615800906e-06,
+      "loss": 0.2545,
+      "step": 123050
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.331728880654707e-06,
+      "loss": 0.2505,
+      "step": 123055
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.330527694061882e-06,
+      "loss": 0.2575,
+      "step": 123060
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3293268018172165e-06,
+      "loss": 0.2356,
+      "step": 123065
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3281262039362984e-06,
+      "loss": 0.2797,
+      "step": 123070
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.326925900434723e-06,
+      "loss": 0.254,
+      "step": 123075
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3257258913280733e-06,
+      "loss": 0.2655,
+      "step": 123080
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3245261766319247e-06,
+      "loss": 0.2621,
+      "step": 123085
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.32332675636186e-06,
+      "loss": 0.2525,
+      "step": 123090
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3221276305334603e-06,
+      "loss": 0.2358,
+      "step": 123095
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3209287991622865e-06,
+      "loss": 0.2428,
+      "step": 123100
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3197302622639027e-06,
+      "loss": 0.2508,
+      "step": 123105
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.318532019853875e-06,
+      "loss": 0.2375,
+      "step": 123110
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3173340719477632e-06,
+      "loss": 0.2348,
+      "step": 123115
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.316136418561121e-06,
+      "loss": 0.242,
+      "step": 123120
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3149390597094976e-06,
+      "loss": 0.2382,
+      "step": 123125
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.313741995408436e-06,
+      "loss": 0.2548,
+      "step": 123130
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3125452256734877e-06,
+      "loss": 0.2716,
+      "step": 123135
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3113487505201843e-06,
+      "loss": 0.2397,
+      "step": 123140
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3101525699640693e-06,
+      "loss": 0.263,
+      "step": 123145
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.308956684020666e-06,
+      "loss": 0.2503,
+      "step": 123150
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.30776109270551e-06,
+      "loss": 0.2225,
+      "step": 123155
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.306565796034124e-06,
+      "loss": 0.257,
+      "step": 123160
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.3053707940220188e-06,
+      "loss": 0.2497,
+      "step": 123165
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.30417608668472e-06,
+      "loss": 0.2441,
+      "step": 123170
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.302981674037741e-06,
+      "loss": 0.2544,
+      "step": 123175
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.301787556096585e-06,
+      "loss": 0.2273,
+      "step": 123180
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.300593732876757e-06,
+      "loss": 0.2501,
+      "step": 123185
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.2994002043937617e-06,
+      "loss": 0.2558,
+      "step": 123190
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.2982069706630916e-06,
+      "loss": 0.2451,
+      "step": 123195
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.2970140317002455e-06,
+      "loss": 0.2257,
+      "step": 123200
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.295821387520708e-06,
+      "loss": 0.2497,
+      "step": 123205
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.2946290381399698e-06,
+      "loss": 0.2344,
+      "step": 123210
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2934369835735132e-06,
+      "loss": 0.2731,
+      "step": 123215
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2922452238368115e-06,
+      "loss": 0.2455,
+      "step": 123220
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2910537589453355e-06,
+      "loss": 0.2615,
+      "step": 123225
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2898625889145653e-06,
+      "loss": 0.2281,
+      "step": 123230
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2886717137599626e-06,
+      "loss": 0.2764,
+      "step": 123235
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.287481133496991e-06,
+      "loss": 0.2407,
+      "step": 123240
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2862908481411037e-06,
+      "loss": 0.2603,
+      "step": 123245
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.285100857707764e-06,
+      "loss": 0.2479,
+      "step": 123250
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2839111622124184e-06,
+      "loss": 0.2535,
+      "step": 123255
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2827217616705145e-06,
+      "loss": 0.2414,
+      "step": 123260
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2815326560975015e-06,
+      "loss": 0.2627,
+      "step": 123265
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.280343845508817e-06,
+      "loss": 0.2377,
+      "step": 123270
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2791553299198936e-06,
+      "loss": 0.2502,
+      "step": 123275
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2779671093461574e-06,
+      "loss": 0.2412,
+      "step": 123280
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2767791838030515e-06,
+      "loss": 0.2599,
+      "step": 123285
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.275591553305992e-06,
+      "loss": 0.2541,
+      "step": 123290
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.274404217870402e-06,
+      "loss": 0.2362,
+      "step": 123295
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2732171775116884e-06,
+      "loss": 0.2717,
+      "step": 123300
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2720304322452786e-06,
+      "loss": 0.2824,
+      "step": 123305
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2708439820865706e-06,
+      "loss": 0.2555,
+      "step": 123310
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2696578270509804e-06,
+      "loss": 0.261,
+      "step": 123315
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2684719671538983e-06,
+      "loss": 0.2478,
+      "step": 123320
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2672864024107287e-06,
+      "loss": 0.2405,
+      "step": 123325
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2661011328368674e-06,
+      "loss": 0.2556,
+      "step": 123330
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2649161584476998e-06,
+      "loss": 0.2402,
+      "step": 123335
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.263731479258607e-06,
+      "loss": 0.2424,
+      "step": 123340
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2625470952849836e-06,
+      "loss": 0.258,
+      "step": 123345
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2613630065421997e-06,
+      "loss": 0.2495,
+      "step": 123350
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2601792130456295e-06,
+      "loss": 0.2442,
+      "step": 123355
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2589957148106493e-06,
+      "loss": 0.2352,
+      "step": 123360
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2578125118526166e-06,
+      "loss": 0.2491,
+      "step": 123365
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.256629604186908e-06,
+      "loss": 0.2595,
+      "step": 123370
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.255446991828869e-06,
+      "loss": 0.236,
+      "step": 123375
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2542646747938682e-06,
+      "loss": 0.2478,
+      "step": 123380
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2530826530972493e-06,
+      "loss": 0.244,
+      "step": 123385
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2519009267543605e-06,
+      "loss": 0.2586,
+      "step": 123390
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2507194957805455e-06,
+      "loss": 0.2511,
+      "step": 123395
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2495383601911474e-06,
+      "loss": 0.2539,
+      "step": 123400
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.248357520001501e-06,
+      "loss": 0.2443,
+      "step": 123405
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.247176975226939e-06,
+      "loss": 0.2526,
+      "step": 123410
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.245996725882782e-06,
+      "loss": 0.2532,
+      "step": 123415
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.244816771984365e-06,
+      "loss": 0.2536,
+      "step": 123420
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.243637113547009e-06,
+      "loss": 0.2413,
+      "step": 123425
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.24245775058603e-06,
+      "loss": 0.2432,
+      "step": 123430
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.241278683116732e-06,
+      "loss": 0.2602,
+      "step": 123435
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2400999111544392e-06,
+      "loss": 0.2423,
+      "step": 123440
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.238921434714447e-06,
+      "loss": 0.2605,
+      "step": 123445
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.237743253812061e-06,
+      "loss": 0.2329,
+      "step": 123450
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2365653684625732e-06,
+      "loss": 0.2502,
+      "step": 123455
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2353877786812887e-06,
+      "loss": 0.2332,
+      "step": 123460
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.23421048448349e-06,
+      "loss": 0.2649,
+      "step": 123465
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2330334858844615e-06,
+      "loss": 0.2413,
+      "step": 123470
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2318567828994858e-06,
+      "loss": 0.2451,
+      "step": 123475
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2306803755438533e-06,
+      "loss": 0.2572,
+      "step": 123480
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2295042638328294e-06,
+      "loss": 0.2657,
+      "step": 123485
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.22832844778168e-06,
+      "loss": 0.2541,
+      "step": 123490
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.2271529274056873e-06,
+      "loss": 0.2277,
+      "step": 123495
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2259777027201027e-06,
+      "loss": 0.2265,
+      "step": 123500
+    },
+    {
+      "epoch": 4.35,
+      "eval_loss": 0.24895407259464264,
+      "eval_runtime": 10.5376,
+      "eval_samples_per_second": 9.49,
+      "eval_steps_per_second": 9.49,
+      "step": 123500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2248027737401896e-06,
+      "loss": 0.2476,
+      "step": 123505
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2236281404811988e-06,
+      "loss": 0.256,
+      "step": 123510
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.222453802958391e-06,
+      "loss": 0.2702,
+      "step": 123515
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.221279761187009e-06,
+      "loss": 0.2348,
+      "step": 123520
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.220106015182294e-06,
+      "loss": 0.2549,
+      "step": 123525
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.218932564959489e-06,
+      "loss": 0.2425,
+      "step": 123530
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2177594105338346e-06,
+      "loss": 0.2566,
+      "step": 123535
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.216586551920563e-06,
+      "loss": 0.2586,
+      "step": 123540
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.21564847802509e-06,
+      "loss": 0.2426,
+      "step": 123545
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2144761519124757e-06,
+      "loss": 0.2442,
+      "step": 123550
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2133041216548717e-06,
+      "loss": 0.2617,
+      "step": 123555
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2121323872674932e-06,
+      "loss": 0.2533,
+      "step": 123560
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.210960948765556e-06,
+      "loss": 0.2569,
+      "step": 123565
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.209789806164278e-06,
+      "loss": 0.2602,
+      "step": 123570
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.208618959478859e-06,
+      "loss": 0.2582,
+      "step": 123575
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2074484087244996e-06,
+      "loss": 0.2369,
+      "step": 123580
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.206278153916405e-06,
+      "loss": 0.2599,
+      "step": 123585
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.205108195069769e-06,
+      "loss": 0.2393,
+      "step": 123590
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2039385321997784e-06,
+      "loss": 0.2642,
+      "step": 123595
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2027691653216194e-06,
+      "loss": 0.2589,
+      "step": 123600
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2016000944504842e-06,
+      "loss": 0.268,
+      "step": 123605
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.2004313196015454e-06,
+      "loss": 0.2664,
+      "step": 123610
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1992628407899757e-06,
+      "loss": 0.2459,
+      "step": 123615
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1980946580309524e-06,
+      "loss": 0.2592,
+      "step": 123620
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1969267713396495e-06,
+      "loss": 0.256,
+      "step": 123625
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1957591807312215e-06,
+      "loss": 0.2237,
+      "step": 123630
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1945918862208343e-06,
+      "loss": 0.2536,
+      "step": 123635
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1934248878236364e-06,
+      "loss": 0.2541,
+      "step": 123640
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.192258185554791e-06,
+      "loss": 0.2483,
+      "step": 123645
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.191091779429441e-06,
+      "loss": 0.2494,
+      "step": 123650
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.189925669462736e-06,
+      "loss": 0.2473,
+      "step": 123655
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.188759855669806e-06,
+      "loss": 0.251,
+      "step": 123660
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1875943380658014e-06,
+      "loss": 0.2533,
+      "step": 123665
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.186429116665847e-06,
+      "loss": 0.2504,
+      "step": 123670
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1852641914850757e-06,
+      "loss": 0.2285,
+      "step": 123675
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1840995625386163e-06,
+      "loss": 0.26,
+      "step": 123680
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1829352298415875e-06,
+      "loss": 0.2388,
+      "step": 123685
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.18177119340911e-06,
+      "loss": 0.2301,
+      "step": 123690
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.180607453256289e-06,
+      "loss": 0.2317,
+      "step": 123695
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1794440093982488e-06,
+      "loss": 0.2423,
+      "step": 123700
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1782808618500877e-06,
+      "loss": 0.2753,
+      "step": 123705
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.177118010626908e-06,
+      "loss": 0.2499,
+      "step": 123710
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1759554557438083e-06,
+      "loss": 0.266,
+      "step": 123715
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.174793197215888e-06,
+      "loss": 0.2148,
+      "step": 123720
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.173631235058232e-06,
+      "loss": 0.2433,
+      "step": 123725
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.172469569285934e-06,
+      "loss": 0.2612,
+      "step": 123730
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.171308199914071e-06,
+      "loss": 0.2477,
+      "step": 123735
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.170147126957733e-06,
+      "loss": 0.2531,
+      "step": 123740
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.168986350431987e-06,
+      "loss": 0.2647,
+      "step": 123745
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.167825870351908e-06,
+      "loss": 0.2689,
+      "step": 123750
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.166665686732558e-06,
+      "loss": 0.2332,
+      "step": 123755
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1655057995890126e-06,
+      "loss": 0.244,
+      "step": 123760
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1643462089363243e-06,
+      "loss": 0.2512,
+      "step": 123765
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.16318691478955e-06,
+      "loss": 0.2412,
+      "step": 123770
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.162027917163742e-06,
+      "loss": 0.2264,
+      "step": 123775
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1608692160739465e-06,
+      "loss": 0.2461,
+      "step": 123780
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1597108115352204e-06,
+      "loss": 0.2477,
+      "step": 123785
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1585527035625913e-06,
+      "loss": 0.2551,
+      "step": 123790
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.157394892171108e-06,
+      "loss": 0.2482,
+      "step": 123795
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.156237377375797e-06,
+      "loss": 0.2461,
+      "step": 123800
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1550801591916912e-06,
+      "loss": 0.2364,
+      "step": 123805
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1539232376338065e-06,
+      "loss": 0.2472,
+      "step": 123810
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1527666127171802e-06,
+      "loss": 0.2429,
+      "step": 123815
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1516102844568204e-06,
+      "loss": 0.2437,
+      "step": 123820
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.150454252867745e-06,
+      "loss": 0.2514,
+      "step": 123825
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.149298517964957e-06,
+      "loss": 0.2665,
+      "step": 123830
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1481430797634735e-06,
+      "loss": 0.2393,
+      "step": 123835
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.146987938278286e-06,
+      "loss": 0.2507,
+      "step": 123840
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.145833093524405e-06,
+      "loss": 0.2336,
+      "step": 123845
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.144678545516815e-06,
+      "loss": 0.2427,
+      "step": 123850
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1435242942705187e-06,
+      "loss": 0.255,
+      "step": 123855
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1423703398004954e-06,
+      "loss": 0.243,
+      "step": 123860
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.141216682121727e-06,
+      "loss": 0.2506,
+      "step": 123865
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1400633212491944e-06,
+      "loss": 0.2229,
+      "step": 123870
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1389102571978764e-06,
+      "loss": 0.266,
+      "step": 123875
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.137757489982742e-06,
+      "loss": 0.2635,
+      "step": 123880
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1366050196187564e-06,
+      "loss": 0.2646,
+      "step": 123885
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1354528461208944e-06,
+      "loss": 0.2281,
+      "step": 123890
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.134300969504102e-06,
+      "loss": 0.2389,
+      "step": 123895
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1331493897833452e-06,
+      "loss": 0.2276,
+      "step": 123900
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1319981069735706e-06,
+      "loss": 0.246,
+      "step": 123905
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1308471210897352e-06,
+      "loss": 0.2725,
+      "step": 123910
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.12969643214678e-06,
+      "loss": 0.2385,
+      "step": 123915
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.12854604015964e-06,
+      "loss": 0.2329,
+      "step": 123920
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1273959451432567e-06,
+      "loss": 0.25,
+      "step": 123925
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.126246147112565e-06,
+      "loss": 0.2414,
+      "step": 123930
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1250966460824943e-06,
+      "loss": 0.255,
+      "step": 123935
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.123947442067964e-06,
+      "loss": 0.2632,
+      "step": 123940
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.122798535083906e-06,
+      "loss": 0.2888,
+      "step": 123945
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.121649925145225e-06,
+      "loss": 0.24,
+      "step": 123950
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.120501612266848e-06,
+      "loss": 0.2417,
+      "step": 123955
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.11935359646368e-06,
+      "loss": 0.2631,
+      "step": 123960
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.118205877750623e-06,
+      "loss": 0.2194,
+      "step": 123965
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.117058456142587e-06,
+      "loss": 0.2435,
+      "step": 123970
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.115911331654466e-06,
+      "loss": 0.2673,
+      "step": 123975
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1147645043011533e-06,
+      "loss": 0.247,
+      "step": 123980
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.11361797409754e-06,
+      "loss": 0.2425,
+      "step": 123985
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1124717410585164e-06,
+      "loss": 0.2498,
+      "step": 123990
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.111325805198966e-06,
+      "loss": 0.2905,
+      "step": 123995
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1101801665337595e-06,
+      "loss": 0.254,
+      "step": 124000
+    },
+    {
+      "epoch": 4.36,
+      "eval_loss": 0.24889464676380157,
+      "eval_runtime": 10.5436,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 124000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1090348250777797e-06,
+      "loss": 0.2245,
+      "step": 124005
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.107889780845904e-06,
+      "loss": 0.2225,
+      "step": 124010
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1067450338529887e-06,
+      "loss": 0.2623,
+      "step": 124015
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1056005841139007e-06,
+      "loss": 0.2521,
+      "step": 124020
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.104456431643503e-06,
+      "loss": 0.2633,
+      "step": 124025
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.103312576456651e-06,
+      "loss": 0.2417,
+      "step": 124030
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1021690185681952e-06,
+      "loss": 0.2539,
+      "step": 124035
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.1010257579929806e-06,
+      "loss": 0.2413,
+      "step": 124040
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.099882794745861e-06,
+      "loss": 0.2429,
+      "step": 124045
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.0987401288416697e-06,
+      "loss": 0.2523,
+      "step": 124050
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.0975977602952418e-06,
+      "loss": 0.2338,
+      "step": 124055
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.0964556891214127e-06,
+      "loss": 0.2462,
+      "step": 124060
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.0953139153350175e-06,
+      "loss": 0.2491,
+      "step": 124065
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0941724389508783e-06,
+      "loss": 0.237,
+      "step": 124070
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0930312599838132e-06,
+      "loss": 0.2343,
+      "step": 124075
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0918903784486355e-06,
+      "loss": 0.2433,
+      "step": 124080
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0907497943601696e-06,
+      "loss": 0.2395,
+      "step": 124085
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0896095077332202e-06,
+      "loss": 0.2481,
+      "step": 124090
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.088469518582592e-06,
+      "loss": 0.2496,
+      "step": 124095
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0873298269230816e-06,
+      "loss": 0.2363,
+      "step": 124100
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0861904327695e-06,
+      "loss": 0.2617,
+      "step": 124105
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0850513361366313e-06,
+      "loss": 0.2712,
+      "step": 124110
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0839125370392732e-06,
+      "loss": 0.2404,
+      "step": 124115
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.082774035492202e-06,
+      "loss": 0.2523,
+      "step": 124120
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0816358315102147e-06,
+      "loss": 0.2495,
+      "step": 124125
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0804979251080795e-06,
+      "loss": 0.2443,
+      "step": 124130
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0793603163005713e-06,
+      "loss": 0.263,
+      "step": 124135
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.078223005102467e-06,
+      "loss": 0.26,
+      "step": 124140
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.077085991528532e-06,
+      "loss": 0.251,
+      "step": 124145
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0759492755935304e-06,
+      "loss": 0.2627,
+      "step": 124150
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0748128573122138e-06,
+      "loss": 0.2459,
+      "step": 124155
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.073676736699348e-06,
+      "loss": 0.2482,
+      "step": 124160
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0725409137696774e-06,
+      "loss": 0.2548,
+      "step": 124165
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0714053885379585e-06,
+      "loss": 0.2491,
+      "step": 124170
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0702701610189223e-06,
+      "loss": 0.2515,
+      "step": 124175
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.069135231227323e-06,
+      "loss": 0.2509,
+      "step": 124180
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.068000599177891e-06,
+      "loss": 0.2418,
+      "step": 124185
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0668662648853587e-06,
+      "loss": 0.2332,
+      "step": 124190
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0657322283644503e-06,
+      "loss": 0.2485,
+      "step": 124195
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0645984896298986e-06,
+      "loss": 0.2504,
+      "step": 124200
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0634650486964225e-06,
+      "loss": 0.2497,
+      "step": 124205
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.062331905578735e-06,
+      "loss": 0.2594,
+      "step": 124210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0611990602915464e-06,
+      "loss": 0.2553,
+      "step": 124215
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0600665128495707e-06,
+      "loss": 0.2484,
+      "step": 124220
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0589342632675173e-06,
+      "loss": 0.2325,
+      "step": 124225
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0578023115600866e-06,
+      "loss": 0.2434,
+      "step": 124230
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.056670657741966e-06,
+      "loss": 0.2457,
+      "step": 124235
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0555393018278635e-06,
+      "loss": 0.2459,
+      "step": 124240
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.054408243832459e-06,
+      "loss": 0.2463,
+      "step": 124245
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.053277483770444e-06,
+      "loss": 0.2368,
+      "step": 124250
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0521470216564946e-06,
+      "loss": 0.2472,
+      "step": 124255
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0510168575052945e-06,
+      "loss": 0.2458,
+      "step": 124260
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0498869913315206e-06,
+      "loss": 0.2398,
+      "step": 124265
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.048757423149833e-06,
+      "loss": 0.2281,
+      "step": 124270
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0476281529749092e-06,
+      "loss": 0.268,
+      "step": 124275
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.046499180821401e-06,
+      "loss": 0.259,
+      "step": 124280
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.045370506703981e-06,
+      "loss": 0.2406,
+      "step": 124285
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0442421306372945e-06,
+      "loss": 0.258,
+      "step": 124290
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.043114052635997e-06,
+      "loss": 0.2724,
+      "step": 124295
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0419862727147356e-06,
+      "loss": 0.2524,
+      "step": 124300
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0408587908881505e-06,
+      "loss": 0.2328,
+      "step": 124305
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0397316071708806e-06,
+      "loss": 0.2231,
+      "step": 124310
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0386047215775692e-06,
+      "loss": 0.2343,
+      "step": 124315
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0374781341228444e-06,
+      "loss": 0.2617,
+      "step": 124320
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.036351844821327e-06,
+      "loss": 0.2502,
+      "step": 124325
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0352258536876528e-06,
+      "loss": 0.2625,
+      "step": 124330
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.03410016073643e-06,
+      "loss": 0.2404,
+      "step": 124335
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0329747659822873e-06,
+      "loss": 0.2491,
+      "step": 124340
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.031849669439831e-06,
+      "loss": 0.24,
+      "step": 124345
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.030724871123668e-06,
+      "loss": 0.2722,
+      "step": 124350
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.029600371048407e-06,
+      "loss": 0.2395,
+      "step": 124355
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0284761692286493e-06,
+      "loss": 0.2606,
+      "step": 124360
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.027352265678986e-06,
+      "loss": 0.2531,
+      "step": 124365
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0262286604140144e-06,
+      "loss": 0.2336,
+      "step": 124370
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0251053534483254e-06,
+      "loss": 0.2459,
+      "step": 124375
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0239823447964963e-06,
+      "loss": 0.2519,
+      "step": 124380
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0228596344731205e-06,
+      "loss": 0.2721,
+      "step": 124385
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.021737222492767e-06,
+      "loss": 0.2573,
+      "step": 124390
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0206151088700137e-06,
+      "loss": 0.2838,
+      "step": 124395
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.019493293619432e-06,
+      "loss": 0.2519,
+      "step": 124400
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.018371776755579e-06,
+      "loss": 0.2379,
+      "step": 124405
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0172505582930295e-06,
+      "loss": 0.2365,
+      "step": 124410
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0161296382463335e-06,
+      "loss": 0.2537,
+      "step": 124415
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0150090166300484e-06,
+      "loss": 0.2313,
+      "step": 124420
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.013888693458718e-06,
+      "loss": 0.2522,
+      "step": 124425
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0127686687469004e-06,
+      "loss": 0.2342,
+      "step": 124430
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.011648942509131e-06,
+      "loss": 0.2482,
+      "step": 124435
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.010529514759946e-06,
+      "loss": 0.2342,
+      "step": 124440
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0094103855138852e-06,
+      "loss": 0.23,
+      "step": 124445
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0082915547854824e-06,
+      "loss": 0.2378,
+      "step": 124450
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.007173022589262e-06,
+      "loss": 0.2624,
+      "step": 124455
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0060547889397484e-06,
+      "loss": 0.2536,
+      "step": 124460
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.004936853851455e-06,
+      "loss": 0.2516,
+      "step": 124465
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0038192173389063e-06,
+      "loss": 0.2552,
+      "step": 124470
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0027018794166102e-06,
+      "loss": 0.2291,
+      "step": 124475
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0015848400990716e-06,
+      "loss": 0.257,
+      "step": 124480
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.000468099400796e-06,
+      "loss": 0.2471,
+      "step": 124485
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9993516573362904e-06,
+      "loss": 0.2406,
+      "step": 124490
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.998235513920038e-06,
+      "loss": 0.2725,
+      "step": 124495
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9971196691665415e-06,
+      "loss": 0.245,
+      "step": 124500
+    },
+    {
+      "epoch": 4.38,
+      "eval_loss": 0.24900418519973755,
+      "eval_runtime": 10.5316,
+      "eval_samples_per_second": 9.495,
+      "eval_steps_per_second": 9.495,
+      "step": 124500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9960041230902893e-06,
+      "loss": 0.2467,
+      "step": 124505
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9948888757057637e-06,
+      "loss": 0.2583,
+      "step": 124510
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9937739270274454e-06,
+      "loss": 0.2436,
+      "step": 124515
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9926592770698057e-06,
+      "loss": 0.2666,
+      "step": 124520
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9915449258473274e-06,
+      "loss": 0.2608,
+      "step": 124525
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.990430873374477e-06,
+      "loss": 0.2429,
+      "step": 124530
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.989317119665715e-06,
+      "loss": 0.2524,
+      "step": 124535
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9882036647355025e-06,
+      "loss": 0.2692,
+      "step": 124540
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.987090508598305e-06,
+      "loss": 0.2596,
+      "step": 124545
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9859776512685667e-06,
+      "loss": 0.2396,
+      "step": 124550
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.984865092760746e-06,
+      "loss": 0.2326,
+      "step": 124555
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.983752833089278e-06,
+      "loss": 0.2454,
+      "step": 124560
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9826408722686178e-06,
+      "loss": 0.2596,
+      "step": 124565
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9815292103131988e-06,
+      "loss": 0.273,
+      "step": 124570
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9804178472374506e-06,
+      "loss": 0.2503,
+      "step": 124575
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.979306783055801e-06,
+      "loss": 0.248,
+      "step": 124580
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.978196017782688e-06,
+      "loss": 0.2314,
+      "step": 124585
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9770855514325284e-06,
+      "loss": 0.2342,
+      "step": 124590
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9759753840197374e-06,
+      "loss": 0.2402,
+      "step": 124595
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.97486551555873e-06,
+      "loss": 0.2625,
+      "step": 124600
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.973755946063918e-06,
+      "loss": 0.2473,
+      "step": 124605
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9726466755497163e-06,
+      "loss": 0.2347,
+      "step": 124610
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9715377040305182e-06,
+      "loss": 0.2508,
+      "step": 124615
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.970429031520729e-06,
+      "loss": 0.2427,
+      "step": 124620
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.969320658034743e-06,
+      "loss": 0.2439,
+      "step": 124625
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.9682125835869507e-06,
+      "loss": 0.249,
+      "step": 124630
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.967104808191736e-06,
+      "loss": 0.2222,
+      "step": 124635
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9659973318634896e-06,
+      "loss": 0.2549,
+      "step": 124640
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.964890154616586e-06,
+      "loss": 0.2388,
+      "step": 124645
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9637832764654057e-06,
+      "loss": 0.27,
+      "step": 124650
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.962676697424312e-06,
+      "loss": 0.2355,
+      "step": 124655
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.96157041750768e-06,
+      "loss": 0.2612,
+      "step": 124660
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.960464436729878e-06,
+      "loss": 0.2444,
+      "step": 124665
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.959358755105262e-06,
+      "loss": 0.2326,
+      "step": 124670
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9582533726481833e-06,
+      "loss": 0.2414,
+      "step": 124675
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9571482893730007e-06,
+      "loss": 0.2675,
+      "step": 124680
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9560435052940667e-06,
+      "loss": 0.2569,
+      "step": 124685
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9549390204257194e-06,
+      "loss": 0.2463,
+      "step": 124690
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.953834834782295e-06,
+      "loss": 0.2412,
+      "step": 124695
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.952730948378145e-06,
+      "loss": 0.2457,
+      "step": 124700
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.951627361227593e-06,
+      "loss": 0.2618,
+      "step": 124705
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.950524073344967e-06,
+      "loss": 0.2526,
+      "step": 124710
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.949421084744599e-06,
+      "loss": 0.2455,
+      "step": 124715
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.948318395440804e-06,
+      "loss": 0.2467,
+      "step": 124720
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9472160054479077e-06,
+      "loss": 0.2602,
+      "step": 124725
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.946113914780212e-06,
+      "loss": 0.2352,
+      "step": 124730
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.945012123452042e-06,
+      "loss": 0.2524,
+      "step": 124735
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.943910631477694e-06,
+      "loss": 0.245,
+      "step": 124740
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9428094388714707e-06,
+      "loss": 0.2418,
+      "step": 124745
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9417085456476693e-06,
+      "loss": 0.2375,
+      "step": 124750
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.940607951820592e-06,
+      "loss": 0.2728,
+      "step": 124755
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9395076574045213e-06,
+      "loss": 0.2674,
+      "step": 124760
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.938407662413741e-06,
+      "loss": 0.2358,
+      "step": 124765
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.937307966862545e-06,
+      "loss": 0.2655,
+      "step": 124770
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.936208570765202e-06,
+      "loss": 0.2538,
+      "step": 124775
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.935109474135996e-06,
+      "loss": 0.2464,
+      "step": 124780
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9340106769891924e-06,
+      "loss": 0.2625,
+      "step": 124785
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9329121793390554e-06,
+      "loss": 0.2676,
+      "step": 124790
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9318139811998537e-06,
+      "loss": 0.2386,
+      "step": 124795
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9307160825858483e-06,
+      "loss": 0.2545,
+      "step": 124800
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.929618483511289e-06,
+      "loss": 0.2529,
+      "step": 124805
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9285211839904256e-06,
+      "loss": 0.2464,
+      "step": 124810
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9274241840375155e-06,
+      "loss": 0.2499,
+      "step": 124815
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9263274836667927e-06,
+      "loss": 0.2586,
+      "step": 124820
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9252310828925035e-06,
+      "loss": 0.2359,
+      "step": 124825
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9241349817288812e-06,
+      "loss": 0.2371,
+      "step": 124830
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9230391801901588e-06,
+      "loss": 0.2598,
+      "step": 124835
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9219436782905666e-06,
+      "loss": 0.2648,
+      "step": 124840
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9208484760443235e-06,
+      "loss": 0.2486,
+      "step": 124845
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9197535734656545e-06,
+      "loss": 0.2699,
+      "step": 124850
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9186589705687762e-06,
+      "loss": 0.2401,
+      "step": 124855
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.917564667367902e-06,
+      "loss": 0.2449,
+      "step": 124860
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9164706638772316e-06,
+      "loss": 0.236,
+      "step": 124865
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9153769601109844e-06,
+      "loss": 0.26,
+      "step": 124870
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9142835560833515e-06,
+      "loss": 0.2426,
+      "step": 124875
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9131904518085276e-06,
+      "loss": 0.2414,
+      "step": 124880
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.912097647300715e-06,
+      "loss": 0.2513,
+      "step": 124885
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.911005142574099e-06,
+      "loss": 0.2624,
+      "step": 124890
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9099129376428658e-06,
+      "loss": 0.2521,
+      "step": 124895
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.908821032521196e-06,
+      "loss": 0.2464,
+      "step": 124900
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.907729427223265e-06,
+      "loss": 0.2497,
+      "step": 124905
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9066381217632516e-06,
+      "loss": 0.2499,
+      "step": 124910
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.9055471161553262e-06,
+      "loss": 0.2384,
+      "step": 124915
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.9044564104136492e-06,
+      "loss": 0.2524,
+      "step": 124920
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.9033660045523788e-06,
+      "loss": 0.2348,
+      "step": 124925
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.902275898585687e-06,
+      "loss": 0.2543,
+      "step": 124930
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.9011860925277153e-06,
+      "loss": 0.2533,
+      "step": 124935
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.9000965863926245e-06,
+      "loss": 0.231,
+      "step": 124940
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8990073801945508e-06,
+      "loss": 0.2458,
+      "step": 124945
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8979184739476463e-06,
+      "loss": 0.2199,
+      "step": 124950
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8968298676660472e-06,
+      "loss": 0.2734,
+      "step": 124955
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8957415613638813e-06,
+      "loss": 0.2508,
+      "step": 124960
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8946535550552897e-06,
+      "loss": 0.2559,
+      "step": 124965
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8935658487543944e-06,
+      "loss": 0.2305,
+      "step": 124970
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8924784424753179e-06,
+      "loss": 0.2515,
+      "step": 124975
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.891391336232176e-06,
+      "loss": 0.2453,
+      "step": 124980
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8903045300390942e-06,
+      "loss": 0.244,
+      "step": 124985
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8892180239101747e-06,
+      "loss": 0.2466,
+      "step": 124990
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8881318178595314e-06,
+      "loss": 0.2572,
+      "step": 124995
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8870459119012612e-06,
+      "loss": 0.2675,
+      "step": 125000
+    },
+    {
+      "epoch": 4.4,
+      "eval_loss": 0.24892328679561615,
+      "eval_runtime": 10.538,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 125000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8859603060494695e-06,
+      "loss": 0.2666,
+      "step": 125005
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8848750003182537e-06,
+      "loss": 0.2405,
+      "step": 125010
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.883789994721702e-06,
+      "loss": 0.2561,
+      "step": 125015
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8827052892738979e-06,
+      "loss": 0.2539,
+      "step": 125020
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8816208839889382e-06,
+      "loss": 0.2762,
+      "step": 125025
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8805367788808925e-06,
+      "loss": 0.2596,
+      "step": 125030
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.879452973963841e-06,
+      "loss": 0.2361,
+      "step": 125035
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8783694692518506e-06,
+      "loss": 0.2608,
+      "step": 125040
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8772862647589957e-06,
+      "loss": 0.2547,
+      "step": 125045
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8762033604993457e-06,
+      "loss": 0.2269,
+      "step": 125050
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.875120756486956e-06,
+      "loss": 0.2755,
+      "step": 125055
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8740384527358767e-06,
+      "loss": 0.2498,
+      "step": 125060
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8729564492601742e-06,
+      "loss": 0.2621,
+      "step": 125065
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.87187474607389e-06,
+      "loss": 0.249,
+      "step": 125070
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8707933431910657e-06,
+      "loss": 0.2657,
+      "step": 125075
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.869712240625751e-06,
+      "loss": 0.251,
+      "step": 125080
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8686314383919794e-06,
+      "loss": 0.2563,
+      "step": 125085
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.867550936503784e-06,
+      "loss": 0.2529,
+      "step": 125090
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8664707349751926e-06,
+      "loss": 0.2425,
+      "step": 125095
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8653908338202325e-06,
+      "loss": 0.2547,
+      "step": 125100
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8643112330529288e-06,
+      "loss": 0.2334,
+      "step": 125105
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8632319326872982e-06,
+      "loss": 0.2442,
+      "step": 125110
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8621529327373488e-06,
+      "loss": 0.2678,
+      "step": 125115
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8610742332171e-06,
+      "loss": 0.2398,
+      "step": 125120
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8599958341405515e-06,
+      "loss": 0.2348,
+      "step": 125125
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8589177355217092e-06,
+      "loss": 0.2419,
+      "step": 125130
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8578399373745643e-06,
+      "loss": 0.2521,
+      "step": 125135
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8567624397131195e-06,
+      "loss": 0.2476,
+      "step": 125140
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8556852425513637e-06,
+      "loss": 0.2725,
+      "step": 125145
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8546083459032775e-06,
+      "loss": 0.2639,
+      "step": 125150
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8535317497828552e-06,
+      "loss": 0.2557,
+      "step": 125155
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8524554542040634e-06,
+      "loss": 0.2451,
+      "step": 125160
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8513794591808853e-06,
+      "loss": 0.244,
+      "step": 125165
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8503037647272902e-06,
+      "loss": 0.2563,
+      "step": 125170
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8492283708572422e-06,
+      "loss": 0.2534,
+      "step": 125175
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8481532775847077e-06,
+      "loss": 0.2703,
+      "step": 125180
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8470784849236479e-06,
+      "loss": 0.2378,
+      "step": 125185
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8460039928880125e-06,
+      "loss": 0.2739,
+      "step": 125190
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.8449298014917572e-06,
+      "loss": 0.2582,
+      "step": 125195
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.843855910748829e-06,
+      "loss": 0.2524,
+      "step": 125200
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8427823206731699e-06,
+      "loss": 0.2527,
+      "step": 125205
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.841709031278724e-06,
+      "loss": 0.2494,
+      "step": 125210
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.840636042579419e-06,
+      "loss": 0.2502,
+      "step": 125215
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8395633545891965e-06,
+      "loss": 0.2505,
+      "step": 125220
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8384909673219818e-06,
+      "loss": 0.2543,
+      "step": 125225
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8374188807916942e-06,
+      "loss": 0.2378,
+      "step": 125230
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8363470950122614e-06,
+      "loss": 0.2551,
+      "step": 125235
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.835275609997597e-06,
+      "loss": 0.2633,
+      "step": 125240
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8342044257616097e-06,
+      "loss": 0.2501,
+      "step": 125245
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8331335423182106e-06,
+      "loss": 0.2508,
+      "step": 125250
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8320629596813078e-06,
+      "loss": 0.2496,
+      "step": 125255
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8309926778647928e-06,
+      "loss": 0.2322,
+      "step": 125260
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8299226968825745e-06,
+      "loss": 0.2623,
+      "step": 125265
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8288530167485385e-06,
+      "loss": 0.2539,
+      "step": 125270
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8277836374765767e-06,
+      "loss": 0.2352,
+      "step": 125275
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.826714559080575e-06,
+      "loss": 0.2707,
+      "step": 125280
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8256457815744117e-06,
+      "loss": 0.2852,
+      "step": 125285
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8245773049719639e-06,
+      "loss": 0.2284,
+      "step": 125290
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.823509129287107e-06,
+      "loss": 0.2353,
+      "step": 125295
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8224412545337132e-06,
+      "loss": 0.2682,
+      "step": 125300
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8213736807256382e-06,
+      "loss": 0.262,
+      "step": 125305
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8203064078767568e-06,
+      "loss": 0.284,
+      "step": 125310
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8192394360009218e-06,
+      "loss": 0.239,
+      "step": 125315
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8181727651119807e-06,
+      "loss": 0.245,
+      "step": 125320
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8171063952237888e-06,
+      "loss": 0.2323,
+      "step": 125325
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8160403263501963e-06,
+      "loss": 0.2541,
+      "step": 125330
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8149745585050421e-06,
+      "loss": 0.2466,
+      "step": 125335
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8139090917021623e-06,
+      "loss": 0.2559,
+      "step": 125340
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8128439259553875e-06,
+      "loss": 0.2354,
+      "step": 125345
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8117790612785596e-06,
+      "loss": 0.2612,
+      "step": 125350
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8107144976855006e-06,
+      "loss": 0.251,
+      "step": 125355
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8096502351900274e-06,
+      "loss": 0.2372,
+      "step": 125360
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8085862738059622e-06,
+      "loss": 0.274,
+      "step": 125365
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8075226135471247e-06,
+      "loss": 0.2536,
+      "step": 125370
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8064592544273146e-06,
+      "loss": 0.2435,
+      "step": 125375
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8053961964603517e-06,
+      "loss": 0.2335,
+      "step": 125380
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8043334396600304e-06,
+      "loss": 0.2588,
+      "step": 125385
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8032709840401563e-06,
+      "loss": 0.2283,
+      "step": 125390
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8022088296145212e-06,
+      "loss": 0.2606,
+      "step": 125395
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.801146976396914e-06,
+      "loss": 0.2641,
+      "step": 125400
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.8000854244011239e-06,
+      "loss": 0.2468,
+      "step": 125405
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7990241736409369e-06,
+      "loss": 0.244,
+      "step": 125410
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7979632241301337e-06,
+      "loss": 0.2569,
+      "step": 125415
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7969025758824864e-06,
+      "loss": 0.2388,
+      "step": 125420
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.795842228911762e-06,
+      "loss": 0.2548,
+      "step": 125425
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7947821832317357e-06,
+      "loss": 0.243,
+      "step": 125430
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.793722438856174e-06,
+      "loss": 0.257,
+      "step": 125435
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7926629957988272e-06,
+      "loss": 0.244,
+      "step": 125440
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7916038540734648e-06,
+      "loss": 0.2558,
+      "step": 125445
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7905450136938284e-06,
+      "loss": 0.2988,
+      "step": 125450
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7894864746736684e-06,
+      "loss": 0.274,
+      "step": 125455
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7884282370267291e-06,
+      "loss": 0.2545,
+      "step": 125460
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7873703007667553e-06,
+      "loss": 0.2523,
+      "step": 125465
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7863126659074802e-06,
+      "loss": 0.2514,
+      "step": 125470
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7852553324626376e-06,
+      "loss": 0.2401,
+      "step": 125475
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7841983004459495e-06,
+      "loss": 0.2538,
+      "step": 125480
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.7831415698711467e-06,
+      "loss": 0.2336,
+      "step": 125485
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7820851407519572e-06,
+      "loss": 0.273,
+      "step": 125490
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7810290131020868e-06,
+      "loss": 0.2575,
+      "step": 125495
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.779973186935252e-06,
+      "loss": 0.2638,
+      "step": 125500
+    },
+    {
+      "epoch": 4.42,
+      "eval_loss": 0.24885064363479614,
+      "eval_runtime": 10.5403,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 125500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7789176622651642e-06,
+      "loss": 0.2569,
+      "step": 125505
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7778624391055293e-06,
+      "loss": 0.2524,
+      "step": 125510
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.776807517470047e-06,
+      "loss": 0.2501,
+      "step": 125515
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7757528973724096e-06,
+      "loss": 0.2501,
+      "step": 125520
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7746985788263199e-06,
+      "loss": 0.2399,
+      "step": 125525
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.773644561845461e-06,
+      "loss": 0.2688,
+      "step": 125530
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7725908464435198e-06,
+      "loss": 0.262,
+      "step": 125535
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7715374326341767e-06,
+      "loss": 0.2531,
+      "step": 125540
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7704843204311179e-06,
+      "loss": 0.2601,
+      "step": 125545
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7694315098480102e-06,
+      "loss": 0.2438,
+      "step": 125550
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7683790008985206e-06,
+      "loss": 0.2567,
+      "step": 125555
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7675372109243276e-06,
+      "loss": 0.2601,
+      "step": 125560
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7664852449497975e-06,
+      "loss": 0.2437,
+      "step": 125565
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7654335806471439e-06,
+      "loss": 0.2388,
+      "step": 125570
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7643822180300196e-06,
+      "loss": 0.2325,
+      "step": 125575
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.763331157112086e-06,
+      "loss": 0.2332,
+      "step": 125580
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.762280397906982e-06,
+      "loss": 0.2463,
+      "step": 125585
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7612299404283494e-06,
+      "loss": 0.2564,
+      "step": 125590
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.760179784689836e-06,
+      "loss": 0.2581,
+      "step": 125595
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7591299307050723e-06,
+      "loss": 0.266,
+      "step": 125600
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7580803784876893e-06,
+      "loss": 0.2848,
+      "step": 125605
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.757031128051312e-06,
+      "loss": 0.2355,
+      "step": 125610
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.755982179409571e-06,
+      "loss": 0.2384,
+      "step": 125615
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7549335325760834e-06,
+      "loss": 0.2755,
+      "step": 125620
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7538851875644606e-06,
+      "loss": 0.2615,
+      "step": 125625
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7528371443883195e-06,
+      "loss": 0.2423,
+      "step": 125630
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.751789403061274e-06,
+      "loss": 0.2409,
+      "step": 125635
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.750741963596919e-06,
+      "loss": 0.2515,
+      "step": 125640
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7496948260088547e-06,
+      "loss": 0.2494,
+      "step": 125645
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.748647990310684e-06,
+      "loss": 0.2517,
+      "step": 125650
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.747601456515996e-06,
+      "loss": 0.2673,
+      "step": 125655
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.746555224638377e-06,
+      "loss": 0.2398,
+      "step": 125660
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7455092946914136e-06,
+      "loss": 0.2578,
+      "step": 125665
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7444636666886894e-06,
+      "loss": 0.2711,
+      "step": 125670
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7434183406437793e-06,
+      "loss": 0.2458,
+      "step": 125675
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7423733165702478e-06,
+      "loss": 0.2469,
+      "step": 125680
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7413285944816754e-06,
+      "loss": 0.2754,
+      "step": 125685
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7402841743916266e-06,
+      "loss": 0.2554,
+      "step": 125690
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7392400563136568e-06,
+      "loss": 0.2446,
+      "step": 125695
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7381962402613277e-06,
+      "loss": 0.2441,
+      "step": 125700
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.737152726248184e-06,
+      "loss": 0.2483,
+      "step": 125705
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.736109514287787e-06,
+      "loss": 0.2573,
+      "step": 125710
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7350666043936786e-06,
+      "loss": 0.2383,
+      "step": 125715
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7340239965793954e-06,
+      "loss": 0.2672,
+      "step": 125720
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7329816908584734e-06,
+      "loss": 0.2704,
+      "step": 125725
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.731939687244455e-06,
+      "loss": 0.2382,
+      "step": 125730
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7308979857508623e-06,
+      "loss": 0.2316,
+      "step": 125735
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.729856586391229e-06,
+      "loss": 0.2461,
+      "step": 125740
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.728815489179067e-06,
+      "loss": 0.2689,
+      "step": 125745
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7277746941279038e-06,
+      "loss": 0.2622,
+      "step": 125750
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7267342012512484e-06,
+      "loss": 0.2526,
+      "step": 125755
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.725694010562609e-06,
+      "loss": 0.2388,
+      "step": 125760
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.7246541220754975e-06,
+      "loss": 0.2314,
+      "step": 125765
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.723614535803414e-06,
+      "loss": 0.2311,
+      "step": 125770
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7225752517598558e-06,
+      "loss": 0.2388,
+      "step": 125775
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7215362699583154e-06,
+      "loss": 0.2332,
+      "step": 125780
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7204975904122872e-06,
+      "loss": 0.2405,
+      "step": 125785
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.719459213135252e-06,
+      "loss": 0.2476,
+      "step": 125790
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7184211381407023e-06,
+      "loss": 0.2752,
+      "step": 125795
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7173833654421046e-06,
+      "loss": 0.2647,
+      "step": 125800
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7163458950529454e-06,
+      "loss": 0.2338,
+      "step": 125805
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.715308726986692e-06,
+      "loss": 0.228,
+      "step": 125810
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7142718612568053e-06,
+      "loss": 0.2639,
+      "step": 125815
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.713235297876753e-06,
+      "loss": 0.2404,
+      "step": 125820
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7121990368599961e-06,
+      "loss": 0.2597,
+      "step": 125825
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.711163078219985e-06,
+      "loss": 0.2799,
+      "step": 125830
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7101274219701756e-06,
+      "loss": 0.2636,
+      "step": 125835
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7090920681240102e-06,
+      "loss": 0.2583,
+      "step": 125840
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7080570166949333e-06,
+      "loss": 0.2649,
+      "step": 125845
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7070222676963897e-06,
+      "loss": 0.2289,
+      "step": 125850
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7059878211418107e-06,
+      "loss": 0.2291,
+      "step": 125855
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7049536770446295e-06,
+      "loss": 0.2542,
+      "step": 125860
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7039198354182718e-06,
+      "loss": 0.2395,
+      "step": 125865
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7028862962761654e-06,
+      "loss": 0.2525,
+      "step": 125870
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7018530596317195e-06,
+      "loss": 0.2491,
+      "step": 125875
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.7008201254983647e-06,
+      "loss": 0.2712,
+      "step": 125880
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6997874938895043e-06,
+      "loss": 0.2482,
+      "step": 125885
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6987551648185495e-06,
+      "loss": 0.2483,
+      "step": 125890
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6977231382988956e-06,
+      "loss": 0.2378,
+      "step": 125895
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6966914143439566e-06,
+      "loss": 0.2375,
+      "step": 125900
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.695659992967119e-06,
+      "loss": 0.2365,
+      "step": 125905
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6946288741817806e-06,
+      "loss": 0.2516,
+      "step": 125910
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6935980580013221e-06,
+      "loss": 0.2581,
+      "step": 125915
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6925675444391386e-06,
+      "loss": 0.262,
+      "step": 125920
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6915373335086054e-06,
+      "loss": 0.2466,
+      "step": 125925
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6905074252231007e-06,
+      "loss": 0.2644,
+      "step": 125930
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6894778195959887e-06,
+      "loss": 0.2811,
+      "step": 125935
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6884485166406504e-06,
+      "loss": 0.2388,
+      "step": 125940
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6874195163704475e-06,
+      "loss": 0.2606,
+      "step": 125945
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6863908187987305e-06,
+      "loss": 0.2765,
+      "step": 125950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6853624239388716e-06,
+      "loss": 0.2499,
+      "step": 125955
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6843343318042104e-06,
+      "loss": 0.2536,
+      "step": 125960
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6833065424081086e-06,
+      "loss": 0.2379,
+      "step": 125965
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6822790557638995e-06,
+      "loss": 0.2696,
+      "step": 125970
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.681251871884934e-06,
+      "loss": 0.2442,
+      "step": 125975
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6802249907845457e-06,
+      "loss": 0.2633,
+      "step": 125980
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6791984124760657e-06,
+      "loss": 0.2424,
+      "step": 125985
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6781721369728248e-06,
+      "loss": 0.2456,
+      "step": 125990
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6771461642881515e-06,
+      "loss": 0.2539,
+      "step": 125995
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6761204944353626e-06,
+      "loss": 0.2602,
+      "step": 126000
+    },
+    {
+      "epoch": 4.43,
+      "eval_loss": 0.24884916841983795,
+      "eval_runtime": 10.5407,
+      "eval_samples_per_second": 9.487,
+      "eval_steps_per_second": 9.487,
+      "step": 126000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.675095127427781e-06,
+      "loss": 0.236,
+      "step": 126005
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6740700632787127e-06,
+      "loss": 0.2544,
+      "step": 126010
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6730453020014718e-06,
+      "loss": 0.2377,
+      "step": 126015
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6720208436093705e-06,
+      "loss": 0.2608,
+      "step": 126020
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.670996688115703e-06,
+      "loss": 0.2792,
+      "step": 126025
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6699728355337645e-06,
+      "loss": 0.2465,
+      "step": 126030
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6689492858768584e-06,
+      "loss": 0.2511,
+      "step": 126035
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6679260391582713e-06,
+      "loss": 0.2508,
+      "step": 126040
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6669030953912869e-06,
+      "loss": 0.2346,
+      "step": 126045
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.665880454589186e-06,
+      "loss": 0.2555,
+      "step": 126050
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.6648581167652498e-06,
+      "loss": 0.2365,
+      "step": 126055
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6638360819327569e-06,
+      "loss": 0.2425,
+      "step": 126060
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6628143501049687e-06,
+      "loss": 0.2536,
+      "step": 126065
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.661792921295155e-06,
+      "loss": 0.2817,
+      "step": 126070
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.660771795516583e-06,
+      "loss": 0.2533,
+      "step": 126075
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.659750972782509e-06,
+      "loss": 0.2689,
+      "step": 126080
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6587304531061831e-06,
+      "loss": 0.2694,
+      "step": 126085
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6577102365008644e-06,
+      "loss": 0.2329,
+      "step": 126090
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6566903229797954e-06,
+      "loss": 0.2602,
+      "step": 126095
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6556707125562177e-06,
+      "loss": 0.2298,
+      "step": 126100
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.654651405243368e-06,
+      "loss": 0.2428,
+      "step": 126105
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.653632401054489e-06,
+      "loss": 0.2488,
+      "step": 126110
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6526137000028058e-06,
+      "loss": 0.2422,
+      "step": 126115
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6515953021015441e-06,
+      "loss": 0.2439,
+      "step": 126120
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6505772073639291e-06,
+      "loss": 0.2258,
+      "step": 126125
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6495594158031868e-06,
+      "loss": 0.2332,
+      "step": 126130
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6485419274325259e-06,
+      "loss": 0.2456,
+      "step": 126135
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6475247422651608e-06,
+      "loss": 0.2648,
+      "step": 126140
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.646507860314292e-06,
+      "loss": 0.2597,
+      "step": 126145
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6454912815931312e-06,
+      "loss": 0.2382,
+      "step": 126150
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6444750061148762e-06,
+      "loss": 0.277,
+      "step": 126155
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6434590338927191e-06,
+      "loss": 0.2687,
+      "step": 126160
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6424433649398524e-06,
+      "loss": 0.2389,
+      "step": 126165
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6414279992694682e-06,
+      "loss": 0.2455,
+      "step": 126170
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6404129368947445e-06,
+      "loss": 0.2309,
+      "step": 126175
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6393981778288658e-06,
+      "loss": 0.2248,
+      "step": 126180
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6383837220850045e-06,
+      "loss": 0.2573,
+      "step": 126185
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6373695696763391e-06,
+      "loss": 0.2841,
+      "step": 126190
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.636355720616034e-06,
+      "loss": 0.2451,
+      "step": 126195
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6353421749172453e-06,
+      "loss": 0.2418,
+      "step": 126200
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6343289325931489e-06,
+      "loss": 0.2396,
+      "step": 126205
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6333159936568893e-06,
+      "loss": 0.2439,
+      "step": 126210
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6323033581216228e-06,
+      "loss": 0.2512,
+      "step": 126215
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6312910260004949e-06,
+      "loss": 0.2396,
+      "step": 126220
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6302789973066556e-06,
+      "loss": 0.2277,
+      "step": 126225
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6292672720532365e-06,
+      "loss": 0.2511,
+      "step": 126230
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6282558502533879e-06,
+      "loss": 0.2438,
+      "step": 126235
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6272447319202272e-06,
+      "loss": 0.248,
+      "step": 126240
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6262339170668966e-06,
+      "loss": 0.245,
+      "step": 126245
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6252234057065136e-06,
+      "loss": 0.2582,
+      "step": 126250
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6242131978522007e-06,
+      "loss": 0.2563,
+      "step": 126255
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.62320329351707e-06,
+      "loss": 0.2532,
+      "step": 126260
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6221936927142412e-06,
+      "loss": 0.2486,
+      "step": 126265
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6211843954568206e-06,
+      "loss": 0.258,
+      "step": 126270
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6201754017579147e-06,
+      "loss": 0.2594,
+      "step": 126275
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6191667116306208e-06,
+      "loss": 0.2616,
+      "step": 126280
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6181583250880372e-06,
+      "loss": 0.2547,
+      "step": 126285
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6171502421432615e-06,
+      "loss": 0.2448,
+      "step": 126290
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6161424628093808e-06,
+      "loss": 0.2552,
+      "step": 126295
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.615134987099476e-06,
+      "loss": 0.2536,
+      "step": 126300
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6141278150266364e-06,
+      "loss": 0.2477,
+      "step": 126305
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6131209466039354e-06,
+      "loss": 0.2782,
+      "step": 126310
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6121143818444429e-06,
+      "loss": 0.2656,
+      "step": 126315
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6111081207612344e-06,
+      "loss": 0.2611,
+      "step": 126320
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6101021633673747e-06,
+      "loss": 0.2602,
+      "step": 126325
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6090965096759225e-06,
+      "loss": 0.2545,
+      "step": 126330
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6080911596999316e-06,
+      "loss": 0.2484,
+      "step": 126335
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.6070861134524661e-06,
+      "loss": 0.2543,
+      "step": 126340
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6060813709465688e-06,
+      "loss": 0.2743,
+      "step": 126345
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6050769321952903e-06,
+      "loss": 0.2608,
+      "step": 126350
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.604072797211667e-06,
+      "loss": 0.2474,
+      "step": 126355
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6030689660087416e-06,
+      "loss": 0.2777,
+      "step": 126360
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6020654385995482e-06,
+      "loss": 0.2635,
+      "step": 126365
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.6010622149971178e-06,
+      "loss": 0.2895,
+      "step": 126370
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.600059295214465e-06,
+      "loss": 0.2526,
+      "step": 126375
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5990566792646293e-06,
+      "loss": 0.2474,
+      "step": 126380
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5980543671606173e-06,
+      "loss": 0.2581,
+      "step": 126385
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5970523589154462e-06,
+      "loss": 0.2419,
+      "step": 126390
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5960506545421305e-06,
+      "loss": 0.2387,
+      "step": 126395
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5950492540536681e-06,
+      "loss": 0.2547,
+      "step": 126400
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5940481574630712e-06,
+      "loss": 0.2519,
+      "step": 126405
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5930473647833315e-06,
+      "loss": 0.2622,
+      "step": 126410
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5920468760274449e-06,
+      "loss": 0.2612,
+      "step": 126415
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.591046691208406e-06,
+      "loss": 0.2612,
+      "step": 126420
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5900468103391992e-06,
+      "loss": 0.2397,
+      "step": 126425
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5890472334328028e-06,
+      "loss": 0.2415,
+      "step": 126430
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5880479605022037e-06,
+      "loss": 0.2416,
+      "step": 126435
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.587048991560372e-06,
+      "loss": 0.2499,
+      "step": 126440
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.586050326620281e-06,
+      "loss": 0.2477,
+      "step": 126445
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.585051965694892e-06,
+      "loss": 0.2689,
+      "step": 126450
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5840539087971701e-06,
+      "loss": 0.2361,
+      "step": 126455
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5830561559400825e-06,
+      "loss": 0.2429,
+      "step": 126460
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5820587071365772e-06,
+      "loss": 0.2768,
+      "step": 126465
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.581061562399605e-06,
+      "loss": 0.234,
+      "step": 126470
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5800647217421166e-06,
+      "loss": 0.275,
+      "step": 126475
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5790681851770544e-06,
+      "loss": 0.2451,
+      "step": 126480
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5780719527173555e-06,
+      "loss": 0.2236,
+      "step": 126485
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5770760243759536e-06,
+      "loss": 0.2471,
+      "step": 126490
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5760804001657858e-06,
+      "loss": 0.2583,
+      "step": 126495
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.575085080099778e-06,
+      "loss": 0.2628,
+      "step": 126500
+    },
+    {
+      "epoch": 4.45,
+      "eval_loss": 0.2488119751214981,
+      "eval_runtime": 10.5384,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 126500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5740900641908501e-06,
+      "loss": 0.2759,
+      "step": 126505
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5730953524519227e-06,
+      "loss": 0.2509,
+      "step": 126510
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5721009448959184e-06,
+      "loss": 0.242,
+      "step": 126515
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.571106841535744e-06,
+      "loss": 0.2429,
+      "step": 126520
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5701130423843053e-06,
+      "loss": 0.2498,
+      "step": 126525
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5691195474545062e-06,
+      "loss": 0.2651,
+      "step": 126530
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5681263567592502e-06,
+      "loss": 0.2381,
+      "step": 126535
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5671334703114327e-06,
+      "loss": 0.2313,
+      "step": 126540
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5661408881239404e-06,
+      "loss": 0.2446,
+      "step": 126545
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5651486102096686e-06,
+      "loss": 0.2733,
+      "step": 126550
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5641566365814958e-06,
+      "loss": 0.262,
+      "step": 126555
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5631649672523036e-06,
+      "loss": 0.2356,
+      "step": 126560
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5621736022349676e-06,
+      "loss": 0.2432,
+      "step": 126565
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5611825415423637e-06,
+      "loss": 0.269,
+      "step": 126570
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5601917851873593e-06,
+      "loss": 0.2697,
+      "step": 126575
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5592013331828193e-06,
+      "loss": 0.2502,
+      "step": 126580
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5582111855415943e-06,
+      "loss": 0.23,
+      "step": 126585
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5572213422765548e-06,
+      "loss": 0.2646,
+      "step": 126590
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5562318034005457e-06,
+      "loss": 0.2634,
+      "step": 126595
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5552425689264156e-06,
+      "loss": 0.2501,
+      "step": 126600
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5542536388670065e-06,
+      "loss": 0.2284,
+      "step": 126605
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5532650132351667e-06,
+      "loss": 0.2591,
+      "step": 126610
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5522766920437275e-06,
+      "loss": 0.2355,
+      "step": 126615
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5512886753055233e-06,
+      "loss": 0.2518,
+      "step": 126620
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5503009630333799e-06,
+      "loss": 0.2802,
+      "step": 126625
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5493135552401284e-06,
+      "loss": 0.2514,
+      "step": 126630
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5483264519385864e-06,
+      "loss": 0.2342,
+      "step": 126635
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.547339653141569e-06,
+      "loss": 0.2499,
+      "step": 126640
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5463531588618878e-06,
+      "loss": 0.2665,
+      "step": 126645
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.545366969112355e-06,
+      "loss": 0.2529,
+      "step": 126650
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.544381083905777e-06,
+      "loss": 0.2425,
+      "step": 126655
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5433955032549546e-06,
+      "loss": 0.248,
+      "step": 126660
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5424102271726776e-06,
+      "loss": 0.2423,
+      "step": 126665
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.541425255671744e-06,
+      "loss": 0.2585,
+      "step": 126670
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5404405887649493e-06,
+      "loss": 0.2577,
+      "step": 126675
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5394562264650665e-06,
+      "loss": 0.2412,
+      "step": 126680
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5384721687848908e-06,
+      "loss": 0.2255,
+      "step": 126685
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.53748841573719e-06,
+      "loss": 0.2443,
+      "step": 126690
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5365049673347397e-06,
+      "loss": 0.2799,
+      "step": 126695
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5355218235903051e-06,
+      "loss": 0.2415,
+      "step": 126700
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.534538984516662e-06,
+      "loss": 0.2554,
+      "step": 126705
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5335564501265637e-06,
+      "loss": 0.2415,
+      "step": 126710
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5325742204327725e-06,
+      "loss": 0.2521,
+      "step": 126715
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5315922954480338e-06,
+      "loss": 0.2632,
+      "step": 126720
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5306106751851013e-06,
+      "loss": 0.2467,
+      "step": 126725
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5296293596567284e-06,
+      "loss": 0.2427,
+      "step": 126730
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5286483488756499e-06,
+      "loss": 0.2531,
+      "step": 126735
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5276676428546022e-06,
+      "loss": 0.2361,
+      "step": 126740
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5266872416063226e-06,
+      "loss": 0.2521,
+      "step": 126745
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.525707145143543e-06,
+      "loss": 0.2528,
+      "step": 126750
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.524727353478983e-06,
+      "loss": 0.2141,
+      "step": 126755
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5237478666253635e-06,
+      "loss": 0.2405,
+      "step": 126760
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.522768684595413e-06,
+      "loss": 0.2489,
+      "step": 126765
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5217898074018356e-06,
+      "loss": 0.2644,
+      "step": 126770
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5208112350573428e-06,
+      "loss": 0.2701,
+      "step": 126775
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.519832967574647e-06,
+      "loss": 0.2505,
+      "step": 126780
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5188550049664434e-06,
+      "loss": 0.2438,
+      "step": 126785
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5178773472454333e-06,
+      "loss": 0.2178,
+      "step": 126790
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5168999944243089e-06,
+      "loss": 0.2501,
+      "step": 126795
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.515922946515766e-06,
+      "loss": 0.2513,
+      "step": 126800
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5149462035324857e-06,
+      "loss": 0.2651,
+      "step": 126805
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5139697654871527e-06,
+      "loss": 0.2726,
+      "step": 126810
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5129936323924426e-06,
+      "loss": 0.2419,
+      "step": 126815
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5120178042610317e-06,
+      "loss": 0.2407,
+      "step": 126820
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.511042281105593e-06,
+      "loss": 0.2472,
+      "step": 126825
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5100670629387858e-06,
+      "loss": 0.2563,
+      "step": 126830
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5090921497732808e-06,
+      "loss": 0.2501,
+      "step": 126835
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5081175416217313e-06,
+      "loss": 0.2744,
+      "step": 126840
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.507143238496797e-06,
+      "loss": 0.2663,
+      "step": 126845
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5061692404111232e-06,
+      "loss": 0.2494,
+      "step": 126850
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5051955473773556e-06,
+      "loss": 0.2423,
+      "step": 126855
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5042221594081447e-06,
+      "loss": 0.2425,
+      "step": 126860
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.503249076516125e-06,
+      "loss": 0.2373,
+      "step": 126865
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.502276298713931e-06,
+      "loss": 0.2575,
+      "step": 126870
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5013038260141887e-06,
+      "loss": 0.239,
+      "step": 126875
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5003316584295352e-06,
+      "loss": 0.2341,
+      "step": 126880
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.499359795972588e-06,
+      "loss": 0.2364,
+      "step": 126885
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.4983882386559622e-06,
+      "loss": 0.2315,
+      "step": 126890
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.4974169864922782e-06,
+      "loss": 0.2696,
+      "step": 126895
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.4964460394941482e-06,
+      "loss": 0.2497,
+      "step": 126900
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.4954753976741786e-06,
+      "loss": 0.2626,
+      "step": 126905
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.494505061044965e-06,
+      "loss": 0.2441,
+      "step": 126910
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.493535029619117e-06,
+      "loss": 0.2493,
+      "step": 126915
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4925653034092267e-06,
+      "loss": 0.2521,
+      "step": 126920
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.491595882427882e-06,
+      "loss": 0.2596,
+      "step": 126925
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4906267666876695e-06,
+      "loss": 0.2547,
+      "step": 126930
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4896579562011793e-06,
+      "loss": 0.256,
+      "step": 126935
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4886894509809845e-06,
+      "loss": 0.2824,
+      "step": 126940
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4877212510396615e-06,
+      "loss": 0.2579,
+      "step": 126945
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4867533563897807e-06,
+      "loss": 0.259,
+      "step": 126950
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4857857670439152e-06,
+      "loss": 0.2363,
+      "step": 126955
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4848184830146272e-06,
+      "loss": 0.2565,
+      "step": 126960
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4838515043144734e-06,
+      "loss": 0.2329,
+      "step": 126965
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.482884830956005e-06,
+      "loss": 0.254,
+      "step": 126970
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.481918462951784e-06,
+      "loss": 0.2786,
+      "step": 126975
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4809524003143532e-06,
+      "loss": 0.2525,
+      "step": 126980
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4799866430562554e-06,
+      "loss": 0.2582,
+      "step": 126985
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4790211911900277e-06,
+      "loss": 0.2658,
+      "step": 126990
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.47805604472821e-06,
+      "loss": 0.2376,
+      "step": 126995
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4770912036833313e-06,
+      "loss": 0.2564,
+      "step": 127000
+    },
+    {
+      "epoch": 4.47,
+      "eval_loss": 0.2487480491399765,
+      "eval_runtime": 10.5483,
+      "eval_samples_per_second": 9.48,
+      "eval_steps_per_second": 9.48,
+      "step": 127000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.476126668067923e-06,
+      "loss": 0.2605,
+      "step": 127005
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.475162437894506e-06,
+      "loss": 0.2506,
+      "step": 127010
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4741985131756037e-06,
+      "loss": 0.2621,
+      "step": 127015
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4732348939237307e-06,
+      "loss": 0.2506,
+      "step": 127020
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4722715801513937e-06,
+      "loss": 0.2394,
+      "step": 127025
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4713085718711078e-06,
+      "loss": 0.2506,
+      "step": 127030
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4703458690953742e-06,
+      "loss": 0.2659,
+      "step": 127035
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.469383471836694e-06,
+      "loss": 0.2494,
+      "step": 127040
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.468421380107557e-06,
+      "loss": 0.238,
+      "step": 127045
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4674595939204643e-06,
+      "loss": 0.2531,
+      "step": 127050
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.466498113287898e-06,
+      "loss": 0.2549,
+      "step": 127055
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4655369382223478e-06,
+      "loss": 0.2312,
+      "step": 127060
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4645760687362842e-06,
+      "loss": 0.2283,
+      "step": 127065
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4636155048421974e-06,
+      "loss": 0.2538,
+      "step": 127070
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4626552465525495e-06,
+      "loss": 0.2794,
+      "step": 127075
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.461695293879814e-06,
+      "loss": 0.274,
+      "step": 127080
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4607356468364475e-06,
+      "loss": 0.2674,
+      "step": 127085
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4597763054349179e-06,
+      "loss": 0.2542,
+      "step": 127090
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4588172696876817e-06,
+      "loss": 0.261,
+      "step": 127095
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4578585396071904e-06,
+      "loss": 0.2515,
+      "step": 127100
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4569001152058837e-06,
+      "loss": 0.2512,
+      "step": 127105
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4559419964962156e-06,
+      "loss": 0.272,
+      "step": 127110
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4549841834906292e-06,
+      "loss": 0.235,
+      "step": 127115
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4540266762015531e-06,
+      "loss": 0.2546,
+      "step": 127120
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.453069474641422e-06,
+      "loss": 0.239,
+      "step": 127125
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4521125788226703e-06,
+      "loss": 0.2469,
+      "step": 127130
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4511559887577158e-06,
+      "loss": 0.2428,
+      "step": 127135
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4501997044589793e-06,
+      "loss": 0.2392,
+      "step": 127140
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.449243725938884e-06,
+      "loss": 0.2742,
+      "step": 127145
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4482880532098369e-06,
+      "loss": 0.2383,
+      "step": 127150
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4473326862842474e-06,
+      "loss": 0.2642,
+      "step": 127155
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4463776251745191e-06,
+      "loss": 0.2464,
+      "step": 127160
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4454228698930567e-06,
+      "loss": 0.2648,
+      "step": 127165
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4444684204522551e-06,
+      "loss": 0.2388,
+      "step": 127170
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4435142768645106e-06,
+      "loss": 0.2547,
+      "step": 127175
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4425604391422043e-06,
+      "loss": 0.265,
+      "step": 127180
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4416069072977266e-06,
+      "loss": 0.2466,
+      "step": 127185
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.4406536813434624e-06,
+      "loss": 0.2674,
+      "step": 127190
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4397007612917817e-06,
+      "loss": 0.2396,
+      "step": 127195
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4387481471550557e-06,
+      "loss": 0.2615,
+      "step": 127200
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4377958389456604e-06,
+      "loss": 0.2407,
+      "step": 127205
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4368438366759606e-06,
+      "loss": 0.2464,
+      "step": 127210
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.435892140358308e-06,
+      "loss": 0.2432,
+      "step": 127215
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4349407500050731e-06,
+      "loss": 0.2628,
+      "step": 127220
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4339896656285989e-06,
+      "loss": 0.2422,
+      "step": 127225
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4330388872412392e-06,
+      "loss": 0.2442,
+      "step": 127230
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4320884148553427e-06,
+      "loss": 0.2485,
+      "step": 127235
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.431138248483238e-06,
+      "loss": 0.2313,
+      "step": 127240
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4301883881372768e-06,
+      "loss": 0.2543,
+      "step": 127245
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4292388338297852e-06,
+      "loss": 0.2537,
+      "step": 127250
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.428289585573092e-06,
+      "loss": 0.2665,
+      "step": 127255
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4273406433795267e-06,
+      "loss": 0.223,
+      "step": 127260
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4263920072614068e-06,
+      "loss": 0.2692,
+      "step": 127265
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4254436772310503e-06,
+      "loss": 0.2491,
+      "step": 127270
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4244956533007726e-06,
+      "loss": 0.2294,
+      "step": 127275
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.42354793548288e-06,
+      "loss": 0.2604,
+      "step": 127280
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4226005237896827e-06,
+      "loss": 0.2476,
+      "step": 127285
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.421653418233479e-06,
+      "loss": 0.254,
+      "step": 127290
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4207066188265645e-06,
+      "loss": 0.256,
+      "step": 127295
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4197601255812376e-06,
+      "loss": 0.2693,
+      "step": 127300
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.418813938509786e-06,
+      "loss": 0.2401,
+      "step": 127305
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4178680576244967e-06,
+      "loss": 0.234,
+      "step": 127310
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4169224829376438e-06,
+      "loss": 0.2692,
+      "step": 127315
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4159772144615142e-06,
+      "loss": 0.2418,
+      "step": 127320
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4150322522083787e-06,
+      "loss": 0.227,
+      "step": 127325
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4140875961905026e-06,
+      "loss": 0.2393,
+      "step": 127330
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4131432464201565e-06,
+      "loss": 0.2483,
+      "step": 127335
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4121992029096031e-06,
+      "loss": 0.2745,
+      "step": 127340
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4112554656710992e-06,
+      "loss": 0.2644,
+      "step": 127345
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4103120347168958e-06,
+      "loss": 0.2589,
+      "step": 127350
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4093689100592416e-06,
+      "loss": 0.2414,
+      "step": 127355
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4084260917103881e-06,
+      "loss": 0.2681,
+      "step": 127360
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4074835796825753e-06,
+      "loss": 0.2525,
+      "step": 127365
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.406541373988038e-06,
+      "loss": 0.2467,
+      "step": 127370
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4055994746390132e-06,
+      "loss": 0.2416,
+      "step": 127375
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4046578816477306e-06,
+      "loss": 0.2252,
+      "step": 127380
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4037165950264136e-06,
+      "loss": 0.2356,
+      "step": 127385
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4027756147872856e-06,
+      "loss": 0.2436,
+      "step": 127390
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4018349409425674e-06,
+      "loss": 0.2723,
+      "step": 127395
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.4008945735044716e-06,
+      "loss": 0.2374,
+      "step": 127400
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3999545124852077e-06,
+      "loss": 0.2445,
+      "step": 127405
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.399014757896977e-06,
+      "loss": 0.2511,
+      "step": 127410
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3980753097519894e-06,
+      "loss": 0.2439,
+      "step": 127415
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3971361680624406e-06,
+      "loss": 0.2561,
+      "step": 127420
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3961973328405236e-06,
+      "loss": 0.2491,
+      "step": 127425
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3952588040984232e-06,
+      "loss": 0.2633,
+      "step": 127430
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.394320581848338e-06,
+      "loss": 0.2349,
+      "step": 127435
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3933826661024386e-06,
+      "loss": 0.2662,
+      "step": 127440
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3924450568729098e-06,
+      "loss": 0.2524,
+      "step": 127445
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3915077541719224e-06,
+      "loss": 0.2417,
+      "step": 127450
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.39057075801165e-06,
+      "loss": 0.2401,
+      "step": 127455
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.389634068404258e-06,
+      "loss": 0.2373,
+      "step": 127460
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3886976853619087e-06,
+      "loss": 0.2342,
+      "step": 127465
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.3877616088967565e-06,
+      "loss": 0.2402,
+      "step": 127470
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.386825839020961e-06,
+      "loss": 0.2766,
+      "step": 127475
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3858903757466706e-06,
+      "loss": 0.2329,
+      "step": 127480
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3849552190860259e-06,
+      "loss": 0.2594,
+      "step": 127485
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.384020369051181e-06,
+      "loss": 0.2463,
+      "step": 127490
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3830858256542623e-06,
+      "loss": 0.2369,
+      "step": 127495
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3821515889074154e-06,
+      "loss": 0.2386,
+      "step": 127500
+    },
+    {
+      "epoch": 4.49,
+      "eval_loss": 0.2487904578447342,
+      "eval_runtime": 10.5517,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 127500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3812176588227615e-06,
+      "loss": 0.2426,
+      "step": 127505
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3802840354124353e-06,
+      "loss": 0.2382,
+      "step": 127510
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3793507186885546e-06,
+      "loss": 0.2469,
+      "step": 127515
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3784177086632377e-06,
+      "loss": 0.2207,
+      "step": 127520
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3774850053485944e-06,
+      "loss": 0.2492,
+      "step": 127525
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3765526087567455e-06,
+      "loss": 0.2256,
+      "step": 127530
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3756205188997922e-06,
+      "loss": 0.2693,
+      "step": 127535
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3746887357898392e-06,
+      "loss": 0.2465,
+      "step": 127540
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3737572594389765e-06,
+      "loss": 0.254,
+      "step": 127545
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3728260898593082e-06,
+      "loss": 0.2624,
+      "step": 127550
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.371895227062925e-06,
+      "loss": 0.2333,
+      "step": 127555
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3709646710619113e-06,
+      "loss": 0.2297,
+      "step": 127560
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3700344218683437e-06,
+      "loss": 0.2366,
+      "step": 127565
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3691044794943125e-06,
+      "loss": 0.2414,
+      "step": 127570
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.368174843951886e-06,
+      "loss": 0.2431,
+      "step": 127575
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.367245515253135e-06,
+      "loss": 0.2397,
+      "step": 127580
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3663164934101219e-06,
+      "loss": 0.2398,
+      "step": 127585
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3653877784349178e-06,
+      "loss": 0.2439,
+      "step": 127590
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.364459370339577e-06,
+      "loss": 0.2407,
+      "step": 127595
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3635312691361508e-06,
+      "loss": 0.2473,
+      "step": 127600
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3626034748366962e-06,
+      "loss": 0.2755,
+      "step": 127605
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3616759874532592e-06,
+      "loss": 0.258,
+      "step": 127610
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.360748806997883e-06,
+      "loss": 0.2314,
+      "step": 127615
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3598219334826022e-06,
+      "loss": 0.238,
+      "step": 127620
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3588953669194548e-06,
+      "loss": 0.2529,
+      "step": 127625
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3579691073204726e-06,
+      "loss": 0.2615,
+      "step": 127630
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3570431546976792e-06,
+      "loss": 0.244,
+      "step": 127635
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3561175090630984e-06,
+      "loss": 0.2943,
+      "step": 127640
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3551921704287513e-06,
+      "loss": 0.237,
+      "step": 127645
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.354267138806653e-06,
+      "loss": 0.2492,
+      "step": 127650
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.353342414208808e-06,
+      "loss": 0.2577,
+      "step": 127655
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3524179966472317e-06,
+      "loss": 0.2467,
+      "step": 127660
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3514938861339199e-06,
+      "loss": 0.2468,
+      "step": 127665
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.350570082680877e-06,
+      "loss": 0.236,
+      "step": 127670
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.349646586300099e-06,
+      "loss": 0.251,
+      "step": 127675
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3487233970035679e-06,
+      "loss": 0.2442,
+      "step": 127680
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3478005148032824e-06,
+      "loss": 0.2409,
+      "step": 127685
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3468779397112191e-06,
+      "loss": 0.2736,
+      "step": 127690
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3459556717393574e-06,
+      "loss": 0.2542,
+      "step": 127695
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.345033710899668e-06,
+      "loss": 0.2343,
+      "step": 127700
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3441120572041305e-06,
+      "loss": 0.244,
+      "step": 127705
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3431907106647073e-06,
+      "loss": 0.2564,
+      "step": 127710
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3422696712933642e-06,
+      "loss": 0.242,
+      "step": 127715
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3413489391020551e-06,
+      "loss": 0.2257,
+      "step": 127720
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.34042851410274e-06,
+      "loss": 0.269,
+      "step": 127725
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3395083963073707e-06,
+      "loss": 0.2503,
+      "step": 127730
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3385885857278874e-06,
+      "loss": 0.2621,
+      "step": 127735
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3376690823762417e-06,
+      "loss": 0.2473,
+      "step": 127740
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3367498862643685e-06,
+      "loss": 0.2326,
+      "step": 127745
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3358309974042027e-06,
+      "loss": 0.2687,
+      "step": 127750
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3349124158076736e-06,
+      "loss": 0.2496,
+      "step": 127755
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.3339941414867162e-06,
+      "loss": 0.2329,
+      "step": 127760
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.333076174453246e-06,
+      "loss": 0.2658,
+      "step": 127765
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3321585147191812e-06,
+      "loss": 0.2538,
+      "step": 127770
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.33124116229644e-06,
+      "loss": 0.2431,
+      "step": 127775
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3303241171969406e-06,
+      "loss": 0.2353,
+      "step": 127780
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3294073794325818e-06,
+      "loss": 0.2328,
+      "step": 127785
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3284909490152681e-06,
+      "loss": 0.2453,
+      "step": 127790
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3275748259568955e-06,
+      "loss": 0.2366,
+      "step": 127795
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3266590102693655e-06,
+      "loss": 0.254,
+      "step": 127800
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3257435019645687e-06,
+      "loss": 0.2509,
+      "step": 127805
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3248283010543927e-06,
+      "loss": 0.2444,
+      "step": 127810
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3240963616583557e-06,
+      "loss": 0.2366,
+      "step": 127815
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3231817140884339e-06,
+      "loss": 0.2728,
+      "step": 127820
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3222673739463926e-06,
+      "loss": 0.2542,
+      "step": 127825
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3213533412441054e-06,
+      "loss": 0.2533,
+      "step": 127830
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3204396159934406e-06,
+      "loss": 0.2539,
+      "step": 127835
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.319526198206264e-06,
+      "loss": 0.251,
+      "step": 127840
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3186130878944296e-06,
+      "loss": 0.2461,
+      "step": 127845
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3177002850698006e-06,
+      "loss": 0.2506,
+      "step": 127850
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3167877897442227e-06,
+      "loss": 0.2423,
+      "step": 127855
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.315875601929553e-06,
+      "loss": 0.2454,
+      "step": 127860
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3149637216376243e-06,
+      "loss": 0.256,
+      "step": 127865
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3140521488802876e-06,
+      "loss": 0.2749,
+      "step": 127870
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3131408836693726e-06,
+      "loss": 0.2542,
+      "step": 127875
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3122299260167143e-06,
+      "loss": 0.2321,
+      "step": 127880
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3113192759341363e-06,
+      "loss": 0.2417,
+      "step": 127885
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3104089334334686e-06,
+      "loss": 0.2423,
+      "step": 127890
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3094988985265287e-06,
+      "loss": 0.2395,
+      "step": 127895
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.308589171225133e-06,
+      "loss": 0.2731,
+      "step": 127900
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.307679751541091e-06,
+      "loss": 0.2482,
+      "step": 127905
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3067706394862156e-06,
+      "loss": 0.2195,
+      "step": 127910
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3058618350723085e-06,
+      "loss": 0.2456,
+      "step": 127915
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3049533383111685e-06,
+      "loss": 0.251,
+      "step": 127920
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3040451492145972e-06,
+      "loss": 0.2635,
+      "step": 127925
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3031372677943827e-06,
+      "loss": 0.2577,
+      "step": 127930
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3022296940623153e-06,
+      "loss": 0.2534,
+      "step": 127935
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3013224280301745e-06,
+      "loss": 0.2489,
+      "step": 127940
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.3004154697097454e-06,
+      "loss": 0.2611,
+      "step": 127945
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2995088191128046e-06,
+      "loss": 0.2496,
+      "step": 127950
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2986024762511235e-06,
+      "loss": 0.2583,
+      "step": 127955
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2976964411364644e-06,
+      "loss": 0.2547,
+      "step": 127960
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.296790713780599e-06,
+      "loss": 0.2385,
+      "step": 127965
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2958852941952843e-06,
+      "loss": 0.2369,
+      "step": 127970
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2949801823922802e-06,
+      "loss": 0.2403,
+      "step": 127975
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2940753783833303e-06,
+      "loss": 0.2501,
+      "step": 127980
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2931708821801946e-06,
+      "loss": 0.2313,
+      "step": 127985
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.292266693794611e-06,
+      "loss": 0.2635,
+      "step": 127990
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2913628132383226e-06,
+      "loss": 0.2462,
+      "step": 127995
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2904592405230565e-06,
+      "loss": 0.2526,
+      "step": 128000
+    },
+    {
+      "epoch": 4.5,
+      "eval_loss": 0.24867750704288483,
+      "eval_runtime": 10.5298,
+      "eval_samples_per_second": 9.497,
+      "eval_steps_per_second": 9.497,
+      "step": 128000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2895559756605586e-06,
+      "loss": 0.255,
+      "step": 128005
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2886530186625473e-06,
+      "loss": 0.2401,
+      "step": 128010
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2877503695407495e-06,
+      "loss": 0.2524,
+      "step": 128015
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.286848028306889e-06,
+      "loss": 0.2278,
+      "step": 128020
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.285945994972676e-06,
+      "loss": 0.2532,
+      "step": 128025
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2850442695498287e-06,
+      "loss": 0.2428,
+      "step": 128030
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2841428520500492e-06,
+      "loss": 0.2495,
+      "step": 128035
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.28324174248505e-06,
+      "loss": 0.2322,
+      "step": 128040
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.282340940866525e-06,
+      "loss": 0.2704,
+      "step": 128045
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2814404472061752e-06,
+      "loss": 0.2555,
+      "step": 128050
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2805402615156837e-06,
+      "loss": 0.2477,
+      "step": 128055
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2796403838067517e-06,
+      "loss": 0.2378,
+      "step": 128060
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2787408140910535e-06,
+      "loss": 0.2618,
+      "step": 128065
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2778415523802767e-06,
+      "loss": 0.2622,
+      "step": 128070
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2769425986860873e-06,
+      "loss": 0.253,
+      "step": 128075
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2760439530201645e-06,
+      "loss": 0.2437,
+      "step": 128080
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.27514561539418e-06,
+      "loss": 0.2546,
+      "step": 128085
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2742475858197961e-06,
+      "loss": 0.2535,
+      "step": 128090
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2733498643086677e-06,
+      "loss": 0.2389,
+      "step": 128095
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.272452450872458e-06,
+      "loss": 0.2523,
+      "step": 128100
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2715553455228151e-06,
+      "loss": 0.243,
+      "step": 128105
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2706585482713889e-06,
+      "loss": 0.2398,
+      "step": 128110
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2697620591298199e-06,
+      "loss": 0.2464,
+      "step": 128115
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.268865878109754e-06,
+      "loss": 0.243,
+      "step": 128120
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2679700052228265e-06,
+      "loss": 0.26,
+      "step": 128125
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2670744404806645e-06,
+      "loss": 0.2382,
+      "step": 128130
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2661791838949028e-06,
+      "loss": 0.2534,
+      "step": 128135
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2652842354771627e-06,
+      "loss": 0.2408,
+      "step": 128140
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2643895952390682e-06,
+      "loss": 0.2671,
+      "step": 128145
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.263495263192227e-06,
+      "loss": 0.2214,
+      "step": 128150
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.26260123934826e-06,
+      "loss": 0.2392,
+      "step": 128155
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2617075237187748e-06,
+      "loss": 0.2451,
+      "step": 128160
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.260814116315373e-06,
+      "loss": 0.2509,
+      "step": 128165
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2599210171496484e-06,
+      "loss": 0.2544,
+      "step": 128170
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2590282262332109e-06,
+      "loss": 0.2544,
+      "step": 128175
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2581357435776458e-06,
+      "loss": 0.2139,
+      "step": 128180
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2572435691945355e-06,
+      "loss": 0.2569,
+      "step": 128185
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2563517030954735e-06,
+      "loss": 0.2502,
+      "step": 128190
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2554601452920394e-06,
+      "loss": 0.2751,
+      "step": 128195
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2545688957958046e-06,
+      "loss": 0.2483,
+      "step": 128200
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2536779546183485e-06,
+      "loss": 0.2366,
+      "step": 128205
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2527873217712287e-06,
+      "loss": 0.2486,
+      "step": 128210
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2518969972660194e-06,
+      "loss": 0.2633,
+      "step": 128215
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2510069811142749e-06,
+      "loss": 0.2338,
+      "step": 128220
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2501172733275557e-06,
+      "loss": 0.2527,
+      "step": 128225
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2492278739174079e-06,
+      "loss": 0.2317,
+      "step": 128230
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.248338782895389e-06,
+      "loss": 0.2424,
+      "step": 128235
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2474500002730316e-06,
+      "loss": 0.2638,
+      "step": 128240
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2465615260618874e-06,
+      "loss": 0.2505,
+      "step": 128245
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.245673360273486e-06,
+      "loss": 0.2524,
+      "step": 128250
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2447855029193628e-06,
+      "loss": 0.2336,
+      "step": 128255
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2438979540110445e-06,
+      "loss": 0.2408,
+      "step": 128260
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2430107135600527e-06,
+      "loss": 0.249,
+      "step": 128265
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2421237815779112e-06,
+      "loss": 0.2256,
+      "step": 128270
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.241237158076139e-06,
+      "loss": 0.2579,
+      "step": 128275
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.240350843066243e-06,
+      "loss": 0.2842,
+      "step": 128280
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2394648365597283e-06,
+      "loss": 0.2446,
+      "step": 128285
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2385791385681105e-06,
+      "loss": 0.2459,
+      "step": 128290
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2376937491028778e-06,
+      "loss": 0.2724,
+      "step": 128295
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2368086681755347e-06,
+      "loss": 0.2479,
+      "step": 128300
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2359238957975695e-06,
+      "loss": 0.2443,
+      "step": 128305
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2350394319804726e-06,
+      "loss": 0.2539,
+      "step": 128310
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2341552767357296e-06,
+      "loss": 0.2659,
+      "step": 128315
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.2332714300748144e-06,
+      "loss": 0.267,
+      "step": 128320
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.232387892009207e-06,
+      "loss": 0.283,
+      "step": 128325
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2315046625503785e-06,
+      "loss": 0.2495,
+      "step": 128330
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2306217417098004e-06,
+      "loss": 0.2439,
+      "step": 128335
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2297391294989357e-06,
+      "loss": 0.2537,
+      "step": 128340
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2288568259292394e-06,
+      "loss": 0.2471,
+      "step": 128345
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2279748310121742e-06,
+      "loss": 0.246,
+      "step": 128350
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2270931447591838e-06,
+      "loss": 0.2501,
+      "step": 128355
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2262117671817286e-06,
+      "loss": 0.2576,
+      "step": 128360
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.225330698291241e-06,
+      "loss": 0.2712,
+      "step": 128365
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2244499380991704e-06,
+      "loss": 0.2883,
+      "step": 128370
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.223569486616949e-06,
+      "loss": 0.2656,
+      "step": 128375
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2226893438560038e-06,
+      "loss": 0.2379,
+      "step": 128380
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2218095098277733e-06,
+      "loss": 0.2576,
+      "step": 128385
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2209299845436727e-06,
+      "loss": 0.2686,
+      "step": 128390
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2200507680151268e-06,
+      "loss": 0.2639,
+      "step": 128395
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2191718602535484e-06,
+      "loss": 0.2408,
+      "step": 128400
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2182932612703507e-06,
+      "loss": 0.2649,
+      "step": 128405
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.217414971076941e-06,
+      "loss": 0.2392,
+      "step": 128410
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2165369896847302e-06,
+      "loss": 0.2655,
+      "step": 128415
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2156593171051055e-06,
+      "loss": 0.2692,
+      "step": 128420
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2147819533494753e-06,
+      "loss": 0.2493,
+      "step": 128425
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2139048984292273e-06,
+      "loss": 0.2598,
+      "step": 128430
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2130281523557468e-06,
+      "loss": 0.2561,
+      "step": 128435
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2121517151404193e-06,
+      "loss": 0.2514,
+      "step": 128440
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2112755867946246e-06,
+      "loss": 0.2554,
+      "step": 128445
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2103997673297423e-06,
+      "loss": 0.2531,
+      "step": 128450
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2095242567571358e-06,
+      "loss": 0.2466,
+      "step": 128455
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2086490550881818e-06,
+      "loss": 0.2515,
+      "step": 128460
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2077741623342354e-06,
+      "loss": 0.2532,
+      "step": 128465
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2068995785066678e-06,
+      "loss": 0.282,
+      "step": 128470
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2060253036168284e-06,
+      "loss": 0.2421,
+      "step": 128475
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2051513376760664e-06,
+      "loss": 0.2655,
+      "step": 128480
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2042776806957368e-06,
+      "loss": 0.2595,
+      "step": 128485
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2034043326871775e-06,
+      "loss": 0.2237,
+      "step": 128490
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2025312936617295e-06,
+      "loss": 0.2359,
+      "step": 128495
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.201658563630731e-06,
+      "loss": 0.2484,
+      "step": 128500
+    },
+    {
+      "epoch": 4.52,
+      "eval_loss": 0.24868270754814148,
+      "eval_runtime": 10.5541,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 9.475,
+      "step": 128500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2007861426055146e-06,
+      "loss": 0.2519,
+      "step": 128505
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1999140305974043e-06,
+      "loss": 0.243,
+      "step": 128510
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1990422276177221e-06,
+      "loss": 0.2495,
+      "step": 128515
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1981707336777893e-06,
+      "loss": 0.2229,
+      "step": 128520
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1972995487889299e-06,
+      "loss": 0.255,
+      "step": 128525
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1964286729624492e-06,
+      "loss": 0.2373,
+      "step": 128530
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1955581062096488e-06,
+      "loss": 0.2358,
+      "step": 128535
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.194687848541845e-06,
+      "loss": 0.2546,
+      "step": 128540
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1938178999703287e-06,
+      "loss": 0.2521,
+      "step": 128545
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1929482605064018e-06,
+      "loss": 0.2572,
+      "step": 128550
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1920789301613443e-06,
+      "loss": 0.2352,
+      "step": 128555
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1912099089464584e-06,
+      "loss": 0.2382,
+      "step": 128560
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.190341196873021e-06,
+      "loss": 0.2519,
+      "step": 128565
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1894727939523093e-06,
+      "loss": 0.2644,
+      "step": 128570
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1886047001956003e-06,
+      "loss": 0.2342,
+      "step": 128575
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1879104477958952e-06,
+      "loss": 0.2639,
+      "step": 128580
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.187042910562794e-06,
+      "loss": 0.2323,
+      "step": 128585
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1861756825252534e-06,
+      "loss": 0.2592,
+      "step": 128590
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.185308763694526e-06,
+      "loss": 0.2256,
+      "step": 128595
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1844421540818662e-06,
+      "loss": 0.2473,
+      "step": 128600
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1835758536985347e-06,
+      "loss": 0.2467,
+      "step": 128605
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.1827098625557698e-06,
+      "loss": 0.2349,
+      "step": 128610
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1818441806648262e-06,
+      "loss": 0.2594,
+      "step": 128615
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1809788080369393e-06,
+      "loss": 0.2482,
+      "step": 128620
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1801137446833422e-06,
+      "loss": 0.2539,
+      "step": 128625
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1792489906152726e-06,
+      "loss": 0.2359,
+      "step": 128630
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1783845458439552e-06,
+      "loss": 0.2332,
+      "step": 128635
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.177520410380617e-06,
+      "loss": 0.2727,
+      "step": 128640
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.176656584236474e-06,
+      "loss": 0.246,
+      "step": 128645
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.175793067422748e-06,
+      "loss": 0.2481,
+      "step": 128650
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1749298599506464e-06,
+      "loss": 0.2231,
+      "step": 128655
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.174066961831377e-06,
+      "loss": 0.2451,
+      "step": 128660
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1732043730761478e-06,
+      "loss": 0.2431,
+      "step": 128665
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1723420936961576e-06,
+      "loss": 0.2217,
+      "step": 128670
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1714801237026036e-06,
+      "loss": 0.252,
+      "step": 128675
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1706184631066737e-06,
+      "loss": 0.2514,
+      "step": 128680
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1697571119195643e-06,
+      "loss": 0.229,
+      "step": 128685
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1688960701524503e-06,
+      "loss": 0.2412,
+      "step": 128690
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1680353378165172e-06,
+      "loss": 0.2556,
+      "step": 128695
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.167174914922936e-06,
+      "loss": 0.2428,
+      "step": 128700
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1663148014828873e-06,
+      "loss": 0.2401,
+      "step": 128705
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1654549975075314e-06,
+      "loss": 0.2426,
+      "step": 128710
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1645955030080313e-06,
+      "loss": 0.2618,
+      "step": 128715
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1637363179955535e-06,
+      "loss": 0.2434,
+      "step": 128720
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1628774424812528e-06,
+      "loss": 0.2437,
+      "step": 128725
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1620188764762785e-06,
+      "loss": 0.2514,
+      "step": 128730
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1611606199917773e-06,
+      "loss": 0.2378,
+      "step": 128735
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1603026730388962e-06,
+      "loss": 0.2316,
+      "step": 128740
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1594450356287727e-06,
+      "loss": 0.2539,
+      "step": 128745
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.158587707772546e-06,
+      "loss": 0.2615,
+      "step": 128750
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1577306894813483e-06,
+      "loss": 0.2508,
+      "step": 128755
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1568739807662988e-06,
+      "loss": 0.2585,
+      "step": 128760
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1560175816385298e-06,
+      "loss": 0.2464,
+      "step": 128765
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.155161492109158e-06,
+      "loss": 0.253,
+      "step": 128770
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.154305712189302e-06,
+      "loss": 0.2679,
+      "step": 128775
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1534502418900667e-06,
+      "loss": 0.2551,
+      "step": 128780
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.152595081222571e-06,
+      "loss": 0.2369,
+      "step": 128785
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1517402301979092e-06,
+      "loss": 0.233,
+      "step": 128790
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.150885688827183e-06,
+      "loss": 0.2555,
+      "step": 128795
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1500314571214921e-06,
+      "loss": 0.2557,
+      "step": 128800
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.149177535091922e-06,
+      "loss": 0.2527,
+      "step": 128805
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1483239227495668e-06,
+      "loss": 0.2333,
+      "step": 128810
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1474706201055007e-06,
+      "loss": 0.2304,
+      "step": 128815
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.146617627170815e-06,
+      "loss": 0.2461,
+      "step": 128820
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1457649439565733e-06,
+      "loss": 0.2539,
+      "step": 128825
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1449125704738606e-06,
+      "loss": 0.2542,
+      "step": 128830
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1440605067337295e-06,
+      "loss": 0.2524,
+      "step": 128835
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1432087527472574e-06,
+      "loss": 0.2413,
+      "step": 128840
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1423573085254963e-06,
+      "loss": 0.2529,
+      "step": 128845
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1415061740795014e-06,
+      "loss": 0.2289,
+      "step": 128850
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1406553494203247e-06,
+      "loss": 0.2514,
+      "step": 128855
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1398048345590162e-06,
+      "loss": 0.255,
+      "step": 128860
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1389546295066166e-06,
+      "loss": 0.2527,
+      "step": 128865
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1381047342741674e-06,
+      "loss": 0.2329,
+      "step": 128870
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1372551488726985e-06,
+      "loss": 0.2546,
+      "step": 128875
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1364058733132483e-06,
+      "loss": 0.2448,
+      "step": 128880
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1355569076068411e-06,
+      "loss": 0.2494,
+      "step": 128885
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.134708251764502e-06,
+      "loss": 0.2349,
+      "step": 128890
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.1338599057972437e-06,
+      "loss": 0.245,
+      "step": 128895
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1330118697160914e-06,
+      "loss": 0.2504,
+      "step": 128900
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1321641435320495e-06,
+      "loss": 0.2522,
+      "step": 128905
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1313167272561238e-06,
+      "loss": 0.242,
+      "step": 128910
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1304696208993215e-06,
+      "loss": 0.2493,
+      "step": 128915
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1296228244726426e-06,
+      "loss": 0.2431,
+      "step": 128920
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1287763379870809e-06,
+      "loss": 0.2553,
+      "step": 128925
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.127930161453622e-06,
+      "loss": 0.2363,
+      "step": 128930
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1270842948832599e-06,
+      "loss": 0.2522,
+      "step": 128935
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1262387382869778e-06,
+      "loss": 0.254,
+      "step": 128940
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1253934916757496e-06,
+      "loss": 0.2429,
+      "step": 128945
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1245485550605534e-06,
+      "loss": 0.2606,
+      "step": 128950
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1237039284523605e-06,
+      "loss": 0.2483,
+      "step": 128955
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.122859611862137e-06,
+      "loss": 0.2245,
+      "step": 128960
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1220156053008468e-06,
+      "loss": 0.2884,
+      "step": 128965
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1211719087794447e-06,
+      "loss": 0.2425,
+      "step": 128970
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1203285223088915e-06,
+      "loss": 0.2316,
+      "step": 128975
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.119485445900137e-06,
+      "loss": 0.2686,
+      "step": 128980
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1186426795641192e-06,
+      "loss": 0.2521,
+      "step": 128985
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1178002233117934e-06,
+      "loss": 0.2504,
+      "step": 128990
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.11695807715409e-06,
+      "loss": 0.2675,
+      "step": 128995
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1161162411019471e-06,
+      "loss": 0.2357,
+      "step": 129000
+    },
+    {
+      "epoch": 4.54,
+      "eval_loss": 0.24872316420078278,
+      "eval_runtime": 10.5448,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 129000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.115274715166298e-06,
+      "loss": 0.2395,
+      "step": 129005
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1144334993580613e-06,
+      "loss": 0.2523,
+      "step": 129010
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1135925936881674e-06,
+      "loss": 0.2347,
+      "step": 129015
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1127519981675299e-06,
+      "loss": 0.2594,
+      "step": 129020
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1119117128070649e-06,
+      "loss": 0.2511,
+      "step": 129025
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.111071737617686e-06,
+      "loss": 0.2486,
+      "step": 129030
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1102320726102983e-06,
+      "loss": 0.2397,
+      "step": 129035
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1093927177957985e-06,
+      "loss": 0.241,
+      "step": 129040
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1085536731850947e-06,
+      "loss": 0.2557,
+      "step": 129045
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.10771493878907e-06,
+      "loss": 0.2746,
+      "step": 129050
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1068765146186266e-06,
+      "loss": 0.2543,
+      "step": 129055
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1060384006846475e-06,
+      "loss": 0.2526,
+      "step": 129060
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1052005969980073e-06,
+      "loss": 0.2341,
+      "step": 129065
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1043631035695918e-06,
+      "loss": 0.2538,
+      "step": 129070
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1035259204102756e-06,
+      "loss": 0.2729,
+      "step": 129075
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1026890475309277e-06,
+      "loss": 0.2602,
+      "step": 129080
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1018524849424089e-06,
+      "loss": 0.2441,
+      "step": 129085
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1010162326555912e-06,
+      "loss": 0.2472,
+      "step": 129090
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.100180290681327e-06,
+      "loss": 0.2702,
+      "step": 129095
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0993446590304658e-06,
+      "loss": 0.257,
+      "step": 129100
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0985093377138656e-06,
+      "loss": 0.2181,
+      "step": 129105
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0976743267423705e-06,
+      "loss": 0.2417,
+      "step": 129110
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.096839626126825e-06,
+      "loss": 0.2334,
+      "step": 129115
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0960052358780615e-06,
+      "loss": 0.2583,
+      "step": 129120
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0951711560069161e-06,
+      "loss": 0.254,
+      "step": 129125
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0943373865242218e-06,
+      "loss": 0.2625,
+      "step": 129130
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0935039274408003e-06,
+      "loss": 0.2485,
+      "step": 129135
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0926707787674767e-06,
+      "loss": 0.2437,
+      "step": 129140
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.091837940515064e-06,
+      "loss": 0.2593,
+      "step": 129145
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0910054126943846e-06,
+      "loss": 0.2597,
+      "step": 129150
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.090173195316238e-06,
+      "loss": 0.2385,
+      "step": 129155
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0893412883914378e-06,
+      "loss": 0.2366,
+      "step": 129160
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0885096919307835e-06,
+      "loss": 0.2441,
+      "step": 129165
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0876784059450722e-06,
+      "loss": 0.2596,
+      "step": 129170
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.086847430445101e-06,
+      "loss": 0.25,
+      "step": 129175
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.0860167654416525e-06,
+      "loss": 0.2292,
+      "step": 129180
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0851864109455184e-06,
+      "loss": 0.2542,
+      "step": 129185
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.084356366967479e-06,
+      "loss": 0.2411,
+      "step": 129190
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0835266335183113e-06,
+      "loss": 0.257,
+      "step": 129195
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.082697210608785e-06,
+      "loss": 0.2618,
+      "step": 129200
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0818680982496749e-06,
+      "loss": 0.2573,
+      "step": 129205
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.081039296451744e-06,
+      "loss": 0.2352,
+      "step": 129210
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0802108052257592e-06,
+      "loss": 0.2664,
+      "step": 129215
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0793826245824674e-06,
+      "loss": 0.2425,
+      "step": 129220
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0785547545326318e-06,
+      "loss": 0.2493,
+      "step": 129225
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0777271950869972e-06,
+      "loss": 0.245,
+      "step": 129230
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0768999462563128e-06,
+      "loss": 0.2364,
+      "step": 129235
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0760730080513093e-06,
+      "loss": 0.2364,
+      "step": 129240
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0752463804827361e-06,
+      "loss": 0.2683,
+      "step": 129245
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.074420063561324e-06,
+      "loss": 0.2563,
+      "step": 129250
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.073594057297797e-06,
+      "loss": 0.2401,
+      "step": 129255
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0727683617028805e-06,
+      "loss": 0.2371,
+      "step": 129260
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.071942976787299e-06,
+      "loss": 0.2316,
+      "step": 129265
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0711179025617719e-06,
+      "loss": 0.249,
+      "step": 129270
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.070293139037004e-06,
+      "loss": 0.2508,
+      "step": 129275
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0694686862237152e-06,
+      "loss": 0.2644,
+      "step": 129280
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.068644544132602e-06,
+      "loss": 0.2427,
+      "step": 129285
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.06782071277437e-06,
+      "loss": 0.2476,
+      "step": 129290
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0669971921597105e-06,
+      "loss": 0.263,
+      "step": 129295
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0661739822993234e-06,
+      "loss": 0.2553,
+      "step": 129300
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0653510832038943e-06,
+      "loss": 0.2611,
+      "step": 129305
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0645284948841067e-06,
+      "loss": 0.2628,
+      "step": 129310
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0637062173506406e-06,
+      "loss": 0.2506,
+      "step": 129315
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.062884250614174e-06,
+      "loss": 0.2878,
+      "step": 129320
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0620625946853841e-06,
+      "loss": 0.2495,
+      "step": 129325
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0612412495749348e-06,
+      "loss": 0.2503,
+      "step": 129330
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0604202152934895e-06,
+      "loss": 0.2626,
+      "step": 129335
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0595994918517121e-06,
+      "loss": 0.237,
+      "step": 129340
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0587790792602608e-06,
+      "loss": 0.2439,
+      "step": 129345
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0579589775297827e-06,
+      "loss": 0.2621,
+      "step": 129350
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0571391866709275e-06,
+      "loss": 0.2516,
+      "step": 129355
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0563197066943447e-06,
+      "loss": 0.2473,
+      "step": 129360
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0555005376106708e-06,
+      "loss": 0.2242,
+      "step": 129365
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0546816794305358e-06,
+      "loss": 0.263,
+      "step": 129370
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0538631321645815e-06,
+      "loss": 0.2499,
+      "step": 129375
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0530448958234378e-06,
+      "loss": 0.2622,
+      "step": 129380
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0522269704177218e-06,
+      "loss": 0.2467,
+      "step": 129385
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0514093559580552e-06,
+      "loss": 0.2427,
+      "step": 129390
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0505920524550571e-06,
+      "loss": 0.2242,
+      "step": 129395
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.049775059919339e-06,
+      "loss": 0.251,
+      "step": 129400
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0489583783615086e-06,
+      "loss": 0.2533,
+      "step": 129405
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0481420077921634e-06,
+      "loss": 0.2605,
+      "step": 129410
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.047325948221914e-06,
+      "loss": 0.206,
+      "step": 129415
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0465101996613518e-06,
+      "loss": 0.2622,
+      "step": 129420
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.045694762121066e-06,
+      "loss": 0.2663,
+      "step": 129425
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0448796356116508e-06,
+      "loss": 0.2672,
+      "step": 129430
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0440648201436836e-06,
+      "loss": 0.2433,
+      "step": 129435
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.043250315727748e-06,
+      "loss": 0.2424,
+      "step": 129440
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.042436122374421e-06,
+      "loss": 0.2538,
+      "step": 129445
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0416222400942698e-06,
+      "loss": 0.2515,
+      "step": 129450
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0408086688978663e-06,
+      "loss": 0.2418,
+      "step": 129455
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.0399954087957714e-06,
+      "loss": 0.2414,
+      "step": 129460
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.039182459798549e-06,
+      "loss": 0.2363,
+      "step": 129465
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0383698219167459e-06,
+      "loss": 0.2648,
+      "step": 129470
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0375574951609234e-06,
+      "loss": 0.2565,
+      "step": 129475
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0367454795416205e-06,
+      "loss": 0.2511,
+      "step": 129480
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0359337750693893e-06,
+      "loss": 0.242,
+      "step": 129485
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0351223817547633e-06,
+      "loss": 0.2405,
+      "step": 129490
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0343112996082815e-06,
+      "loss": 0.2332,
+      "step": 129495
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0335005286404742e-06,
+      "loss": 0.2607,
+      "step": 129500
+    },
+    {
+      "epoch": 4.56,
+      "eval_loss": 0.24865983426570892,
+      "eval_runtime": 10.5527,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 129500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0326900688618635e-06,
+      "loss": 0.256,
+      "step": 129505
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0318799202829827e-06,
+      "loss": 0.2483,
+      "step": 129510
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0310700829143427e-06,
+      "loss": 0.238,
+      "step": 129515
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0302605567664631e-06,
+      "loss": 0.2675,
+      "step": 129520
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.029451341849852e-06,
+      "loss": 0.2344,
+      "step": 129525
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0286424381750175e-06,
+      "loss": 0.267,
+      "step": 129530
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0278338457524655e-06,
+      "loss": 0.2232,
+      "step": 129535
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0270255645926902e-06,
+      "loss": 0.2512,
+      "step": 129540
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0262175947061886e-06,
+      "loss": 0.276,
+      "step": 129545
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0254099361034553e-06,
+      "loss": 0.2621,
+      "step": 129550
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0246025887949734e-06,
+      "loss": 0.2586,
+      "step": 129555
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0237955527912264e-06,
+      "loss": 0.2416,
+      "step": 129560
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0229888281026916e-06,
+      "loss": 0.2527,
+      "step": 129565
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0221824147398474e-06,
+      "loss": 0.2497,
+      "step": 129570
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0213763127131625e-06,
+      "loss": 0.2474,
+      "step": 129575
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0205705220331042e-06,
+      "loss": 0.2455,
+      "step": 129580
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0197650427101302e-06,
+      "loss": 0.2321,
+      "step": 129585
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0189598747547074e-06,
+      "loss": 0.2573,
+      "step": 129590
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0181550181772858e-06,
+      "loss": 0.2412,
+      "step": 129595
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0173504729883182e-06,
+      "loss": 0.2519,
+      "step": 129600
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0165462391982432e-06,
+      "loss": 0.2473,
+      "step": 129605
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0157423168175167e-06,
+      "loss": 0.2461,
+      "step": 129610
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.014938705856569e-06,
+      "loss": 0.2548,
+      "step": 129615
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0141354063258335e-06,
+      "loss": 0.2538,
+      "step": 129620
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0133324182357434e-06,
+      "loss": 0.2641,
+      "step": 129625
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.012529741596724e-06,
+      "loss": 0.2556,
+      "step": 129630
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0117273764191998e-06,
+      "loss": 0.2564,
+      "step": 129635
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0109253227135824e-06,
+      "loss": 0.2577,
+      "step": 129640
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.010123580490291e-06,
+      "loss": 0.2703,
+      "step": 129645
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0093221497597367e-06,
+      "loss": 0.2258,
+      "step": 129650
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0085210305323223e-06,
+      "loss": 0.2672,
+      "step": 129655
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0077202228184507e-06,
+      "loss": 0.25,
+      "step": 129660
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0069197266285245e-06,
+      "loss": 0.2373,
+      "step": 129665
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0061195419729302e-06,
+      "loss": 0.2499,
+      "step": 129670
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0053196688620647e-06,
+      "loss": 0.2464,
+      "step": 129675
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.004520107306306e-06,
+      "loss": 0.2409,
+      "step": 129680
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0037208573160428e-06,
+      "loss": 0.2213,
+      "step": 129685
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0029219189016503e-06,
+      "loss": 0.2502,
+      "step": 129690
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0021232920735008e-06,
+      "loss": 0.2466,
+      "step": 129695
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0013249768419635e-06,
+      "loss": 0.2498,
+      "step": 129700
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.0005269732174055e-06,
+      "loss": 0.2472,
+      "step": 129705
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.997292812101905e-07,
+      "loss": 0.2651,
+      "step": 129710
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.98931900830677e-07,
+      "loss": 0.235,
+      "step": 129715
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.981348320892093e-07,
+      "loss": 0.2582,
+      "step": 129720
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.97338074996146e-07,
+      "loss": 0.2462,
+      "step": 129725
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.965416295618313e-07,
+      "loss": 0.2511,
+      "step": 129730
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.957454957966017e-07,
+      "loss": 0.2487,
+      "step": 129735
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.949496737107988e-07,
+      "loss": 0.2294,
+      "step": 129740
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.941541633147562e-07,
+      "loss": 0.2585,
+      "step": 129745
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.933589646187985e-07,
+      "loss": 0.2421,
+      "step": 129750
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.92564077633254e-07,
+      "loss": 0.2785,
+      "step": 129755
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.91769502368442e-07,
+      "loss": 0.2468,
+      "step": 129760
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.90975238834682e-07,
+      "loss": 0.2277,
+      "step": 129765
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.901812870422884e-07,
+      "loss": 0.2453,
+      "step": 129770
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.893876470015661e-07,
+      "loss": 0.2453,
+      "step": 129775
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.885943187228214e-07,
+      "loss": 0.2807,
+      "step": 129780
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.878013022163567e-07,
+      "loss": 0.2355,
+      "step": 129785
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.870085974924698e-07,
+      "loss": 0.2383,
+      "step": 129790
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.862162045614436e-07,
+      "loss": 0.2452,
+      "step": 129795
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.854241234335814e-07,
+      "loss": 0.265,
+      "step": 129800
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.846323541191582e-07,
+      "loss": 0.2254,
+      "step": 129805
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.838408966284545e-07,
+      "loss": 0.2428,
+      "step": 129810
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.830497509717511e-07,
+      "loss": 0.2269,
+      "step": 129815
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.822589171593206e-07,
+      "loss": 0.2352,
+      "step": 129820
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.814683952014292e-07,
+      "loss": 0.2565,
+      "step": 129825
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.806781851083414e-07,
+      "loss": 0.2418,
+      "step": 129830
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.79888286890318e-07,
+      "loss": 0.2701,
+      "step": 129835
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.790987005576152e-07,
+      "loss": 0.2583,
+      "step": 129840
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.783094261204884e-07,
+      "loss": 0.2461,
+      "step": 129845
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.775204635891766e-07,
+      "loss": 0.2528,
+      "step": 129850
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.767318129739355e-07,
+      "loss": 0.2441,
+      "step": 129855
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.759434742849988e-07,
+      "loss": 0.263,
+      "step": 129860
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.751554475326025e-07,
+      "loss": 0.2548,
+      "step": 129865
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.743677327269802e-07,
+      "loss": 0.2489,
+      "step": 129870
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.735803298783569e-07,
+      "loss": 0.2698,
+      "step": 129875
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.727932389969636e-07,
+      "loss": 0.2638,
+      "step": 129880
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.72006460093014e-07,
+      "loss": 0.2491,
+      "step": 129885
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.712199931767223e-07,
+      "loss": 0.2499,
+      "step": 129890
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.704338382583084e-07,
+      "loss": 0.2712,
+      "step": 129895
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.69647995347972e-07,
+      "loss": 0.2672,
+      "step": 129900
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.688624644559218e-07,
+      "loss": 0.2716,
+      "step": 129905
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.680772455923526e-07,
+      "loss": 0.2433,
+      "step": 129910
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.672923387674644e-07,
+      "loss": 0.2583,
+      "step": 129915
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.665077439914461e-07,
+      "loss": 0.2455,
+      "step": 129920
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.65723461274487e-07,
+      "loss": 0.2558,
+      "step": 129925
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.649394906267677e-07,
+      "loss": 0.25,
+      "step": 129930
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.64155832058472e-07,
+      "loss": 0.2606,
+      "step": 129935
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.63372485579775e-07,
+      "loss": 0.2532,
+      "step": 129940
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.625894512008432e-07,
+      "loss": 0.2402,
+      "step": 129945
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.618067289318412e-07,
+      "loss": 0.2522,
+      "step": 129950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.610243187829438e-07,
+      "loss": 0.2268,
+      "step": 129955
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.602422207643013e-07,
+      "loss": 0.2497,
+      "step": 129960
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.59460434886067e-07,
+      "loss": 0.2595,
+      "step": 129965
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.58678961158399e-07,
+      "loss": 0.2403,
+      "step": 129970
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.578977995914424e-07,
+      "loss": 0.2467,
+      "step": 129975
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.57116950195336e-07,
+      "loss": 0.2558,
+      "step": 129980
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.563364129802187e-07,
+      "loss": 0.2688,
+      "step": 129985
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.55556187956233e-07,
+      "loss": 0.2545,
+      "step": 129990
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.547762751335038e-07,
+      "loss": 0.2514,
+      "step": 129995
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.539966745221563e-07,
+      "loss": 0.251,
+      "step": 130000
+    },
+    {
+      "epoch": 4.57,
+      "eval_loss": 0.24862876534461975,
+      "eval_runtime": 10.5504,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 130000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.532173861323129e-07,
+      "loss": 0.2545,
+      "step": 130005
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.524384099740991e-07,
+      "loss": 0.2479,
+      "step": 130010
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.516597460576232e-07,
+      "loss": 0.2336,
+      "step": 130015
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.508813943929967e-07,
+      "loss": 0.2482,
+      "step": 130020
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.501033549903227e-07,
+      "loss": 0.2513,
+      "step": 130025
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.493256278597124e-07,
+      "loss": 0.2543,
+      "step": 130030
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.485482130112549e-07,
+      "loss": 0.2624,
+      "step": 130035
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.477711104550507e-07,
+      "loss": 0.2444,
+      "step": 130040
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.46994320201186e-07,
+      "loss": 0.2581,
+      "step": 130045
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.4621784225975e-07,
+      "loss": 0.2551,
+      "step": 130050
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.454416766408236e-07,
+      "loss": 0.2578,
+      "step": 130055
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.446658233544847e-07,
+      "loss": 0.2461,
+      "step": 130060
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.438902824108059e-07,
+      "loss": 0.2454,
+      "step": 130065
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.431150538198597e-07,
+      "loss": 0.2429,
+      "step": 130070
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.423401375917101e-07,
+      "loss": 0.2445,
+      "step": 130075
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.41565533736416e-07,
+      "loss": 0.2653,
+      "step": 130080
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.407912422640414e-07,
+      "loss": 0.2538,
+      "step": 130085
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.400172631846337e-07,
+      "loss": 0.2458,
+      "step": 130090
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.392435965082491e-07,
+      "loss": 0.2516,
+      "step": 130095
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.384702422449265e-07,
+      "loss": 0.2256,
+      "step": 130100
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.376972004047136e-07,
+      "loss": 0.2582,
+      "step": 130105
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.369244709976465e-07,
+      "loss": 0.239,
+      "step": 130110
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.361520540337564e-07,
+      "loss": 0.2414,
+      "step": 130115
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.353799495230687e-07,
+      "loss": 0.2354,
+      "step": 130120
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.346081574756166e-07,
+      "loss": 0.2258,
+      "step": 130125
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.338366779014174e-07,
+      "loss": 0.2356,
+      "step": 130130
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.330655108104907e-07,
+      "loss": 0.237,
+      "step": 130135
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.322946562128426e-07,
+      "loss": 0.2546,
+      "step": 130140
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.3152411411849e-07,
+      "loss": 0.2514,
+      "step": 130145
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.307538845374358e-07,
+      "loss": 0.2565,
+      "step": 130150
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.299839674796806e-07,
+      "loss": 0.2609,
+      "step": 130155
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.292143629552191e-07,
+      "loss": 0.2513,
+      "step": 130160
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.284450709740461e-07,
+      "loss": 0.2503,
+      "step": 130165
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.276760915461508e-07,
+      "loss": 0.2504,
+      "step": 130170
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.269074246815195e-07,
+      "loss": 0.2378,
+      "step": 130175
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.261390703901251e-07,
+      "loss": 0.2432,
+      "step": 130180
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.25371028681954e-07,
+      "loss": 0.2401,
+      "step": 130185
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.246032995669728e-07,
+      "loss": 0.2513,
+      "step": 130190
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.238358830551519e-07,
+      "loss": 0.2554,
+      "step": 130195
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.230687791564524e-07,
+      "loss": 0.2605,
+      "step": 130200
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.223019878808415e-07,
+      "loss": 0.2507,
+      "step": 130205
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.215355092382694e-07,
+      "loss": 0.2304,
+      "step": 130210
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.207693432386893e-07,
+      "loss": 0.248,
+      "step": 130215
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.200034898920546e-07,
+      "loss": 0.2407,
+      "step": 130220
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.192379492083042e-07,
+      "loss": 0.2477,
+      "step": 130225
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.184727211973776e-07,
+      "loss": 0.2558,
+      "step": 130230
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.177078058692112e-07,
+      "loss": 0.2516,
+      "step": 130235
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.169432032337416e-07,
+      "loss": 0.2311,
+      "step": 130240
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.161789133008914e-07,
+      "loss": 0.2586,
+      "step": 130245
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.15414936080583e-07,
+      "loss": 0.2382,
+      "step": 130250
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.146512715827421e-07,
+      "loss": 0.2545,
+      "step": 130255
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.1388791981728e-07,
+      "loss": 0.237,
+      "step": 130260
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.13124880794114e-07,
+      "loss": 0.2335,
+      "step": 130265
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.12362154523147e-07,
+      "loss": 0.2433,
+      "step": 130270
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.115997410142796e-07,
+      "loss": 0.2211,
+      "step": 130275
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.108376402774177e-07,
+      "loss": 0.2445,
+      "step": 130280
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.100758523224534e-07,
+      "loss": 0.2432,
+      "step": 130285
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.09314377159276e-07,
+      "loss": 0.2565,
+      "step": 130290
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.085532147977749e-07,
+      "loss": 0.2818,
+      "step": 130295
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.077923652478365e-07,
+      "loss": 0.2358,
+      "step": 130300
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.070318285193336e-07,
+      "loss": 0.2563,
+      "step": 130305
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.06271604622147e-07,
+      "loss": 0.2659,
+      "step": 130310
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.055116935661439e-07,
+      "loss": 0.2431,
+      "step": 130315
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.047520953611943e-07,
+      "loss": 0.2275,
+      "step": 130320
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.039928100171569e-07,
+      "loss": 0.2477,
+      "step": 130325
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.032338375438931e-07,
+      "loss": 0.2529,
+      "step": 130330
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.024751779512592e-07,
+      "loss": 0.255,
+      "step": 130335
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.017168312491053e-07,
+      "loss": 0.23,
+      "step": 130340
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.009587974472767e-07,
+      "loss": 0.2536,
+      "step": 130345
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.002010765556124e-07,
+      "loss": 0.23,
+      "step": 130350
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.994436685839602e-07,
+      "loss": 0.2501,
+      "step": 130355
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.986865735421429e-07,
+      "loss": 0.2553,
+      "step": 130360
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.979297914400026e-07,
+      "loss": 0.2381,
+      "step": 130365
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.971733222873563e-07,
+      "loss": 0.2512,
+      "step": 130370
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.964171660940323e-07,
+      "loss": 0.2411,
+      "step": 130375
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.956613228698479e-07,
+      "loss": 0.253,
+      "step": 130380
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.949057926246146e-07,
+      "loss": 0.2265,
+      "step": 130385
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.941505753681412e-07,
+      "loss": 0.2592,
+      "step": 130390
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.933956711102393e-07,
+      "loss": 0.2514,
+      "step": 130395
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.926410798607066e-07,
+      "loss": 0.2592,
+      "step": 130400
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.918868016293435e-07,
+      "loss": 0.2251,
+      "step": 130405
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.911328364259397e-07,
+      "loss": 0.2537,
+      "step": 130410
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.903791842602899e-07,
+      "loss": 0.2395,
+      "step": 130415
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.896258451421752e-07,
+      "loss": 0.2495,
+      "step": 130420
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.888728190813822e-07,
+      "loss": 0.2558,
+      "step": 130425
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.881201060876837e-07,
+      "loss": 0.2303,
+      "step": 130430
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.873677061708551e-07,
+      "loss": 0.2421,
+      "step": 130435
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.866156193406694e-07,
+      "loss": 0.2329,
+      "step": 130440
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.858638456068852e-07,
+      "loss": 0.2282,
+      "step": 130445
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.851123849792698e-07,
+      "loss": 0.2595,
+      "step": 130450
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.843612374675764e-07,
+      "loss": 0.2371,
+      "step": 130455
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.836104030815612e-07,
+      "loss": 0.2655,
+      "step": 130460
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.828598818309663e-07,
+      "loss": 0.248,
+      "step": 130465
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.82109673725548e-07,
+      "loss": 0.2432,
+      "step": 130470
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.813597787750372e-07,
+      "loss": 0.2633,
+      "step": 130475
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.806101969891761e-07,
+      "loss": 0.2779,
+      "step": 130480
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.79860928377696e-07,
+      "loss": 0.2375,
+      "step": 130485
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.791119729503278e-07,
+      "loss": 0.2311,
+      "step": 130490
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.783633307167943e-07,
+      "loss": 0.2546,
+      "step": 130495
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.776150016868157e-07,
+      "loss": 0.2328,
+      "step": 130500
+    },
+    {
+      "epoch": 4.59,
+      "eval_loss": 0.24856102466583252,
+      "eval_runtime": 10.5535,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 9.475,
+      "step": 130500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.768669858701061e-07,
+      "loss": 0.2428,
+      "step": 130505
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.761192832763859e-07,
+      "loss": 0.2483,
+      "step": 130510
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.753718939153555e-07,
+      "loss": 0.2551,
+      "step": 130515
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.746248177967209e-07,
+      "loss": 0.2682,
+      "step": 130520
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.738780549301884e-07,
+      "loss": 0.2427,
+      "step": 130525
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.731316053254446e-07,
+      "loss": 0.2679,
+      "step": 130530
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.723854689921901e-07,
+      "loss": 0.2462,
+      "step": 130535
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.716396459401088e-07,
+      "loss": 0.2441,
+      "step": 130540
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.708941361788847e-07,
+      "loss": 0.2305,
+      "step": 130545
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.701489397181989e-07,
+      "loss": 0.2483,
+      "step": 130550
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.694040565677297e-07,
+      "loss": 0.238,
+      "step": 130555
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.686594867371417e-07,
+      "loss": 0.2371,
+      "step": 130560
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.679152302361104e-07,
+      "loss": 0.2677,
+      "step": 130565
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.671712870742948e-07,
+      "loss": 0.2529,
+      "step": 130570
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.664276572613595e-07,
+      "loss": 0.2513,
+      "step": 130575
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.656843408069493e-07,
+      "loss": 0.2493,
+      "step": 130580
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.64941337720726e-07,
+      "loss": 0.2527,
+      "step": 130585
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.641986480123349e-07,
+      "loss": 0.2418,
+      "step": 130590
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.634562716914179e-07,
+      "loss": 0.2504,
+      "step": 130595
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 8.627142087676121e-07,
+      "loss": 0.257,
+      "step": 130600
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.619724592505596e-07,
+      "loss": 0.2459,
+      "step": 130605
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.61231023149886e-07,
+      "loss": 0.2586,
+      "step": 130610
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.604899004752198e-07,
+      "loss": 0.2216,
+      "step": 130615
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.59749091236181e-07,
+      "loss": 0.2588,
+      "step": 130620
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.590085954423955e-07,
+      "loss": 0.2517,
+      "step": 130625
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.582684131034718e-07,
+      "loss": 0.2439,
+      "step": 130630
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.575285442290193e-07,
+      "loss": 0.2547,
+      "step": 130635
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.567889888286495e-07,
+      "loss": 0.265,
+      "step": 130640
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.56049746911966e-07,
+      "loss": 0.2511,
+      "step": 130645
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.553108184885666e-07,
+      "loss": 0.2476,
+      "step": 130650
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.545722035680409e-07,
+      "loss": 0.2505,
+      "step": 130655
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.538339021599839e-07,
+      "loss": 0.2566,
+      "step": 130660
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.530959142739797e-07,
+      "loss": 0.2655,
+      "step": 130665
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.523582399196151e-07,
+      "loss": 0.2626,
+      "step": 130670
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.5162087910646e-07,
+      "loss": 0.2215,
+      "step": 130675
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.508838318440959e-07,
+      "loss": 0.2521,
+      "step": 130680
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.501470981420928e-07,
+      "loss": 0.26,
+      "step": 130685
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.494106780100098e-07,
+      "loss": 0.229,
+      "step": 130690
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.48674571457414e-07,
+      "loss": 0.2412,
+      "step": 130695
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.479387784938647e-07,
+      "loss": 0.2515,
+      "step": 130700
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.472032991289125e-07,
+      "loss": 0.2331,
+      "step": 130705
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.464681333721109e-07,
+      "loss": 0.2427,
+      "step": 130710
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.457332812329966e-07,
+      "loss": 0.26,
+      "step": 130715
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.449987427211231e-07,
+      "loss": 0.2556,
+      "step": 130720
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.442645178460217e-07,
+      "loss": 0.2508,
+      "step": 130725
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.435306066172238e-07,
+      "loss": 0.2447,
+      "step": 130730
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.427970090442605e-07,
+      "loss": 0.2612,
+      "step": 130735
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.420637251366604e-07,
+      "loss": 0.2698,
+      "step": 130740
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.413307549039406e-07,
+      "loss": 0.2544,
+      "step": 130745
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.405980983556216e-07,
+      "loss": 0.2537,
+      "step": 130750
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.398657555012123e-07,
+      "loss": 0.2554,
+      "step": 130755
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.391337263502247e-07,
+      "loss": 0.2342,
+      "step": 130760
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.384020109121649e-07,
+      "loss": 0.2464,
+      "step": 130765
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.376706091965309e-07,
+      "loss": 0.2523,
+      "step": 130770
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.369395212128206e-07,
+      "loss": 0.2362,
+      "step": 130775
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.362087469705265e-07,
+      "loss": 0.2474,
+      "step": 130780
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.354782864791383e-07,
+      "loss": 0.2297,
+      "step": 130785
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.347481397481344e-07,
+      "loss": 0.2467,
+      "step": 130790
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.340183067870072e-07,
+      "loss": 0.2582,
+      "step": 130795
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.332887876052215e-07,
+      "loss": 0.2395,
+      "step": 130800
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.325595822122556e-07,
+      "loss": 0.273,
+      "step": 130805
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.318306906175771e-07,
+      "loss": 0.2424,
+      "step": 130810
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.311021128306506e-07,
+      "loss": 0.235,
+      "step": 130815
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.303738488609353e-07,
+      "loss": 0.2673,
+      "step": 130820
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.296458987178873e-07,
+      "loss": 0.2706,
+      "step": 130825
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.289182624109549e-07,
+      "loss": 0.275,
+      "step": 130830
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.28190939949594e-07,
+      "loss": 0.2573,
+      "step": 130835
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.274639313432419e-07,
+      "loss": 0.2329,
+      "step": 130840
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.267372366013409e-07,
+      "loss": 0.2559,
+      "step": 130845
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.260108557333251e-07,
+      "loss": 0.2578,
+      "step": 130850
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.252847887486287e-07,
+      "loss": 0.2443,
+      "step": 130855
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.245590356566745e-07,
+      "loss": 0.253,
+      "step": 130860
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.238335964668914e-07,
+      "loss": 0.2273,
+      "step": 130865
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.231084711886939e-07,
+      "loss": 0.2434,
+      "step": 130870
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.223836598315049e-07,
+      "loss": 0.2605,
+      "step": 130875
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.216591624047282e-07,
+      "loss": 0.2579,
+      "step": 130880
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 8.209349789177728e-07,
+      "loss": 0.2405,
+      "step": 130885
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.202111093800396e-07,
+      "loss": 0.236,
+      "step": 130890
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.194875538009322e-07,
+      "loss": 0.2623,
+      "step": 130895
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.187643121898458e-07,
+      "loss": 0.259,
+      "step": 130900
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.180413845561646e-07,
+      "loss": 0.2347,
+      "step": 130905
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.173187709092811e-07,
+      "loss": 0.2467,
+      "step": 130910
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.165964712585739e-07,
+      "loss": 0.2806,
+      "step": 130915
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.158744856134271e-07,
+      "loss": 0.2558,
+      "step": 130920
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.151528139832109e-07,
+      "loss": 0.25,
+      "step": 130925
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.144314563772987e-07,
+      "loss": 0.2682,
+      "step": 130930
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.137104128050549e-07,
+      "loss": 0.2685,
+      "step": 130935
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.129896832758416e-07,
+      "loss": 0.2251,
+      "step": 130940
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.122692677990151e-07,
+      "loss": 0.2476,
+      "step": 130945
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.115491663839348e-07,
+      "loss": 0.2486,
+      "step": 130950
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.108293790399485e-07,
+      "loss": 0.2653,
+      "step": 130955
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.101099057763989e-07,
+      "loss": 0.2658,
+      "step": 130960
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.093907466026313e-07,
+      "loss": 0.2512,
+      "step": 130965
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.086719015279825e-07,
+      "loss": 0.2522,
+      "step": 130970
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.07953370561787e-07,
+      "loss": 0.2256,
+      "step": 130975
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.072351537133732e-07,
+      "loss": 0.275,
+      "step": 130980
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.065172509920671e-07,
+      "loss": 0.2418,
+      "step": 130985
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.057996624071917e-07,
+      "loss": 0.2487,
+      "step": 130990
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.050823879680619e-07,
+      "loss": 0.2264,
+      "step": 130995
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.043654276839951e-07,
+      "loss": 0.2602,
+      "step": 131000
+    },
+    {
+      "epoch": 4.61,
+      "eval_loss": 0.24853986501693726,
+      "eval_runtime": 10.534,
+      "eval_samples_per_second": 9.493,
+      "eval_steps_per_second": 9.493,
+      "step": 131000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.036487815642923e-07,
+      "loss": 0.2605,
+      "step": 131005
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.029324496182683e-07,
+      "loss": 0.2415,
+      "step": 131010
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.022164318552184e-07,
+      "loss": 0.2655,
+      "step": 131015
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.015007282844405e-07,
+      "loss": 0.2641,
+      "step": 131020
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.007853389152276e-07,
+      "loss": 0.2411,
+      "step": 131025
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 8.000702637568719e-07,
+      "loss": 0.2311,
+      "step": 131030
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.99355502818655e-07,
+      "loss": 0.2515,
+      "step": 131035
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.986410561098556e-07,
+      "loss": 0.2698,
+      "step": 131040
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.979269236397552e-07,
+      "loss": 0.2797,
+      "step": 131045
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.97213105417624e-07,
+      "loss": 0.2432,
+      "step": 131050
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.964996014527298e-07,
+      "loss": 0.2413,
+      "step": 131055
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.957864117543373e-07,
+      "loss": 0.2622,
+      "step": 131060
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.950735363317085e-07,
+      "loss": 0.2344,
+      "step": 131065
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.943609751940972e-07,
+      "loss": 0.2349,
+      "step": 131070
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.936487283507543e-07,
+      "loss": 0.2351,
+      "step": 131075
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.929367958109307e-07,
+      "loss": 0.2361,
+      "step": 131080
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.922251775838718e-07,
+      "loss": 0.2419,
+      "step": 131085
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.915138736788147e-07,
+      "loss": 0.261,
+      "step": 131090
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.908028841049992e-07,
+      "loss": 0.2685,
+      "step": 131095
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.900922088716483e-07,
+      "loss": 0.2618,
+      "step": 131100
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.893818479879966e-07,
+      "loss": 0.2682,
+      "step": 131105
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.886718014632699e-07,
+      "loss": 0.2335,
+      "step": 131110
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.879620693066803e-07,
+      "loss": 0.2395,
+      "step": 131115
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.872526515274481e-07,
+      "loss": 0.268,
+      "step": 131120
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.865435481347827e-07,
+      "loss": 0.2562,
+      "step": 131125
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.858347591378907e-07,
+      "loss": 0.228,
+      "step": 131130
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.851262845459756e-07,
+      "loss": 0.2396,
+      "step": 131135
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.844181243682414e-07,
+      "loss": 0.2498,
+      "step": 131140
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.837102786138806e-07,
+      "loss": 0.2512,
+      "step": 131145
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.830027472920803e-07,
+      "loss": 0.2554,
+      "step": 131150
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.822955304120277e-07,
+      "loss": 0.2474,
+      "step": 131155
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.815886279829098e-07,
+      "loss": 0.2653,
+      "step": 131160
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.808820400139028e-07,
+      "loss": 0.2367,
+      "step": 131165
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 7.801757665141824e-07,
+      "loss": 0.2574,
+      "step": 131170
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.794698074929163e-07,
+      "loss": 0.2576,
+      "step": 131175
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.787641629592752e-07,
+      "loss": 0.2521,
+      "step": 131180
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.780588329224154e-07,
+      "loss": 0.2378,
+      "step": 131185
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.773538173915018e-07,
+      "loss": 0.2399,
+      "step": 131190
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.766491163756828e-07,
+      "loss": 0.2391,
+      "step": 131195
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.759447298841149e-07,
+      "loss": 0.2725,
+      "step": 131200
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.75240657925938e-07,
+      "loss": 0.257,
+      "step": 131205
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.745369005102976e-07,
+      "loss": 0.2435,
+      "step": 131210
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.73833457646328e-07,
+      "loss": 0.247,
+      "step": 131215
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.731303293431691e-07,
+      "loss": 0.2392,
+      "step": 131220
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.72427515609947e-07,
+      "loss": 0.2595,
+      "step": 131225
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.717250164557849e-07,
+      "loss": 0.2273,
+      "step": 131230
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.710228318898033e-07,
+      "loss": 0.254,
+      "step": 131235
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.703209619211254e-07,
+      "loss": 0.2277,
+      "step": 131240
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.696194065588635e-07,
+      "loss": 0.2502,
+      "step": 131245
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.689181658121214e-07,
+      "loss": 0.2546,
+      "step": 131250
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.682172396900111e-07,
+      "loss": 0.2075,
+      "step": 131255
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.67516628201631e-07,
+      "loss": 0.2491,
+      "step": 131260
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.668163313560795e-07,
+      "loss": 0.2517,
+      "step": 131265
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.661163491624434e-07,
+      "loss": 0.2478,
+      "step": 131270
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.654166816298159e-07,
+      "loss": 0.239,
+      "step": 131275
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.647173287672865e-07,
+      "loss": 0.2577,
+      "step": 131280
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.640182905839261e-07,
+      "loss": 0.2768,
+      "step": 131285
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.633195670888161e-07,
+      "loss": 0.2681,
+      "step": 131290
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.626211582910326e-07,
+      "loss": 0.2349,
+      "step": 131295
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.61923064199635e-07,
+      "loss": 0.2773,
+      "step": 131300
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.612252848236967e-07,
+      "loss": 0.2587,
+      "step": 131305
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.605278201722715e-07,
+      "loss": 0.2307,
+      "step": 131310
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.598306702544217e-07,
+      "loss": 0.2553,
+      "step": 131315
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.591338350791954e-07,
+      "loss": 0.2531,
+      "step": 131320
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.584373146556411e-07,
+      "loss": 0.2391,
+      "step": 131325
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.577411089927989e-07,
+      "loss": 0.2578,
+      "step": 131330
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.570452180997139e-07,
+      "loss": 0.2569,
+      "step": 131335
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.56349641985421e-07,
+      "loss": 0.2361,
+      "step": 131340
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.556543806589489e-07,
+      "loss": 0.2516,
+      "step": 131345
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.549594341293292e-07,
+      "loss": 0.2621,
+      "step": 131350
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.542648024055826e-07,
+      "loss": 0.2666,
+      "step": 131355
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.535704854967296e-07,
+      "loss": 0.2504,
+      "step": 131360
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.528764834117824e-07,
+      "loss": 0.235,
+      "step": 131365
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.521827961597561e-07,
+      "loss": 0.2512,
+      "step": 131370
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.514894237496573e-07,
+      "loss": 0.2873,
+      "step": 131375
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.507963661904871e-07,
+      "loss": 0.2377,
+      "step": 131380
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.50103623491244e-07,
+      "loss": 0.2497,
+      "step": 131385
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.494111956609262e-07,
+      "loss": 0.2483,
+      "step": 131390
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.48719082708521e-07,
+      "loss": 0.2678,
+      "step": 131395
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.480272846430159e-07,
+      "loss": 0.2581,
+      "step": 131400
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.473358014733949e-07,
+      "loss": 0.2626,
+      "step": 131405
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.466446332086347e-07,
+      "loss": 0.2685,
+      "step": 131410
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.459537798577109e-07,
+      "loss": 0.2595,
+      "step": 131415
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.452632414295946e-07,
+      "loss": 0.2587,
+      "step": 131420
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.445730179332478e-07,
+      "loss": 0.2841,
+      "step": 131425
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.438831093776383e-07,
+      "loss": 0.2513,
+      "step": 131430
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.431935157717201e-07,
+      "loss": 0.2596,
+      "step": 131435
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.425042371244472e-07,
+      "loss": 0.2492,
+      "step": 131440
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.418152734447708e-07,
+      "loss": 0.2612,
+      "step": 131445
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.411266247416365e-07,
+      "loss": 0.2343,
+      "step": 131450
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 7.40438291023987e-07,
+      "loss": 0.244,
+      "step": 131455
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.397502723007571e-07,
+      "loss": 0.254,
+      "step": 131460
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.390625685808811e-07,
+      "loss": 0.2319,
+      "step": 131465
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.383751798732907e-07,
+      "loss": 0.2528,
+      "step": 131470
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.376881061869096e-07,
+      "loss": 0.2371,
+      "step": 131475
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.370013475306581e-07,
+      "loss": 0.2413,
+      "step": 131480
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.363149039134542e-07,
+      "loss": 0.2594,
+      "step": 131485
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.356287753442131e-07,
+      "loss": 0.269,
+      "step": 131490
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.349429618318416e-07,
+      "loss": 0.2414,
+      "step": 131495
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.342574633852406e-07,
+      "loss": 0.2318,
+      "step": 131500
+    },
+    {
+      "epoch": 4.63,
+      "eval_loss": 0.24853749573230743,
+      "eval_runtime": 10.5675,
+      "eval_samples_per_second": 9.463,
+      "eval_steps_per_second": 9.463,
+      "step": 131500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.335722800133171e-07,
+      "loss": 0.2536,
+      "step": 131505
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.328874117249668e-07,
+      "loss": 0.2504,
+      "step": 131510
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.322028585290768e-07,
+      "loss": 0.2521,
+      "step": 131515
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.315186204345403e-07,
+      "loss": 0.2433,
+      "step": 131520
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.308346974502444e-07,
+      "loss": 0.2541,
+      "step": 131525
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.301510895850655e-07,
+      "loss": 0.2189,
+      "step": 131530
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.294677968478797e-07,
+      "loss": 0.2477,
+      "step": 131535
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.287848192475577e-07,
+      "loss": 0.234,
+      "step": 131540
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.281021567929703e-07,
+      "loss": 0.2328,
+      "step": 131545
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.274198094929824e-07,
+      "loss": 0.2481,
+      "step": 131550
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.267377773564509e-07,
+      "loss": 0.2676,
+      "step": 131555
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.2605606039223e-07,
+      "loss": 0.2516,
+      "step": 131560
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.253746586091792e-07,
+      "loss": 0.2411,
+      "step": 131565
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.246935720161358e-07,
+      "loss": 0.2535,
+      "step": 131570
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.24012800621951e-07,
+      "loss": 0.2354,
+      "step": 131575
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.233323444354595e-07,
+      "loss": 0.2391,
+      "step": 131580
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.226522034655015e-07,
+      "loss": 0.2697,
+      "step": 131585
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.219723777209058e-07,
+      "loss": 0.277,
+      "step": 131590
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.212928672104963e-07,
+      "loss": 0.2566,
+      "step": 131595
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.206136719430989e-07,
+      "loss": 0.2366,
+      "step": 131600
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.199347919275345e-07,
+      "loss": 0.2788,
+      "step": 131605
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.192562271726155e-07,
+      "loss": 0.2382,
+      "step": 131610
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.185779776871487e-07,
+      "loss": 0.2471,
+      "step": 131615
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.179000434799493e-07,
+      "loss": 0.2557,
+      "step": 131620
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.172224245598102e-07,
+      "loss": 0.2397,
+      "step": 131625
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.165451209355411e-07,
+      "loss": 0.2317,
+      "step": 131630
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.158681326159239e-07,
+      "loss": 0.2725,
+      "step": 131635
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.151914596097597e-07,
+      "loss": 0.2184,
+      "step": 131640
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.145151019258306e-07,
+      "loss": 0.249,
+      "step": 131645
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.138390595729183e-07,
+      "loss": 0.2474,
+      "step": 131650
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.131633325597964e-07,
+      "loss": 0.2322,
+      "step": 131655
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.124879208952468e-07,
+      "loss": 0.2616,
+      "step": 131660
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.118128245880346e-07,
+      "loss": 0.2338,
+      "step": 131665
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.111380436469279e-07,
+      "loss": 0.2484,
+      "step": 131670
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.104635780806834e-07,
+      "loss": 0.2542,
+      "step": 131675
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.09789427898061e-07,
+      "loss": 0.2621,
+      "step": 131680
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.091155931078175e-07,
+      "loss": 0.2529,
+      "step": 131685
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.084420737187014e-07,
+      "loss": 0.2661,
+      "step": 131690
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.07768869739453e-07,
+      "loss": 0.2306,
+      "step": 131695
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.070959811788181e-07,
+      "loss": 0.2482,
+      "step": 131700
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.06423408045534e-07,
+      "loss": 0.2536,
+      "step": 131705
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.057511503483299e-07,
+      "loss": 0.2493,
+      "step": 131710
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.050792080959351e-07,
+      "loss": 0.2561,
+      "step": 131715
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.044075812970785e-07,
+      "loss": 0.261,
+      "step": 131720
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.037362699604783e-07,
+      "loss": 0.2364,
+      "step": 131725
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.030652740948496e-07,
+      "loss": 0.2581,
+      "step": 131730
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.023945937089077e-07,
+      "loss": 0.2381,
+      "step": 131735
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 7.01724228811354e-07,
+      "loss": 0.2489,
+      "step": 131740
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.010541794109038e-07,
+      "loss": 0.2608,
+      "step": 131745
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 7.003844455162473e-07,
+      "loss": 0.2626,
+      "step": 131750
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.997150271360886e-07,
+      "loss": 0.2393,
+      "step": 131755
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.990459242791125e-07,
+      "loss": 0.2539,
+      "step": 131760
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.983771369540148e-07,
+      "loss": 0.2579,
+      "step": 131765
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.977086651694692e-07,
+      "loss": 0.2256,
+      "step": 131770
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.97040508934163e-07,
+      "loss": 0.2416,
+      "step": 131775
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.963726682567701e-07,
+      "loss": 0.2481,
+      "step": 131780
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.95705143145961e-07,
+      "loss": 0.2529,
+      "step": 131785
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.95037933610404e-07,
+      "loss": 0.2361,
+      "step": 131790
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.943710396587589e-07,
+      "loss": 0.2571,
+      "step": 131795
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.937044612996906e-07,
+      "loss": 0.2358,
+      "step": 131800
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.93038198541851e-07,
+      "loss": 0.2303,
+      "step": 131805
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.923722513938912e-07,
+      "loss": 0.2525,
+      "step": 131810
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.917066198644601e-07,
+      "loss": 0.2429,
+      "step": 131815
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.910413039621977e-07,
+      "loss": 0.2485,
+      "step": 131820
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.903763036957445e-07,
+      "loss": 0.2254,
+      "step": 131825
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.897116190737296e-07,
+      "loss": 0.2648,
+      "step": 131830
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.890472501047934e-07,
+      "loss": 0.2625,
+      "step": 131835
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.88383196797554e-07,
+      "loss": 0.2564,
+      "step": 131840
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.877194591606406e-07,
+      "loss": 0.2375,
+      "step": 131845
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.870560372026629e-07,
+      "loss": 0.2494,
+      "step": 131850
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.86392930932242e-07,
+      "loss": 0.233,
+      "step": 131855
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.857301403579875e-07,
+      "loss": 0.2484,
+      "step": 131860
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.850676654885008e-07,
+      "loss": 0.2281,
+      "step": 131865
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.84405506332389e-07,
+      "loss": 0.2505,
+      "step": 131870
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.837436628982452e-07,
+      "loss": 0.2339,
+      "step": 131875
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.830821351946654e-07,
+      "loss": 0.2606,
+      "step": 131880
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.82420923230237e-07,
+      "loss": 0.2293,
+      "step": 131885
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.817600270135504e-07,
+      "loss": 0.2381,
+      "step": 131890
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.810994465531822e-07,
+      "loss": 0.2254,
+      "step": 131895
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.804391818577089e-07,
+      "loss": 0.2608,
+      "step": 131900
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.797792329357039e-07,
+      "loss": 0.2643,
+      "step": 131905
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.791195997957439e-07,
+      "loss": 0.2592,
+      "step": 131910
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.784602824463859e-07,
+      "loss": 0.2448,
+      "step": 131915
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.778012808961925e-07,
+      "loss": 0.246,
+      "step": 131920
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.77142595153718e-07,
+      "loss": 0.236,
+      "step": 131925
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.764842252275195e-07,
+      "loss": 0.249,
+      "step": 131930
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.758261711261455e-07,
+      "loss": 0.2457,
+      "step": 131935
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.751684328581364e-07,
+      "loss": 0.2405,
+      "step": 131940
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.745110104320329e-07,
+      "loss": 0.2463,
+      "step": 131945
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.738539038563752e-07,
+      "loss": 0.2291,
+      "step": 131950
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.731971131396897e-07,
+      "loss": 0.2499,
+      "step": 131955
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.725406382905086e-07,
+      "loss": 0.2429,
+      "step": 131960
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.718844793173585e-07,
+      "loss": 0.2376,
+      "step": 131965
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.712286362287545e-07,
+      "loss": 0.2402,
+      "step": 131970
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.705731090332123e-07,
+      "loss": 0.2352,
+      "step": 131975
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.699178977392446e-07,
+      "loss": 0.2627,
+      "step": 131980
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.692630023553581e-07,
+      "loss": 0.2318,
+      "step": 131985
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.686084228900602e-07,
+      "loss": 0.2494,
+      "step": 131990
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.679541593518468e-07,
+      "loss": 0.2371,
+      "step": 131995
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.673002117492083e-07,
+      "loss": 0.262,
+      "step": 132000
+    },
+    {
+      "epoch": 4.64,
+      "eval_loss": 0.24853284657001495,
+      "eval_runtime": 10.5548,
+      "eval_samples_per_second": 9.474,
+      "eval_steps_per_second": 9.474,
+      "step": 132000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.666465800906435e-07,
+      "loss": 0.2598,
+      "step": 132005
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.659932643846345e-07,
+      "loss": 0.2609,
+      "step": 132010
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.653402646396717e-07,
+      "loss": 0.2532,
+      "step": 132015
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 6.646875808642233e-07,
+      "loss": 0.2528,
+      "step": 132020
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.640352130667716e-07,
+      "loss": 0.26,
+      "step": 132025
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.633831612557844e-07,
+      "loss": 0.2451,
+      "step": 132030
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.627314254397276e-07,
+      "loss": 0.2449,
+      "step": 132035
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.620800056270638e-07,
+      "loss": 0.2424,
+      "step": 132040
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.614289018262526e-07,
+      "loss": 0.2506,
+      "step": 132045
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.607781140457487e-07,
+      "loss": 0.2562,
+      "step": 132050
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.601276422940006e-07,
+      "loss": 0.2633,
+      "step": 132055
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.594774865794517e-07,
+      "loss": 0.2432,
+      "step": 132060
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.588276469105481e-07,
+      "loss": 0.2413,
+      "step": 132065
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.581781232957273e-07,
+      "loss": 0.2464,
+      "step": 132070
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.575289157434189e-07,
+      "loss": 0.2366,
+      "step": 132075
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.568800242620577e-07,
+      "loss": 0.2498,
+      "step": 132080
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.562314488600674e-07,
+      "loss": 0.2505,
+      "step": 132085
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.555831895458691e-07,
+      "loss": 0.256,
+      "step": 132090
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.549352463278757e-07,
+      "loss": 0.2644,
+      "step": 132095
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.542876192145053e-07,
+      "loss": 0.2679,
+      "step": 132100
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.536403082141679e-07,
+      "loss": 0.2297,
+      "step": 132105
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.529933133352678e-07,
+      "loss": 0.256,
+      "step": 132110
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.523466345861984e-07,
+      "loss": 0.2407,
+      "step": 132115
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.51700271975364e-07,
+      "loss": 0.2212,
+      "step": 132120
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.510542255111579e-07,
+      "loss": 0.2716,
+      "step": 132125
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.504084952019651e-07,
+      "loss": 0.2682,
+      "step": 132130
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.497630810561706e-07,
+      "loss": 0.2413,
+      "step": 132135
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.491179830821537e-07,
+      "loss": 0.2548,
+      "step": 132140
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.484732012882966e-07,
+      "loss": 0.2452,
+      "step": 132145
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.478287356829649e-07,
+      "loss": 0.2513,
+      "step": 132150
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.47184586274524e-07,
+      "loss": 0.2597,
+      "step": 132155
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.465407530713452e-07,
+      "loss": 0.2343,
+      "step": 132160
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.458972360817883e-07,
+      "loss": 0.2566,
+      "step": 132165
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.452540353141995e-07,
+      "loss": 0.249,
+      "step": 132170
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.446111507769414e-07,
+      "loss": 0.2572,
+      "step": 132175
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.439685824783548e-07,
+      "loss": 0.2377,
+      "step": 132180
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.433263304267855e-07,
+      "loss": 0.2539,
+      "step": 132185
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.426843946305716e-07,
+      "loss": 0.2579,
+      "step": 132190
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.420427750980479e-07,
+      "loss": 0.245,
+      "step": 132195
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.414014718375494e-07,
+      "loss": 0.2459,
+      "step": 132200
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.407604848574001e-07,
+      "loss": 0.2422,
+      "step": 132205
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.401198141659182e-07,
+      "loss": 0.2429,
+      "step": 132210
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.394794597714304e-07,
+      "loss": 0.2692,
+      "step": 132215
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.388394216822469e-07,
+      "loss": 0.2535,
+      "step": 132220
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.381996999066775e-07,
+      "loss": 0.2592,
+      "step": 132225
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.375602944530295e-07,
+      "loss": 0.247,
+      "step": 132230
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.369212053296076e-07,
+      "loss": 0.2495,
+      "step": 132235
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.362824325447076e-07,
+      "loss": 0.2421,
+      "step": 132240
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.35643976106623e-07,
+      "loss": 0.2519,
+      "step": 132245
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.350058360236443e-07,
+      "loss": 0.2391,
+      "step": 132250
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.343680123040597e-07,
+      "loss": 0.2558,
+      "step": 132255
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.337305049561481e-07,
+      "loss": 0.2499,
+      "step": 132260
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.330933139881895e-07,
+      "loss": 0.2328,
+      "step": 132265
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.32456439408452e-07,
+      "loss": 0.2415,
+      "step": 132270
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.318198812252124e-07,
+      "loss": 0.244,
+      "step": 132275
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.31183639446728e-07,
+      "loss": 0.2548,
+      "step": 132280
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.305477140812699e-07,
+      "loss": 0.2634,
+      "step": 132285
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.299121051370843e-07,
+      "loss": 0.2385,
+      "step": 132290
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.292768126224341e-07,
+      "loss": 0.2506,
+      "step": 132295
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.286418365455626e-07,
+      "loss": 0.2482,
+      "step": 132300
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 6.280071769147134e-07,
+      "loss": 0.2356,
+      "step": 132305
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.273728337381296e-07,
+      "loss": 0.2472,
+      "step": 132310
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.267388070240493e-07,
+      "loss": 0.2518,
+      "step": 132315
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.261050967807047e-07,
+      "loss": 0.2369,
+      "step": 132320
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.254717030163171e-07,
+      "loss": 0.2795,
+      "step": 132325
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.248386257391187e-07,
+      "loss": 0.2486,
+      "step": 132330
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.242058649573279e-07,
+      "loss": 0.247,
+      "step": 132335
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.235734206791577e-07,
+      "loss": 0.264,
+      "step": 132340
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.229412929128209e-07,
+      "loss": 0.2515,
+      "step": 132345
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.223094816665304e-07,
+      "loss": 0.2593,
+      "step": 132350
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.216779869484823e-07,
+      "loss": 0.2431,
+      "step": 132355
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.210468087668842e-07,
+      "loss": 0.2436,
+      "step": 132360
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.204159471299209e-07,
+      "loss": 0.2392,
+      "step": 132365
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.197854020457944e-07,
+      "loss": 0.2254,
+      "step": 132370
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.191551735226869e-07,
+      "loss": 0.2355,
+      "step": 132375
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.185252615687809e-07,
+      "loss": 0.2576,
+      "step": 132380
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.178956661922531e-07,
+      "loss": 0.2521,
+      "step": 132385
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.172663874012857e-07,
+      "loss": 0.2453,
+      "step": 132390
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.166374252040418e-07,
+      "loss": 0.2552,
+      "step": 132395
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.160087796086955e-07,
+      "loss": 0.234,
+      "step": 132400
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.15380450623404e-07,
+      "loss": 0.2648,
+      "step": 132405
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.147524382563275e-07,
+      "loss": 0.2542,
+      "step": 132410
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.141247425156204e-07,
+      "loss": 0.22,
+      "step": 132415
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.134973634094321e-07,
+      "loss": 0.2267,
+      "step": 132420
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.128703009459114e-07,
+      "loss": 0.25,
+      "step": 132425
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.122435551331962e-07,
+      "loss": 0.2679,
+      "step": 132430
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.116171259794273e-07,
+      "loss": 0.2352,
+      "step": 132435
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.109910134927344e-07,
+      "loss": 0.2364,
+      "step": 132440
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.103652176812524e-07,
+      "loss": 0.2653,
+      "step": 132445
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.097397385531029e-07,
+      "loss": 0.2623,
+      "step": 132450
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.091145761164125e-07,
+      "loss": 0.264,
+      "step": 132455
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.084897303792914e-07,
+      "loss": 0.2523,
+      "step": 132460
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.078652013498609e-07,
+      "loss": 0.2527,
+      "step": 132465
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.07240989036223e-07,
+      "loss": 0.2597,
+      "step": 132470
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.066170934464877e-07,
+      "loss": 0.2469,
+      "step": 132475
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.059935145887485e-07,
+      "loss": 0.2414,
+      "step": 132480
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.05370252471113e-07,
+      "loss": 0.2587,
+      "step": 132485
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.047473071016663e-07,
+      "loss": 0.26,
+      "step": 132490
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.041246784884991e-07,
+      "loss": 0.2351,
+      "step": 132495
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.035023666396938e-07,
+      "loss": 0.2501,
+      "step": 132500
+    },
+    {
+      "epoch": 4.66,
+      "eval_loss": 0.24850960075855255,
+      "eval_runtime": 10.5477,
+      "eval_samples_per_second": 9.481,
+      "eval_steps_per_second": 9.481,
+      "step": 132500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.028803715633302e-07,
+      "loss": 0.2474,
+      "step": 132505
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.022586932674906e-07,
+      "loss": 0.2318,
+      "step": 132510
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.016373317602436e-07,
+      "loss": 0.2425,
+      "step": 132515
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.010162870496522e-07,
+      "loss": 0.24,
+      "step": 132520
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 6.003955591437904e-07,
+      "loss": 0.2605,
+      "step": 132525
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.997751480507102e-07,
+      "loss": 0.2381,
+      "step": 132530
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.991550537784663e-07,
+      "loss": 0.2376,
+      "step": 132535
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.98535276335116e-07,
+      "loss": 0.2289,
+      "step": 132540
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.979158157287057e-07,
+      "loss": 0.2477,
+      "step": 132545
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.972966719672762e-07,
+      "loss": 0.2415,
+      "step": 132550
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.966778450588628e-07,
+      "loss": 0.2521,
+      "step": 132555
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.96059335011509e-07,
+      "loss": 0.2563,
+      "step": 132560
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.954411418332418e-07,
+      "loss": 0.255,
+      "step": 132565
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.948232655320879e-07,
+      "loss": 0.2468,
+      "step": 132570
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.942057061160688e-07,
+      "loss": 0.267,
+      "step": 132575
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.935884635932087e-07,
+      "loss": 0.2616,
+      "step": 132580
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.929715379715151e-07,
+      "loss": 0.2581,
+      "step": 132585
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 5.923549292590008e-07,
+      "loss": 0.2606,
+      "step": 132590
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.917386374636735e-07,
+      "loss": 0.2622,
+      "step": 132595
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.911226625935351e-07,
+      "loss": 0.236,
+      "step": 132600
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.905070046565819e-07,
+      "loss": 0.2401,
+      "step": 132605
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.898916636608077e-07,
+      "loss": 0.2449,
+      "step": 132610
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.892766396142058e-07,
+      "loss": 0.2527,
+      "step": 132615
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.886619325247561e-07,
+      "loss": 0.271,
+      "step": 132620
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.880475424004467e-07,
+      "loss": 0.2544,
+      "step": 132625
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.874334692492517e-07,
+      "loss": 0.2472,
+      "step": 132630
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.868197130791397e-07,
+      "loss": 0.2511,
+      "step": 132635
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.862062738980906e-07,
+      "loss": 0.2521,
+      "step": 132640
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.855931517140617e-07,
+      "loss": 0.249,
+      "step": 132645
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.849803465350162e-07,
+      "loss": 0.261,
+      "step": 132650
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.843678583689116e-07,
+      "loss": 0.2515,
+      "step": 132655
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.837556872236999e-07,
+      "loss": 0.248,
+      "step": 132660
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.831438331073275e-07,
+      "loss": 0.2506,
+      "step": 132665
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.825322960277435e-07,
+      "loss": 0.2607,
+      "step": 132670
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.819210759928833e-07,
+      "loss": 0.2541,
+      "step": 132675
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.813101730106879e-07,
+      "loss": 0.249,
+      "step": 132680
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.806995870890869e-07,
+      "loss": 0.2453,
+      "step": 132685
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.800893182360073e-07,
+      "loss": 0.2662,
+      "step": 132690
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.794793664593762e-07,
+      "loss": 0.2356,
+      "step": 132695
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.788697317671149e-07,
+      "loss": 0.2409,
+      "step": 132700
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.78260414167131e-07,
+      "loss": 0.2677,
+      "step": 132705
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.776514136673433e-07,
+      "loss": 0.257,
+      "step": 132710
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.770427302756564e-07,
+      "loss": 0.2392,
+      "step": 132715
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.764343639999725e-07,
+      "loss": 0.2663,
+      "step": 132720
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.758263148481963e-07,
+      "loss": 0.2483,
+      "step": 132725
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.752185828282159e-07,
+      "loss": 0.2377,
+      "step": 132730
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.746111679479277e-07,
+      "loss": 0.2485,
+      "step": 132735
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.740040702152172e-07,
+      "loss": 0.2491,
+      "step": 132740
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.733972896379669e-07,
+      "loss": 0.2482,
+      "step": 132745
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.727908262240539e-07,
+      "loss": 0.2355,
+      "step": 132750
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.721846799813524e-07,
+      "loss": 0.2578,
+      "step": 132755
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.715788509177367e-07,
+      "loss": 0.2684,
+      "step": 132760
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.709733390410698e-07,
+      "loss": 0.281,
+      "step": 132765
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.703681443592152e-07,
+      "loss": 0.2661,
+      "step": 132770
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.697632668800302e-07,
+      "loss": 0.2444,
+      "step": 132775
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.691587066113668e-07,
+      "loss": 0.2518,
+      "step": 132780
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.685544635610773e-07,
+      "loss": 0.256,
+      "step": 132785
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.679505377370109e-07,
+      "loss": 0.2404,
+      "step": 132790
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.673469291470029e-07,
+      "loss": 0.2286,
+      "step": 132795
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.667436377988944e-07,
+      "loss": 0.2347,
+      "step": 132800
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.661406637005151e-07,
+      "loss": 0.2382,
+      "step": 132805
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.655380068597005e-07,
+      "loss": 0.2313,
+      "step": 132810
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.649356672842721e-07,
+      "loss": 0.2519,
+      "step": 132815
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.643336449820485e-07,
+      "loss": 0.2517,
+      "step": 132820
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.637319399608487e-07,
+      "loss": 0.2417,
+      "step": 132825
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.631305522284858e-07,
+      "loss": 0.2561,
+      "step": 132830
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.625294817927673e-07,
+      "loss": 0.2748,
+      "step": 132835
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.619287286615011e-07,
+      "loss": 0.2655,
+      "step": 132840
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.613282928424806e-07,
+      "loss": 0.2467,
+      "step": 132845
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.607281743435112e-07,
+      "loss": 0.2592,
+      "step": 132850
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.601283731723806e-07,
+      "loss": 0.2365,
+      "step": 132855
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.595288893368744e-07,
+      "loss": 0.25,
+      "step": 132860
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.589297228447781e-07,
+      "loss": 0.2686,
+      "step": 132865
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.583308737038745e-07,
+      "loss": 0.2597,
+      "step": 132870
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 5.577323419219349e-07,
+      "loss": 0.2482,
+      "step": 132875
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.571341275067338e-07,
+      "loss": 0.2365,
+      "step": 132880
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.565362304660371e-07,
+      "loss": 0.2679,
+      "step": 132885
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.559386508076053e-07,
+      "loss": 0.2361,
+      "step": 132890
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.553413885392044e-07,
+      "loss": 0.2445,
+      "step": 132895
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.547444436685866e-07,
+      "loss": 0.2383,
+      "step": 132900
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.541478162035013e-07,
+      "loss": 0.2589,
+      "step": 132905
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.535515061516977e-07,
+      "loss": 0.2598,
+      "step": 132910
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.529555135209142e-07,
+      "loss": 0.2544,
+      "step": 132915
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.523598383188916e-07,
+      "loss": 0.2531,
+      "step": 132920
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.517644805533684e-07,
+      "loss": 0.2508,
+      "step": 132925
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.511694402320716e-07,
+      "loss": 0.2439,
+      "step": 132930
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.505747173627257e-07,
+      "loss": 0.2572,
+      "step": 132935
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.499803119530522e-07,
+      "loss": 0.2283,
+      "step": 132940
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.4938622401077e-07,
+      "loss": 0.2578,
+      "step": 132945
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.487924535435979e-07,
+      "loss": 0.2574,
+      "step": 132950
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.481990005592435e-07,
+      "loss": 0.2252,
+      "step": 132955
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.476058650654065e-07,
+      "loss": 0.2651,
+      "step": 132960
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.470130470697943e-07,
+      "loss": 0.2429,
+      "step": 132965
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.464205465801037e-07,
+      "loss": 0.2431,
+      "step": 132970
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.458283636040257e-07,
+      "loss": 0.2552,
+      "step": 132975
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.452364981492486e-07,
+      "loss": 0.2313,
+      "step": 132980
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.446449502234635e-07,
+      "loss": 0.246,
+      "step": 132985
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.440537198343449e-07,
+      "loss": 0.2323,
+      "step": 132990
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.434628069895698e-07,
+      "loss": 0.2413,
+      "step": 132995
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.428722116968127e-07,
+      "loss": 0.263,
+      "step": 133000
+    },
+    {
+      "epoch": 4.68,
+      "eval_loss": 0.24850255250930786,
+      "eval_runtime": 10.5511,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 133000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.422819339637453e-07,
+      "loss": 0.2391,
+      "step": 133005
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.41691973798028e-07,
+      "loss": 0.2412,
+      "step": 133010
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.411023312073188e-07,
+      "loss": 0.2447,
+      "step": 133015
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.405130061992808e-07,
+      "loss": 0.2358,
+      "step": 133020
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.399239987815607e-07,
+      "loss": 0.2476,
+      "step": 133025
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.393353089618081e-07,
+      "loss": 0.2601,
+      "step": 133030
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.387469367476638e-07,
+      "loss": 0.2679,
+      "step": 133035
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.38158882146772e-07,
+      "loss": 0.2602,
+      "step": 133040
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.375711451667653e-07,
+      "loss": 0.2363,
+      "step": 133045
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.369837258152766e-07,
+      "loss": 0.2532,
+      "step": 133050
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.363966240999329e-07,
+      "loss": 0.2625,
+      "step": 133055
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.358098400283534e-07,
+      "loss": 0.2409,
+      "step": 133060
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.352233736081652e-07,
+      "loss": 0.2583,
+      "step": 133065
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.346372248469761e-07,
+      "loss": 0.2374,
+      "step": 133070
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.340513937523994e-07,
+      "loss": 0.2515,
+      "step": 133075
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.33465880332043e-07,
+      "loss": 0.2455,
+      "step": 133080
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.328806845935091e-07,
+      "loss": 0.2345,
+      "step": 133085
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.322958065443944e-07,
+      "loss": 0.2455,
+      "step": 133090
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.317112461922902e-07,
+      "loss": 0.2656,
+      "step": 133095
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.311270035447957e-07,
+      "loss": 0.2337,
+      "step": 133100
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.305430786094856e-07,
+      "loss": 0.2505,
+      "step": 133105
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.29959471393951e-07,
+      "loss": 0.2346,
+      "step": 133110
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.293761819057663e-07,
+      "loss": 0.2234,
+      "step": 133115
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.287932101525061e-07,
+      "loss": 0.2616,
+      "step": 133120
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.282105561417366e-07,
+      "loss": 0.2709,
+      "step": 133125
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.276282198810267e-07,
+      "loss": 0.2789,
+      "step": 133130
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.270462013779371e-07,
+      "loss": 0.2462,
+      "step": 133135
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.264645006400254e-07,
+      "loss": 0.2529,
+      "step": 133140
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.258831176748413e-07,
+      "loss": 0.2318,
+      "step": 133145
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.253020524899343e-07,
+      "loss": 0.2744,
+      "step": 133150
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.248374291488805e-07,
+      "loss": 0.2483,
+      "step": 133155
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 5.242569359874882e-07,
+      "loss": 0.235,
+      "step": 133160
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.236767606274867e-07,
+      "loss": 0.2279,
+      "step": 133165
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.230969030764143e-07,
+      "loss": 0.2417,
+      "step": 133170
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.225173633417985e-07,
+      "loss": 0.2398,
+      "step": 133175
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.219381414311636e-07,
+      "loss": 0.2554,
+      "step": 133180
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.21359237352026e-07,
+      "loss": 0.2657,
+      "step": 133185
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.207806511119101e-07,
+      "loss": 0.2319,
+      "step": 133190
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.202023827183239e-07,
+      "loss": 0.242,
+      "step": 133195
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.19624432178778e-07,
+      "loss": 0.2349,
+      "step": 133200
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.190467995007747e-07,
+      "loss": 0.2436,
+      "step": 133205
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.184694846918164e-07,
+      "loss": 0.2547,
+      "step": 133210
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.178924877593971e-07,
+      "loss": 0.2507,
+      "step": 133215
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.17315808711008e-07,
+      "loss": 0.2255,
+      "step": 133220
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.167394475541404e-07,
+      "loss": 0.2469,
+      "step": 133225
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.16163404296277e-07,
+      "loss": 0.2552,
+      "step": 133230
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.155876789448955e-07,
+      "loss": 0.2525,
+      "step": 133235
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.150122715074701e-07,
+      "loss": 0.2423,
+      "step": 133240
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.144371819914757e-07,
+      "loss": 0.2567,
+      "step": 133245
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.138624104043754e-07,
+      "loss": 0.2347,
+      "step": 133250
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.132879567536386e-07,
+      "loss": 0.2455,
+      "step": 133255
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.127138210467175e-07,
+      "loss": 0.2377,
+      "step": 133260
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.1214000329107e-07,
+      "loss": 0.2489,
+      "step": 133265
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.115665034941486e-07,
+      "loss": 0.2669,
+      "step": 133270
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.109933216633972e-07,
+      "loss": 0.2368,
+      "step": 133275
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.104204578062571e-07,
+      "loss": 0.2481,
+      "step": 133280
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.098479119301697e-07,
+      "loss": 0.2422,
+      "step": 133285
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.092756840425678e-07,
+      "loss": 0.257,
+      "step": 133290
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.087037741508816e-07,
+      "loss": 0.2429,
+      "step": 133295
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.081321822625329e-07,
+      "loss": 0.2547,
+      "step": 133300
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.075609083849491e-07,
+      "loss": 0.2409,
+      "step": 133305
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.069899525255467e-07,
+      "loss": 0.2559,
+      "step": 133310
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.064193146917362e-07,
+      "loss": 0.2601,
+      "step": 133315
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.058489948909312e-07,
+      "loss": 0.24,
+      "step": 133320
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.052789931305341e-07,
+      "loss": 0.2504,
+      "step": 133325
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.047093094179472e-07,
+      "loss": 0.2488,
+      "step": 133330
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.041399437605649e-07,
+      "loss": 0.251,
+      "step": 133335
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.035708961657836e-07,
+      "loss": 0.2721,
+      "step": 133340
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.030021666409895e-07,
+      "loss": 0.2156,
+      "step": 133345
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.02433755193571e-07,
+      "loss": 0.2312,
+      "step": 133350
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.018656618309025e-07,
+      "loss": 0.266,
+      "step": 133355
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.012978865603646e-07,
+      "loss": 0.2396,
+      "step": 133360
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.007304293893262e-07,
+      "loss": 0.2459,
+      "step": 133365
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 5.001632903251619e-07,
+      "loss": 0.2543,
+      "step": 133370
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.99596469375227e-07,
+      "loss": 0.2621,
+      "step": 133375
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.990299665468879e-07,
+      "loss": 0.2688,
+      "step": 133380
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.984637818474996e-07,
+      "loss": 0.2269,
+      "step": 133385
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.978979152844093e-07,
+      "loss": 0.2381,
+      "step": 133390
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.973323668649665e-07,
+      "loss": 0.2326,
+      "step": 133395
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.967671365965181e-07,
+      "loss": 0.276,
+      "step": 133400
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.962022244863973e-07,
+      "loss": 0.2454,
+      "step": 133405
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.956376305419425e-07,
+      "loss": 0.2617,
+      "step": 133410
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.950733547704866e-07,
+      "loss": 0.2602,
+      "step": 133415
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.94509397179349e-07,
+      "loss": 0.2497,
+      "step": 133420
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.939457577758627e-07,
+      "loss": 0.2693,
+      "step": 133425
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.933824365673356e-07,
+      "loss": 0.2398,
+      "step": 133430
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.928194335610897e-07,
+      "loss": 0.2697,
+      "step": 133435
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.922567487644331e-07,
+      "loss": 0.2417,
+      "step": 133440
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.916943821846737e-07,
+      "loss": 0.2278,
+      "step": 133445
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.911323338291057e-07,
+      "loss": 0.2754,
+      "step": 133450
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.905706037050345e-07,
+      "loss": 0.2477,
+      "step": 133455
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.900091918197541e-07,
+      "loss": 0.2517,
+      "step": 133460
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.894480981805505e-07,
+      "loss": 0.2734,
+      "step": 133465
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.888873227947066e-07,
+      "loss": 0.2455,
+      "step": 133470
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.883268656695111e-07,
+      "loss": 0.2346,
+      "step": 133475
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.87766726812236e-07,
+      "loss": 0.2484,
+      "step": 133480
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.872069062301588e-07,
+      "loss": 0.2498,
+      "step": 133485
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.866474039305402e-07,
+      "loss": 0.2736,
+      "step": 133490
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.860882199206523e-07,
+      "loss": 0.252,
+      "step": 133495
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.85529354207756e-07,
+      "loss": 0.2395,
+      "step": 133500
+    },
+    {
+      "epoch": 4.7,
+      "eval_loss": 0.24847909808158875,
+      "eval_runtime": 10.5573,
+      "eval_samples_per_second": 9.472,
+      "eval_steps_per_second": 9.472,
+      "step": 133500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.849708067991038e-07,
+      "loss": 0.2376,
+      "step": 133505
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.844125777019481e-07,
+      "loss": 0.2373,
+      "step": 133510
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.838546669235416e-07,
+      "loss": 0.2411,
+      "step": 133515
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.832970744711257e-07,
+      "loss": 0.2565,
+      "step": 133520
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.827398003519363e-07,
+      "loss": 0.2392,
+      "step": 133525
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.821828445732147e-07,
+      "loss": 0.2737,
+      "step": 133530
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.816262071421912e-07,
+      "loss": 0.2578,
+      "step": 133535
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.810698880660935e-07,
+      "loss": 0.2449,
+      "step": 133540
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.805138873521436e-07,
+      "loss": 0.2407,
+      "step": 133545
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.799582050075635e-07,
+      "loss": 0.2536,
+      "step": 133550
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.794028410395695e-07,
+      "loss": 0.2539,
+      "step": 133555
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.788477954553672e-07,
+      "loss": 0.2721,
+      "step": 133560
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.782930682621644e-07,
+      "loss": 0.2498,
+      "step": 133565
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.777386594671668e-07,
+      "loss": 0.2588,
+      "step": 133570
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.771845690775739e-07,
+      "loss": 0.2659,
+      "step": 133575
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7663079710057455e-07,
+      "loss": 0.2421,
+      "step": 133580
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7607734354336296e-07,
+      "loss": 0.2476,
+      "step": 133585
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7552420841312786e-07,
+      "loss": 0.2486,
+      "step": 133590
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7497139171704954e-07,
+      "loss": 0.2392,
+      "step": 133595
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7441889346230284e-07,
+      "loss": 0.2418,
+      "step": 133600
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.738667136560654e-07,
+      "loss": 0.2407,
+      "step": 133605
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7331485230550363e-07,
+      "loss": 0.2377,
+      "step": 133610
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.727633094177869e-07,
+      "loss": 0.2626,
+      "step": 133615
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.722120850000733e-07,
+      "loss": 0.2364,
+      "step": 133620
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.71661179059521e-07,
+      "loss": 0.2398,
+      "step": 133625
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.711105916032854e-07,
+      "loss": 0.2548,
+      "step": 133630
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.70560322638508e-07,
+      "loss": 0.2547,
+      "step": 133635
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.7001037217234424e-07,
+      "loss": 0.2344,
+      "step": 133640
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6946074021193e-07,
+      "loss": 0.2463,
+      "step": 133645
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6891142676440125e-07,
+      "loss": 0.2588,
+      "step": 133650
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6836243183689123e-07,
+      "loss": 0.2377,
+      "step": 133655
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6781375543652475e-07,
+      "loss": 0.2665,
+      "step": 133660
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.672653975704322e-07,
+      "loss": 0.2636,
+      "step": 133665
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.667173582457329e-07,
+      "loss": 0.2751,
+      "step": 133670
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.661696374695379e-07,
+      "loss": 0.2446,
+      "step": 133675
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.656222352489609e-07,
+      "loss": 0.2516,
+      "step": 133680
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.650751515911128e-07,
+      "loss": 0.2605,
+      "step": 133685
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.645283865030936e-07,
+      "loss": 0.2383,
+      "step": 133690
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.639819399920031e-07,
+      "loss": 0.2682,
+      "step": 133695
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6343581206493844e-07,
+      "loss": 0.251,
+      "step": 133700
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.628900027289884e-07,
+      "loss": 0.2662,
+      "step": 133705
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.623445119912445e-07,
+      "loss": 0.2338,
+      "step": 133710
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.617993398587844e-07,
+      "loss": 0.2535,
+      "step": 133715
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.6125448633868586e-07,
+      "loss": 0.274,
+      "step": 133720
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.607099514380292e-07,
+      "loss": 0.253,
+      "step": 133725
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.601657351638811e-07,
+      "loss": 0.2562,
+      "step": 133730
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.59621837523308e-07,
+      "loss": 0.2455,
+      "step": 133735
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5907825852337106e-07,
+      "loss": 0.2697,
+      "step": 133740
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.585349981711312e-07,
+      "loss": 0.2471,
+      "step": 133745
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5799205647364116e-07,
+      "loss": 0.2394,
+      "step": 133750
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5744943343795364e-07,
+      "loss": 0.2628,
+      "step": 133755
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5690712907110465e-07,
+      "loss": 0.2439,
+      "step": 133760
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.563651433801469e-07,
+      "loss": 0.2465,
+      "step": 133765
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5582347637211367e-07,
+      "loss": 0.2499,
+      "step": 133770
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.55390372222117e-07,
+      "loss": 0.2433,
+      "step": 133775
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.548492788610637e-07,
+      "loss": 0.2583,
+      "step": 133780
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5430850420261904e-07,
+      "loss": 0.2323,
+      "step": 133785
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.537680482537998e-07,
+      "loss": 0.2464,
+      "step": 133790
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5322791102162807e-07,
+      "loss": 0.2542,
+      "step": 133795
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5268809251311496e-07,
+      "loss": 0.259,
+      "step": 133800
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5214859273527146e-07,
+      "loss": 0.2701,
+      "step": 133805
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5160941169509476e-07,
+      "loss": 0.2358,
+      "step": 133810
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.510705493995987e-07,
+      "loss": 0.2536,
+      "step": 133815
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.5053200585576937e-07,
+      "loss": 0.2644,
+      "step": 133820
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.49993781070604e-07,
+      "loss": 0.2591,
+      "step": 133825
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.494558750510913e-07,
+      "loss": 0.2571,
+      "step": 133830
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4891828780421194e-07,
+      "loss": 0.2387,
+      "step": 133835
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.483810193369492e-07,
+      "loss": 0.2582,
+      "step": 133840
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4784406965628077e-07,
+      "loss": 0.2423,
+      "step": 133845
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4730743876917614e-07,
+      "loss": 0.2634,
+      "step": 133850
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4677112668260467e-07,
+      "loss": 0.2468,
+      "step": 133855
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4623513340352753e-07,
+      "loss": 0.2499,
+      "step": 133860
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4569945893890576e-07,
+      "loss": 0.265,
+      "step": 133865
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.451641032956949e-07,
+      "loss": 0.2362,
+      "step": 133870
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.44629066480845e-07,
+      "loss": 0.2777,
+      "step": 133875
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4409434850130607e-07,
+      "loss": 0.2626,
+      "step": 133880
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.435599493640169e-07,
+      "loss": 0.2508,
+      "step": 133885
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4302586907591926e-07,
+      "loss": 0.2379,
+      "step": 133890
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4249210764394643e-07,
+      "loss": 0.2597,
+      "step": 133895
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.419586650750318e-07,
+      "loss": 0.2523,
+      "step": 133900
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.414255413760976e-07,
+      "loss": 0.2325,
+      "step": 133905
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.408927365540688e-07,
+      "loss": 0.2461,
+      "step": 133910
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.4036025061586216e-07,
+      "loss": 0.254,
+      "step": 133915
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3982808356839434e-07,
+      "loss": 0.2598,
+      "step": 133920
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3929623541856813e-07,
+      "loss": 0.244,
+      "step": 133925
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3876470617330026e-07,
+      "loss": 0.2638,
+      "step": 133930
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.382334958394824e-07,
+      "loss": 0.25,
+      "step": 133935
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.377026044240173e-07,
+      "loss": 0.2512,
+      "step": 133940
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3717203193379686e-07,
+      "loss": 0.2599,
+      "step": 133945
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.366417783757071e-07,
+      "loss": 0.2406,
+      "step": 133950
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3611184375663973e-07,
+      "loss": 0.2614,
+      "step": 133955
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3558222808346707e-07,
+      "loss": 0.2675,
+      "step": 133960
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3505293136307523e-07,
+      "loss": 0.2617,
+      "step": 133965
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.34523953602331e-07,
+      "loss": 0.2576,
+      "step": 133970
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.339952948081066e-07,
+      "loss": 0.2348,
+      "step": 133975
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3346695498725776e-07,
+      "loss": 0.2357,
+      "step": 133980
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3293893414665667e-07,
+      "loss": 0.2417,
+      "step": 133985
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.324112322931506e-07,
+      "loss": 0.2454,
+      "step": 133990
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.318838494335925e-07,
+      "loss": 0.2458,
+      "step": 133995
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.313567855748324e-07,
+      "loss": 0.2546,
+      "step": 134000
+    },
+    {
+      "epoch": 4.71,
+      "eval_loss": 0.24852041900157928,
+      "eval_runtime": 10.5387,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 134000
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3083004072371203e-07,
+      "loss": 0.2659,
+      "step": 134005
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.303036148870759e-07,
+      "loss": 0.2601,
+      "step": 134010
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.29777508071752e-07,
+      "loss": 0.2358,
+      "step": 134015
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2925172028457364e-07,
+      "loss": 0.2433,
+      "step": 134020
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2872625153237144e-07,
+      "loss": 0.2395,
+      "step": 134025
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.282011018219678e-07,
+      "loss": 0.2385,
+      "step": 134030
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2767627116017674e-07,
+      "loss": 0.2473,
+      "step": 134035
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.271517595538149e-07,
+      "loss": 0.2381,
+      "step": 134040
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.266275670096936e-07,
+      "loss": 0.243,
+      "step": 134045
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.261036935346213e-07,
+      "loss": 0.2607,
+      "step": 134050
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.255801391353953e-07,
+      "loss": 0.2362,
+      "step": 134055
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.250569038188129e-07,
+      "loss": 0.256,
+      "step": 134060
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2453398759167704e-07,
+      "loss": 0.2454,
+      "step": 134065
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.240113904607684e-07,
+      "loss": 0.2605,
+      "step": 134070
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.23489112432876e-07,
+      "loss": 0.2446,
+      "step": 134075
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2296715351478335e-07,
+      "loss": 0.2687,
+      "step": 134080
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.224455137132627e-07,
+      "loss": 0.2663,
+      "step": 134085
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.21924193035092e-07,
+      "loss": 0.2411,
+      "step": 134090
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.2140319148703535e-07,
+      "loss": 0.2379,
+      "step": 134095
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.208825090758622e-07,
+      "loss": 0.2565,
+      "step": 134100
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.203621458083312e-07,
+      "loss": 0.2397,
+      "step": 134105
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.19842101691198e-07,
+      "loss": 0.2508,
+      "step": 134110
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1932237673121545e-07,
+      "loss": 0.2467,
+      "step": 134115
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1880297093513654e-07,
+      "loss": 0.2712,
+      "step": 134120
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1828388430969755e-07,
+      "loss": 0.254,
+      "step": 134125
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.177651168616459e-07,
+      "loss": 0.2601,
+      "step": 134130
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.172466685977094e-07,
+      "loss": 0.2542,
+      "step": 134135
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.167285395246273e-07,
+      "loss": 0.2537,
+      "step": 134140
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1621072964912464e-07,
+      "loss": 0.2534,
+      "step": 134145
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1569323897792387e-07,
+      "loss": 0.2512,
+      "step": 134150
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.151760675177446e-07,
+      "loss": 0.2476,
+      "step": 134155
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1465921527530095e-07,
+      "loss": 0.2304,
+      "step": 134160
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.141426822573041e-07,
+      "loss": 0.2545,
+      "step": 134165
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.136264684704655e-07,
+      "loss": 0.2259,
+      "step": 134170
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1311057392147966e-07,
+      "loss": 0.248,
+      "step": 134175
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1259499861705243e-07,
+      "loss": 0.2331,
+      "step": 134180
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.120797425638756e-07,
+      "loss": 0.2626,
+      "step": 134185
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.115648057686411e-07,
+      "loss": 0.2415,
+      "step": 134190
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1105018823802966e-07,
+      "loss": 0.247,
+      "step": 134195
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.105358899787276e-07,
+      "loss": 0.2723,
+      "step": 134200
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1002191099741296e-07,
+      "loss": 0.2401,
+      "step": 134205
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.095082513007581e-07,
+      "loss": 0.2487,
+      "step": 134210
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.0899491089543276e-07,
+      "loss": 0.228,
+      "step": 134215
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.08481889788101e-07,
+      "loss": 0.238,
+      "step": 134220
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.0796918798542694e-07,
+      "loss": 0.2298,
+      "step": 134225
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.074568054940636e-07,
+      "loss": 0.2284,
+      "step": 134230
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.069447423206696e-07,
+      "loss": 0.2346,
+      "step": 134235
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.064329984718923e-07,
+      "loss": 0.2512,
+      "step": 134240
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.059215739543737e-07,
+      "loss": 0.239,
+      "step": 134245
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.0541046877475285e-07,
+      "loss": 0.2488,
+      "step": 134250
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.0489968293966896e-07,
+      "loss": 0.2513,
+      "step": 134255
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.0438921645575555e-07,
+      "loss": 0.2548,
+      "step": 134260
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.038790693296407e-07,
+      "loss": 0.2413,
+      "step": 134265
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.033692415679413e-07,
+      "loss": 0.265,
+      "step": 134270
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.028597331772854e-07,
+      "loss": 0.2508,
+      "step": 134275
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.023505441642872e-07,
+      "loss": 0.225,
+      "step": 134280
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.018416745355552e-07,
+      "loss": 0.2656,
+      "step": 134285
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.013331242976981e-07,
+      "loss": 0.2575,
+      "step": 134290
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.008248934573189e-07,
+      "loss": 0.2526,
+      "step": 134295
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.0031698202101784e-07,
+      "loss": 0.2413,
+      "step": 134300
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9980938999538975e-07,
+      "loss": 0.2438,
+      "step": 134305
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9930211738702374e-07,
+      "loss": 0.2287,
+      "step": 134310
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.987951642025062e-07,
+      "loss": 0.2725,
+      "step": 134315
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.982885304484235e-07,
+      "loss": 0.2506,
+      "step": 134320
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9778221613134823e-07,
+      "loss": 0.2348,
+      "step": 134325
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.972762212578557e-07,
+      "loss": 0.2513,
+      "step": 134330
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.967705458345211e-07,
+      "loss": 0.2566,
+      "step": 134335
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.962651898679087e-07,
+      "loss": 0.2325,
+      "step": 134340
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.957601533645744e-07,
+      "loss": 0.2445,
+      "step": 134345
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9525543633108233e-07,
+      "loss": 0.2519,
+      "step": 134350
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9475103877398e-07,
+      "loss": 0.2446,
+      "step": 134355
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9424696069982334e-07,
+      "loss": 0.2499,
+      "step": 134360
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.937432021151516e-07,
+      "loss": 0.255,
+      "step": 134365
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9323976302650946e-07,
+      "loss": 0.268,
+      "step": 134370
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9273664344043337e-07,
+      "loss": 0.2577,
+      "step": 134375
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9223384336345424e-07,
+      "loss": 0.2665,
+      "step": 134380
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.91731362802103e-07,
+      "loss": 0.26,
+      "step": 134385
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9122920176289934e-07,
+      "loss": 0.2487,
+      "step": 134390
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.9072736025236866e-07,
+      "loss": 0.2559,
+      "step": 134395
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.902258382770252e-07,
+      "loss": 0.2562,
+      "step": 134400
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.897246358433804e-07,
+      "loss": 0.2477,
+      "step": 134405
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.892237529579429e-07,
+      "loss": 0.2262,
+      "step": 134410
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8872318962721586e-07,
+      "loss": 0.2661,
+      "step": 134415
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.882229458576997e-07,
+      "loss": 0.2361,
+      "step": 134420
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.877230216558864e-07,
+      "loss": 0.2421,
+      "step": 134425
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.872234170282707e-07,
+      "loss": 0.2562,
+      "step": 134430
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.867241319813364e-07,
+      "loss": 0.2601,
+      "step": 134435
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8622516652156993e-07,
+      "loss": 0.2331,
+      "step": 134440
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.857265206554495e-07,
+      "loss": 0.2541,
+      "step": 134445
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8522819438944757e-07,
+      "loss": 0.2329,
+      "step": 134450
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8473018773003967e-07,
+      "loss": 0.2487,
+      "step": 134455
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8423250068368165e-07,
+      "loss": 0.2591,
+      "step": 134460
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8373513325684897e-07,
+      "loss": 0.25,
+      "step": 134465
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8323808545598915e-07,
+      "loss": 0.2557,
+      "step": 134470
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.82741357287561e-07,
+      "loss": 0.2413,
+      "step": 134475
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.822449487580121e-07,
+      "loss": 0.2602,
+      "step": 134480
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8174885987378725e-07,
+      "loss": 0.2503,
+      "step": 134485
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.812530906413342e-07,
+      "loss": 0.2533,
+      "step": 134490
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.80757641067081e-07,
+      "loss": 0.24,
+      "step": 134495
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.8026251115746427e-07,
+      "loss": 0.2486,
+      "step": 134500
+    },
+    {
+      "epoch": 4.73,
+      "eval_loss": 0.24848829209804535,
+      "eval_runtime": 10.5443,
+      "eval_samples_per_second": 9.484,
+      "eval_steps_per_second": 9.484,
+      "step": 134500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.7976770091891766e-07,
+      "loss": 0.2469,
+      "step": 134505
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.7927321035786113e-07,
+      "loss": 0.2598,
+      "step": 134510
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.7877903948071445e-07,
+      "loss": 0.2534,
+      "step": 134515
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.782851882938976e-07,
+      "loss": 0.2452,
+      "step": 134520
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.777916568038192e-07,
+      "loss": 0.2526,
+      "step": 134525
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.772984450168937e-07,
+      "loss": 0.2298,
+      "step": 134530
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.768055529395187e-07,
+      "loss": 0.2505,
+      "step": 134535
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.763129805780946e-07,
+      "loss": 0.2471,
+      "step": 134540
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.758207279390191e-07,
+      "loss": 0.2605,
+      "step": 134545
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.7532879502868433e-07,
+      "loss": 0.2401,
+      "step": 134550
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.7483718185347683e-07,
+      "loss": 0.2561,
+      "step": 134555
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.7434588841978034e-07,
+      "loss": 0.2412,
+      "step": 134560
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.738549147339732e-07,
+      "loss": 0.2506,
+      "step": 134565
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.733642608024307e-07,
+      "loss": 0.2368,
+      "step": 134570
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.7287392663152286e-07,
+      "loss": 0.2515,
+      "step": 134575
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 3.7238391222761957e-07,
+      "loss": 0.2606,
+      "step": 134580
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.7189421759708245e-07,
+      "loss": 0.2508,
+      "step": 134585
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.7140484274626466e-07,
+      "loss": 0.265,
+      "step": 134590
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.70915787681525e-07,
+      "loss": 0.263,
+      "step": 134595
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.7042705240921127e-07,
+      "loss": 0.2538,
+      "step": 134600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.699386369356711e-07,
+      "loss": 0.2524,
+      "step": 134605
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6945054126724945e-07,
+      "loss": 0.2637,
+      "step": 134610
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.689627654102773e-07,
+      "loss": 0.2613,
+      "step": 134615
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6847530937109133e-07,
+      "loss": 0.2424,
+      "step": 134620
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.679881731560225e-07,
+      "loss": 0.2353,
+      "step": 134625
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.675013567713936e-07,
+      "loss": 0.2516,
+      "step": 134630
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.670148602235246e-07,
+      "loss": 0.252,
+      "step": 134635
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.665286835187354e-07,
+      "loss": 0.2366,
+      "step": 134640
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6604282666333754e-07,
+      "loss": 0.2468,
+      "step": 134645
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6555728966363725e-07,
+      "loss": 0.2369,
+      "step": 134650
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6507207252594046e-07,
+      "loss": 0.2279,
+      "step": 134655
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6458717525654773e-07,
+      "loss": 0.2566,
+      "step": 134660
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.641025978617568e-07,
+      "loss": 0.2486,
+      "step": 134665
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6361834034785714e-07,
+      "loss": 0.2552,
+      "step": 134670
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.631344027211381e-07,
+      "loss": 0.2381,
+      "step": 134675
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6265078498788353e-07,
+      "loss": 0.2367,
+      "step": 134680
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6216748715437176e-07,
+      "loss": 0.2521,
+      "step": 134685
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.616845092268756e-07,
+      "loss": 0.2516,
+      "step": 134690
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.612018512116705e-07,
+      "loss": 0.2339,
+      "step": 134695
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.6071951311502374e-07,
+      "loss": 0.2278,
+      "step": 134700
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.602374949431941e-07,
+      "loss": 0.2603,
+      "step": 134705
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5975579670244065e-07,
+      "loss": 0.2373,
+      "step": 134710
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.592744183990221e-07,
+      "loss": 0.2669,
+      "step": 134715
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.587933600391863e-07,
+      "loss": 0.2454,
+      "step": 134720
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.583126216291782e-07,
+      "loss": 0.2355,
+      "step": 134725
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5783220317524293e-07,
+      "loss": 0.2398,
+      "step": 134730
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5735210468361426e-07,
+      "loss": 0.2453,
+      "step": 134735
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5687232616053166e-07,
+      "loss": 0.2474,
+      "step": 134740
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.563928676122208e-07,
+      "loss": 0.2625,
+      "step": 134745
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.559137290449044e-07,
+      "loss": 0.2214,
+      "step": 134750
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5543491046481084e-07,
+      "loss": 0.2508,
+      "step": 134755
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.549564118781518e-07,
+      "loss": 0.2269,
+      "step": 134760
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.544782332911417e-07,
+      "loss": 0.2652,
+      "step": 134765
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.540003747099896e-07,
+      "loss": 0.2392,
+      "step": 134770
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.535228361409015e-07,
+      "loss": 0.2644,
+      "step": 134775
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5304561759007813e-07,
+      "loss": 0.2172,
+      "step": 134780
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.525687190637117e-07,
+      "loss": 0.2304,
+      "step": 134785
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5209214056799997e-07,
+      "loss": 0.2548,
+      "step": 134790
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5161588210912697e-07,
+      "loss": 0.2445,
+      "step": 134795
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.511399436932766e-07,
+      "loss": 0.2448,
+      "step": 134800
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5066432532663005e-07,
+      "loss": 0.2474,
+      "step": 134805
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.5018902701536296e-07,
+      "loss": 0.2618,
+      "step": 134810
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4971404876564806e-07,
+      "loss": 0.2629,
+      "step": 134815
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4923939058364994e-07,
+      "loss": 0.2302,
+      "step": 134820
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4876505247553315e-07,
+      "loss": 0.237,
+      "step": 134825
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4829103444745383e-07,
+      "loss": 0.242,
+      "step": 134830
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4781733650557366e-07,
+      "loss": 0.2565,
+      "step": 134835
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.4734395865603787e-07,
+      "loss": 0.2636,
+      "step": 134840
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.468709009049914e-07,
+      "loss": 0.2631,
+      "step": 134845
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.463981632585822e-07,
+      "loss": 0.2821,
+      "step": 134850
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.459257457229442e-07,
+      "loss": 0.2558,
+      "step": 134855
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.454536483042142e-07,
+      "loss": 0.2601,
+      "step": 134860
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 3.449818710085179e-07,
+      "loss": 0.26,
+      "step": 134865
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4451041384198635e-07,
+      "loss": 0.2316,
+      "step": 134870
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.440392768107342e-07,
+      "loss": 0.254,
+      "step": 134875
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4356845992088715e-07,
+      "loss": 0.2416,
+      "step": 134880
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.430979631785514e-07,
+      "loss": 0.2333,
+      "step": 134885
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4262778658984416e-07,
+      "loss": 0.2421,
+      "step": 134890
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.421579301608607e-07,
+      "loss": 0.2682,
+      "step": 134895
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4168839389770725e-07,
+      "loss": 0.2293,
+      "step": 134900
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.4121917780648163e-07,
+      "loss": 0.2583,
+      "step": 134905
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.407502818932734e-07,
+      "loss": 0.2395,
+      "step": 134910
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.402817061641722e-07,
+      "loss": 0.2444,
+      "step": 134915
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.398134506252593e-07,
+      "loss": 0.2272,
+      "step": 134920
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.393455152826186e-07,
+      "loss": 0.2682,
+      "step": 134925
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.388779001423259e-07,
+      "loss": 0.2648,
+      "step": 134930
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3841060521044856e-07,
+      "loss": 0.2421,
+      "step": 134935
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3794363049305667e-07,
+      "loss": 0.2558,
+      "step": 134940
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3747697599621764e-07,
+      "loss": 0.2466,
+      "step": 134945
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3701064172598497e-07,
+      "loss": 0.2599,
+      "step": 134950
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.365446276884149e-07,
+      "loss": 0.2467,
+      "step": 134955
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3607893388955815e-07,
+      "loss": 0.2531,
+      "step": 134960
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3561356033546266e-07,
+      "loss": 0.2548,
+      "step": 134965
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.351485070321736e-07,
+      "loss": 0.2526,
+      "step": 134970
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.34683773985725e-07,
+      "loss": 0.2468,
+      "step": 134975
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.34219361202151e-07,
+      "loss": 0.232,
+      "step": 134980
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3375526868748284e-07,
+      "loss": 0.2708,
+      "step": 134985
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3329149644774904e-07,
+      "loss": 0.2592,
+      "step": 134990
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3282804448896697e-07,
+      "loss": 0.2699,
+      "step": 134995
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3236491281715685e-07,
+      "loss": 0.2617,
+      "step": 135000
+    },
+    {
+      "epoch": 4.75,
+      "eval_loss": 0.24845683574676514,
+      "eval_runtime": 10.5389,
+      "eval_samples_per_second": 9.489,
+      "eval_steps_per_second": 9.489,
+      "step": 135000
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3190210143833336e-07,
+      "loss": 0.235,
+      "step": 135005
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3143961035850545e-07,
+      "loss": 0.2608,
+      "step": 135010
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.30977439583674e-07,
+      "loss": 0.2415,
+      "step": 135015
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3051558911984247e-07,
+      "loss": 0.2501,
+      "step": 135020
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.3005405897301165e-07,
+      "loss": 0.2308,
+      "step": 135025
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.295928491491684e-07,
+      "loss": 0.2542,
+      "step": 135030
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2913195965430245e-07,
+      "loss": 0.2402,
+      "step": 135035
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2867139049440334e-07,
+      "loss": 0.246,
+      "step": 135040
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.282111416754441e-07,
+      "loss": 0.2597,
+      "step": 135045
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.277512132034033e-07,
+      "loss": 0.2654,
+      "step": 135050
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2729160508425396e-07,
+      "loss": 0.2464,
+      "step": 135055
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.268323173239635e-07,
+      "loss": 0.2554,
+      "step": 135060
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2637334992849656e-07,
+      "loss": 0.2478,
+      "step": 135065
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.259147029038123e-07,
+      "loss": 0.2364,
+      "step": 135070
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2545637625586154e-07,
+      "loss": 0.2548,
+      "step": 135075
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2499836999059786e-07,
+      "loss": 0.2601,
+      "step": 135080
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.24540684113972e-07,
+      "loss": 0.2602,
+      "step": 135085
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2408331863192363e-07,
+      "loss": 0.2577,
+      "step": 135090
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2362627355038975e-07,
+      "loss": 0.241,
+      "step": 135095
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.231695488753045e-07,
+      "loss": 0.2389,
+      "step": 135100
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2271314461260195e-07,
+      "loss": 0.2322,
+      "step": 135105
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2225706076820796e-07,
+      "loss": 0.2624,
+      "step": 135110
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2180129734804e-07,
+      "loss": 0.2631,
+      "step": 135115
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2134585435802113e-07,
+      "loss": 0.2308,
+      "step": 135120
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.208907318040605e-07,
+      "loss": 0.2534,
+      "step": 135125
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.2043592969207005e-07,
+      "loss": 0.22,
+      "step": 135130
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.1998144802795616e-07,
+      "loss": 0.2446,
+      "step": 135135
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.195272868176169e-07,
+      "loss": 0.2606,
+      "step": 135140
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.190734460669531e-07,
+      "loss": 0.2732,
+      "step": 135145
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 3.186199257818517e-07,
+      "loss": 0.2327,
+      "step": 135150
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.1816672596820527e-07,
+      "loss": 0.2567,
+      "step": 135155
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.1771384663190075e-07,
+      "loss": 0.2659,
+      "step": 135160
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.172612877788139e-07,
+      "loss": 0.2662,
+      "step": 135165
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.168090494148207e-07,
+      "loss": 0.2372,
+      "step": 135170
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.163571315457997e-07,
+      "loss": 0.268,
+      "step": 135175
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.1590553417761294e-07,
+      "loss": 0.2421,
+      "step": 135180
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.154542573161251e-07,
+      "loss": 0.2636,
+      "step": 135185
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.150033009671927e-07,
+      "loss": 0.2537,
+      "step": 135190
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.145526651366776e-07,
+      "loss": 0.2599,
+      "step": 135195
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.1410234983042796e-07,
+      "loss": 0.2666,
+      "step": 135200
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.136523550542891e-07,
+      "loss": 0.2494,
+      "step": 135205
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.1320268081410365e-07,
+      "loss": 0.2346,
+      "step": 135210
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.127533271157168e-07,
+      "loss": 0.2193,
+      "step": 135215
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.123042939649573e-07,
+      "loss": 0.2747,
+      "step": 135220
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.1185558136765656e-07,
+      "loss": 0.2646,
+      "step": 135225
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.114071893296378e-07,
+      "loss": 0.2353,
+      "step": 135230
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.1095911785672946e-07,
+      "loss": 0.2685,
+      "step": 135235
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.105113669547466e-07,
+      "loss": 0.2474,
+      "step": 135240
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.1006393662950106e-07,
+      "loss": 0.2549,
+      "step": 135245
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.096168268868049e-07,
+      "loss": 0.2491,
+      "step": 135250
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.091700377324647e-07,
+      "loss": 0.242,
+      "step": 135255
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0872356917227565e-07,
+      "loss": 0.2686,
+      "step": 135260
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.082774212120415e-07,
+      "loss": 0.2521,
+      "step": 135265
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.078315938575521e-07,
+      "loss": 0.2656,
+      "step": 135270
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.073860871145945e-07,
+      "loss": 0.2623,
+      "step": 135275
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0694090098895846e-07,
+      "loss": 0.2488,
+      "step": 135280
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.064960354864199e-07,
+      "loss": 0.2451,
+      "step": 135285
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0605149061275763e-07,
+      "loss": 0.2686,
+      "step": 135290
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.056072663737447e-07,
+      "loss": 0.2509,
+      "step": 135295
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0516336277514323e-07,
+      "loss": 0.2616,
+      "step": 135300
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.047197798227236e-07,
+      "loss": 0.257,
+      "step": 135305
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.042765175222423e-07,
+      "loss": 0.2525,
+      "step": 135310
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0383357587945307e-07,
+      "loss": 0.2538,
+      "step": 135315
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0339095490011236e-07,
+      "loss": 0.2413,
+      "step": 135320
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0294865458996293e-07,
+      "loss": 0.2461,
+      "step": 135325
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0250667495475005e-07,
+      "loss": 0.2333,
+      "step": 135330
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.020650160002109e-07,
+      "loss": 0.2433,
+      "step": 135335
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0162367773208244e-07,
+      "loss": 0.236,
+      "step": 135340
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.011826601560935e-07,
+      "loss": 0.262,
+      "step": 135345
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.007419632779701e-07,
+      "loss": 0.2551,
+      "step": 135350
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 3.0030158710343537e-07,
+      "loss": 0.2543,
+      "step": 135355
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9986153163820696e-07,
+      "loss": 0.2461,
+      "step": 135360
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.994217968879942e-07,
+      "loss": 0.2425,
+      "step": 135365
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9898238285851756e-07,
+      "loss": 0.2627,
+      "step": 135370
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.985432895554724e-07,
+      "loss": 0.2568,
+      "step": 135375
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9810451698456254e-07,
+      "loss": 0.2551,
+      "step": 135380
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.97666065151489e-07,
+      "loss": 0.2592,
+      "step": 135385
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.972279340619416e-07,
+      "loss": 0.2342,
+      "step": 135390
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.967901237216103e-07,
+      "loss": 0.2552,
+      "step": 135395
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9635263413617664e-07,
+      "loss": 0.2617,
+      "step": 135400
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9591546531132773e-07,
+      "loss": 0.2556,
+      "step": 135405
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9547861725273405e-07,
+      "loss": 0.2481,
+      "step": 135410
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9504208996606885e-07,
+      "loss": 0.281,
+      "step": 135415
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.946058834569998e-07,
+      "loss": 0.2579,
+      "step": 135420
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9416999773119456e-07,
+      "loss": 0.2726,
+      "step": 135425
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.937344327943098e-07,
+      "loss": 0.2413,
+      "step": 135430
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9329918865199925e-07,
+      "loss": 0.2489,
+      "step": 135435
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.928642653099167e-07,
+      "loss": 0.245,
+      "step": 135440
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.924296627737133e-07,
+      "loss": 0.2226,
+      "step": 135445
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.9199538104902335e-07,
+      "loss": 0.2536,
+      "step": 135450
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.9156142014149237e-07,
+      "loss": 0.2443,
+      "step": 135455
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.911277800567519e-07,
+      "loss": 0.2464,
+      "step": 135460
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.906944608004336e-07,
+      "loss": 0.2466,
+      "step": 135465
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.9026146237816354e-07,
+      "loss": 0.2641,
+      "step": 135470
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.89828784795565e-07,
+      "loss": 0.2644,
+      "step": 135475
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8939642805825564e-07,
+      "loss": 0.271,
+      "step": 135480
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.889643921718449e-07,
+      "loss": 0.2541,
+      "step": 135485
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.885326771419505e-07,
+      "loss": 0.2691,
+      "step": 135490
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8810128297417083e-07,
+      "loss": 0.2418,
+      "step": 135495
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.876702096741124e-07,
+      "loss": 0.262,
+      "step": 135500
+    },
+    {
+      "epoch": 4.77,
+      "eval_loss": 0.24846675992012024,
+      "eval_runtime": 10.5434,
+      "eval_samples_per_second": 9.485,
+      "eval_steps_per_second": 9.485,
+      "step": 135500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.87239457247368e-07,
+      "loss": 0.2446,
+      "step": 135505
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8680902569953595e-07,
+      "loss": 0.2705,
+      "step": 135510
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8637891503620064e-07,
+      "loss": 0.2406,
+      "step": 135515
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8594912526294925e-07,
+      "loss": 0.2783,
+      "step": 135520
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.855196563853607e-07,
+      "loss": 0.2388,
+      "step": 135525
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.850905084090111e-07,
+      "loss": 0.2583,
+      "step": 135530
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8466168133947367e-07,
+      "loss": 0.2553,
+      "step": 135535
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.842331751823135e-07,
+      "loss": 0.2385,
+      "step": 135540
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8380498994310113e-07,
+      "loss": 0.2284,
+      "step": 135545
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8337712562739316e-07,
+      "loss": 0.2567,
+      "step": 135550
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8294958224074073e-07,
+      "loss": 0.2624,
+      "step": 135555
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8252235978870054e-07,
+      "loss": 0.2405,
+      "step": 135560
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.820954582768209e-07,
+      "loss": 0.2596,
+      "step": 135565
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.816688777106391e-07,
+      "loss": 0.2357,
+      "step": 135570
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8124261809569785e-07,
+      "loss": 0.2375,
+      "step": 135575
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.8081667943752885e-07,
+      "loss": 0.2561,
+      "step": 135580
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.803910617416666e-07,
+      "loss": 0.2425,
+      "step": 135585
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7996576501363445e-07,
+      "loss": 0.239,
+      "step": 135590
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.795407892589558e-07,
+      "loss": 0.2516,
+      "step": 135595
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.791161344831511e-07,
+      "loss": 0.2466,
+      "step": 135600
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7869180069172997e-07,
+      "loss": 0.2325,
+      "step": 135605
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.782677878902018e-07,
+      "loss": 0.2447,
+      "step": 135610
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.778440960840761e-07,
+      "loss": 0.2456,
+      "step": 135615
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.774207252788541e-07,
+      "loss": 0.2659,
+      "step": 135620
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.769976754800313e-07,
+      "loss": 0.2339,
+      "step": 135625
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7657494669309783e-07,
+      "loss": 0.2383,
+      "step": 135630
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7615253892354643e-07,
+      "loss": 0.2633,
+      "step": 135635
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.757304521768617e-07,
+      "loss": 0.2628,
+      "step": 135640
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.753086864585197e-07,
+      "loss": 0.2574,
+      "step": 135645
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.748872417740023e-07,
+      "loss": 0.2639,
+      "step": 135650
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7446611812877997e-07,
+      "loss": 0.2581,
+      "step": 135655
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.740453155283207e-07,
+      "loss": 0.2548,
+      "step": 135660
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7362483397808667e-07,
+      "loss": 0.255,
+      "step": 135665
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.732046734835403e-07,
+      "loss": 0.2429,
+      "step": 135670
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7278483405013545e-07,
+      "loss": 0.2406,
+      "step": 135675
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7236531568332334e-07,
+      "loss": 0.2339,
+      "step": 135680
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.7194611838855243e-07,
+      "loss": 0.2244,
+      "step": 135685
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.715272421712628e-07,
+      "loss": 0.2684,
+      "step": 135690
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.711086870368973e-07,
+      "loss": 0.2579,
+      "step": 135695
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.706904529908849e-07,
+      "loss": 0.2527,
+      "step": 135700
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.70272540038663e-07,
+      "loss": 0.2475,
+      "step": 135705
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6985494818565215e-07,
+      "loss": 0.2505,
+      "step": 135710
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6943767743728145e-07,
+      "loss": 0.2667,
+      "step": 135715
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6902072779896314e-07,
+      "loss": 0.2407,
+      "step": 135720
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.686040992761096e-07,
+      "loss": 0.2654,
+      "step": 135725
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6818779187413324e-07,
+      "loss": 0.2428,
+      "step": 135730
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.677718055984435e-07,
+      "loss": 0.249,
+      "step": 135735
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6735614045443615e-07,
+      "loss": 0.2608,
+      "step": 135740
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.669407964475068e-07,
+      "loss": 0.2707,
+      "step": 135745
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.66525773583054e-07,
+      "loss": 0.2643,
+      "step": 135750
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.661110718664622e-07,
+      "loss": 0.2534,
+      "step": 135755
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.656966913031217e-07,
+      "loss": 0.2655,
+      "step": 135760
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6528263189840583e-07,
+      "loss": 0.2535,
+      "step": 135765
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6486889365769374e-07,
+      "loss": 0.2756,
+      "step": 135770
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.644554765863616e-07,
+      "loss": 0.2767,
+      "step": 135775
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6404238068977184e-07,
+      "loss": 0.2491,
+      "step": 135780
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6362960597328954e-07,
+      "loss": 0.2487,
+      "step": 135785
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6321715244227443e-07,
+      "loss": 0.2472,
+      "step": 135790
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6280502010208594e-07,
+      "loss": 0.2413,
+      "step": 135795
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6239320895806997e-07,
+      "loss": 0.2438,
+      "step": 135800
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6198171901557213e-07,
+      "loss": 0.2769,
+      "step": 135805
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.615705502799437e-07,
+      "loss": 0.2551,
+      "step": 135810
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.611597027565138e-07,
+      "loss": 0.2648,
+      "step": 135815
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6074917645062537e-07,
+      "loss": 0.2381,
+      "step": 135820
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6033897136760186e-07,
+      "loss": 0.2514,
+      "step": 135825
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.599290875127752e-07,
+      "loss": 0.2506,
+      "step": 135830
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5951952489146613e-07,
+      "loss": 0.2424,
+      "step": 135835
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5911028350898704e-07,
+      "loss": 0.2303,
+      "step": 135840
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5870136337065863e-07,
+      "loss": 0.2387,
+      "step": 135845
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5829276448179063e-07,
+      "loss": 0.284,
+      "step": 135850
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.578844868476815e-07,
+      "loss": 0.2156,
+      "step": 135855
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.574765304736382e-07,
+      "loss": 0.2494,
+      "step": 135860
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.570688953649564e-07,
+      "loss": 0.2516,
+      "step": 135865
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5666158152692907e-07,
+      "loss": 0.2611,
+      "step": 135870
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.562545889648438e-07,
+      "loss": 0.2535,
+      "step": 135875
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.558479176839851e-07,
+      "loss": 0.2218,
+      "step": 135880
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.554415676896349e-07,
+      "loss": 0.2752,
+      "step": 135885
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.550355389870723e-07,
+      "loss": 0.2576,
+      "step": 135890
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.546298315815626e-07,
+      "loss": 0.2479,
+      "step": 135895
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5422444547837376e-07,
+      "loss": 0.2629,
+      "step": 135900
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5381938068277654e-07,
+      "loss": 0.2659,
+      "step": 135905
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.534146372000251e-07,
+      "loss": 0.2494,
+      "step": 135910
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.530102150353764e-07,
+      "loss": 0.2525,
+      "step": 135915
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5260611419408163e-07,
+      "loss": 0.2237,
+      "step": 135920
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.52202334681384e-07,
+      "loss": 0.2504,
+      "step": 135925
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5179887650253476e-07,
+      "loss": 0.232,
+      "step": 135930
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.513957396627631e-07,
+      "loss": 0.253,
+      "step": 135935
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5099292416730925e-07,
+      "loss": 0.231,
+      "step": 135940
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5059043002140235e-07,
+      "loss": 0.2375,
+      "step": 135945
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.5018825723026884e-07,
+      "loss": 0.2539,
+      "step": 135950
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.4978640579912947e-07,
+      "loss": 0.2531,
+      "step": 135955
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.4938487573320234e-07,
+      "loss": 0.2811,
+      "step": 135960
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.489836670377027e-07,
+      "loss": 0.2536,
+      "step": 135965
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.485827797178375e-07,
+      "loss": 0.2492,
+      "step": 135970
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.481822137788137e-07,
+      "loss": 0.2356,
+      "step": 135975
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.4778196922582984e-07,
+      "loss": 0.2409,
+      "step": 135980
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.4738204606408465e-07,
+      "loss": 0.2646,
+      "step": 135985
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.469824442987739e-07,
+      "loss": 0.2426,
+      "step": 135990
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.465831639350824e-07,
+      "loss": 0.2461,
+      "step": 135995
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.4618420497819484e-07,
+      "loss": 0.2414,
+      "step": 136000
+    },
+    {
+      "epoch": 4.78,
+      "eval_loss": 0.2484423816204071,
+      "eval_runtime": 10.5512,
+      "eval_samples_per_second": 9.478,
+      "eval_steps_per_second": 9.478,
+      "step": 136000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.457855674332932e-07,
+      "loss": 0.2498,
+      "step": 136005
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.453872513055511e-07,
+      "loss": 0.2329,
+      "step": 136010
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.449892566001394e-07,
+      "loss": 0.2374,
+      "step": 136015
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.4459158332223173e-07,
+      "loss": 0.2284,
+      "step": 136020
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.4419423147698793e-07,
+      "loss": 0.244,
+      "step": 136025
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.437972010695649e-07,
+      "loss": 0.2669,
+      "step": 136030
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.434004921051197e-07,
+      "loss": 0.2353,
+      "step": 136035
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.4300410458880654e-07,
+      "loss": 0.2529,
+      "step": 136040
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.426080385257684e-07,
+      "loss": 0.26,
+      "step": 136045
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.422122939211513e-07,
+      "loss": 0.2691,
+      "step": 136050
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.418168707800872e-07,
+      "loss": 0.2767,
+      "step": 136055
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.414217691077192e-07,
+      "loss": 0.2778,
+      "step": 136060
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.41026988909171e-07,
+      "loss": 0.2329,
+      "step": 136065
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.40632530189569e-07,
+      "loss": 0.2373,
+      "step": 136070
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.402383929540397e-07,
+      "loss": 0.2503,
+      "step": 136075
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.398445772076957e-07,
+      "loss": 0.2404,
+      "step": 136080
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.394510829556523e-07,
+      "loss": 0.2588,
+      "step": 136085
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.390579102030166e-07,
+      "loss": 0.2546,
+      "step": 136090
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3866505895489833e-07,
+      "loss": 0.2649,
+      "step": 136095
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.382725292163962e-07,
+      "loss": 0.2536,
+      "step": 136100
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3788032099260893e-07,
+      "loss": 0.2585,
+      "step": 136105
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.374884342886241e-07,
+      "loss": 0.237,
+      "step": 136110
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.370968691095321e-07,
+      "loss": 0.2487,
+      "step": 136115
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3670562546042053e-07,
+      "loss": 0.2351,
+      "step": 136120
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3631470334636584e-07,
+      "loss": 0.2577,
+      "step": 136125
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.359241027724418e-07,
+      "loss": 0.2367,
+      "step": 136130
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.355338237437277e-07,
+      "loss": 0.2543,
+      "step": 136135
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3514386626528328e-07,
+      "loss": 0.2224,
+      "step": 136140
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3475423034217402e-07,
+      "loss": 0.256,
+      "step": 136145
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3436491597946253e-07,
+      "loss": 0.2579,
+      "step": 136150
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3397592318220308e-07,
+      "loss": 0.2449,
+      "step": 136155
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3358725195544162e-07,
+      "loss": 0.2583,
+      "step": 136160
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3319890230422968e-07,
+      "loss": 0.2448,
+      "step": 136165
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.328108742336077e-07,
+      "loss": 0.2609,
+      "step": 136170
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3242316774861604e-07,
+      "loss": 0.2442,
+      "step": 136175
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3203578285428406e-07,
+      "loss": 0.2456,
+      "step": 136180
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.316487195556466e-07,
+      "loss": 0.2538,
+      "step": 136185
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3126197785772742e-07,
+      "loss": 0.2818,
+      "step": 136190
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3087555776554747e-07,
+      "loss": 0.2886,
+      "step": 136195
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3048945928412502e-07,
+      "loss": 0.2577,
+      "step": 136200
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.3010368241847268e-07,
+      "loss": 0.2543,
+      "step": 136205
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2971822717360038e-07,
+      "loss": 0.2468,
+      "step": 136210
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2933309355451515e-07,
+      "loss": 0.253,
+      "step": 136215
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2894828156621305e-07,
+      "loss": 0.2237,
+      "step": 136220
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2856379121369288e-07,
+      "loss": 0.2673,
+      "step": 136225
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.281796225019478e-07,
+      "loss": 0.2708,
+      "step": 136230
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2779577543596276e-07,
+      "loss": 0.2639,
+      "step": 136235
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.274122500207282e-07,
+      "loss": 0.2538,
+      "step": 136240
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2702904626121514e-07,
+      "loss": 0.2511,
+      "step": 136245
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.266461641624057e-07,
+      "loss": 0.2426,
+      "step": 136250
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2626360372927092e-07,
+      "loss": 0.269,
+      "step": 136255
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2588136496677625e-07,
+      "loss": 0.2443,
+      "step": 136260
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2549944787988163e-07,
+      "loss": 0.2544,
+      "step": 136265
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2511785247355533e-07,
+      "loss": 0.2678,
+      "step": 136270
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2473657875274334e-07,
+      "loss": 0.2303,
+      "step": 136275
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2435562672240006e-07,
+      "loss": 0.2569,
+      "step": 136280
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.2397499638747155e-07,
+      "loss": 0.2722,
+      "step": 136285
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2359468775289827e-07,
+      "loss": 0.2658,
+      "step": 136290
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2321470082362072e-07,
+      "loss": 0.2705,
+      "step": 136295
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2283503560457108e-07,
+      "loss": 0.2751,
+      "step": 136300
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2245569210068152e-07,
+      "loss": 0.2432,
+      "step": 136305
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2207667031687307e-07,
+      "loss": 0.246,
+      "step": 136310
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2169797025807239e-07,
+      "loss": 0.2459,
+      "step": 136315
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2131959192919215e-07,
+      "loss": 0.2376,
+      "step": 136320
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2094153533514794e-07,
+      "loss": 0.253,
+      "step": 136325
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2056380048084968e-07,
+      "loss": 0.2476,
+      "step": 136330
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.2018638737119902e-07,
+      "loss": 0.2345,
+      "step": 136335
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1980929601109756e-07,
+      "loss": 0.2509,
+      "step": 136340
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1943252640544143e-07,
+      "loss": 0.2467,
+      "step": 136345
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1905607855912115e-07,
+      "loss": 0.2424,
+      "step": 136350
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.186799524770272e-07,
+      "loss": 0.2398,
+      "step": 136355
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1830414816404466e-07,
+      "loss": 0.2574,
+      "step": 136360
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.179286656250501e-07,
+      "loss": 0.2354,
+      "step": 136365
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.175535048649202e-07,
+      "loss": 0.2391,
+      "step": 136370
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.171786658885233e-07,
+      "loss": 0.2304,
+      "step": 136375
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1680414870073053e-07,
+      "loss": 0.2502,
+      "step": 136380
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.164299533064046e-07,
+      "loss": 0.2468,
+      "step": 136385
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1605607971040276e-07,
+      "loss": 0.2589,
+      "step": 136390
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.156825279175767e-07,
+      "loss": 0.2593,
+      "step": 136395
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1530929793278088e-07,
+      "loss": 0.2649,
+      "step": 136400
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.149363897608614e-07,
+      "loss": 0.2745,
+      "step": 136405
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.145638034066588e-07,
+      "loss": 0.2492,
+      "step": 136410
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.141915388750082e-07,
+      "loss": 0.2456,
+      "step": 136415
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1381959617074732e-07,
+      "loss": 0.2561,
+      "step": 136420
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1344797529870565e-07,
+      "loss": 0.2624,
+      "step": 136425
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.130766762637071e-07,
+      "loss": 0.2427,
+      "step": 136430
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1270569907057004e-07,
+      "loss": 0.2615,
+      "step": 136435
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.123350437241156e-07,
+      "loss": 0.2741,
+      "step": 136440
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1196471022915665e-07,
+      "loss": 0.2443,
+      "step": 136445
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1159469859049764e-07,
+      "loss": 0.2567,
+      "step": 136450
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1122500881294582e-07,
+      "loss": 0.262,
+      "step": 136455
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1085564090130294e-07,
+      "loss": 0.2325,
+      "step": 136460
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.104865948603596e-07,
+      "loss": 0.2378,
+      "step": 136465
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.1011787069491195e-07,
+      "loss": 0.2405,
+      "step": 136470
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0974946840974786e-07,
+      "loss": 0.2429,
+      "step": 136475
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.093813880096468e-07,
+      "loss": 0.2424,
+      "step": 136480
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0901362949939384e-07,
+      "loss": 0.2447,
+      "step": 136485
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.086461928837602e-07,
+      "loss": 0.2492,
+      "step": 136490
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0827907816751702e-07,
+      "loss": 0.2411,
+      "step": 136495
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0791228535543272e-07,
+      "loss": 0.2492,
+      "step": 136500
+    },
+    {
+      "epoch": 4.8,
+      "eval_loss": 0.2484273761510849,
+      "eval_runtime": 10.5494,
+      "eval_samples_per_second": 9.479,
+      "eval_steps_per_second": 9.479,
+      "step": 136500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.075458144522674e-07,
+      "loss": 0.2473,
+      "step": 136505
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.071796654627839e-07,
+      "loss": 0.2616,
+      "step": 136510
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0681383839172842e-07,
+      "loss": 0.2791,
+      "step": 136515
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0644833324386104e-07,
+      "loss": 0.2526,
+      "step": 136520
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0608315002391687e-07,
+      "loss": 0.2583,
+      "step": 136525
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0571828873664767e-07,
+      "loss": 0.2542,
+      "step": 136530
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0535374938678575e-07,
+      "loss": 0.2544,
+      "step": 136535
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0498953197906344e-07,
+      "loss": 0.2393,
+      "step": 136540
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.046256365182131e-07,
+      "loss": 0.246,
+      "step": 136545
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.042620630089559e-07,
+      "loss": 0.2716,
+      "step": 136550
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0389881145601863e-07,
+      "loss": 0.2491,
+      "step": 136555
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0353588186411143e-07,
+      "loss": 0.2638,
+      "step": 136560
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0317327423794995e-07,
+      "loss": 0.2367,
+      "step": 136565
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.0281098858224157e-07,
+      "loss": 0.2575,
+      "step": 136570
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0244902490169081e-07,
+      "loss": 0.2421,
+      "step": 136575
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0208738320099675e-07,
+      "loss": 0.2673,
+      "step": 136580
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0172606348485556e-07,
+      "loss": 0.2437,
+      "step": 136585
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0136506575795798e-07,
+      "loss": 0.2469,
+      "step": 136590
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0100439002499473e-07,
+      "loss": 0.258,
+      "step": 136595
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.0064403629064533e-07,
+      "loss": 0.2409,
+      "step": 136600
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.002840045595894e-07,
+      "loss": 0.2655,
+      "step": 136605
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9992429483650098e-07,
+      "loss": 0.2792,
+      "step": 136610
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9956490712605135e-07,
+      "loss": 0.2716,
+      "step": 136615
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9920584143290898e-07,
+      "loss": 0.2667,
+      "step": 136620
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9884709776173404e-07,
+      "loss": 0.2351,
+      "step": 136625
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9848867611718393e-07,
+      "loss": 0.2529,
+      "step": 136630
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.981305765039132e-07,
+      "loss": 0.2524,
+      "step": 136635
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9777279892657375e-07,
+      "loss": 0.2502,
+      "step": 136640
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9741534338980906e-07,
+      "loss": 0.2542,
+      "step": 136645
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9705820989825984e-07,
+      "loss": 0.2528,
+      "step": 136650
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.967013984565641e-07,
+      "loss": 0.2565,
+      "step": 136655
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9634490906935698e-07,
+      "loss": 0.2407,
+      "step": 136660
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9598874174126258e-07,
+      "loss": 0.2541,
+      "step": 136665
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9563289647690775e-07,
+      "loss": 0.2327,
+      "step": 136670
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9527737328091378e-07,
+      "loss": 0.2477,
+      "step": 136675
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9492217215789644e-07,
+      "loss": 0.2629,
+      "step": 136680
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.945672931124659e-07,
+      "loss": 0.2566,
+      "step": 136685
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9421273614923518e-07,
+      "loss": 0.261,
+      "step": 136690
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.938585012728006e-07,
+      "loss": 0.2365,
+      "step": 136695
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9350458848776675e-07,
+      "loss": 0.2402,
+      "step": 136700
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9315099779872726e-07,
+      "loss": 0.2434,
+      "step": 136705
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.927977292102756e-07,
+      "loss": 0.2461,
+      "step": 136710
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9244478272699705e-07,
+      "loss": 0.2583,
+      "step": 136715
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9209215835347404e-07,
+      "loss": 0.2627,
+      "step": 136720
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9173985609428348e-07,
+      "loss": 0.2547,
+      "step": 136725
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.913878759540022e-07,
+      "loss": 0.253,
+      "step": 136730
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9103621793719885e-07,
+      "loss": 0.246,
+      "step": 136735
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.9068488204844193e-07,
+      "loss": 0.256,
+      "step": 136740
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.903338682922917e-07,
+      "loss": 0.2588,
+      "step": 136745
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.899831766733057e-07,
+      "loss": 0.2472,
+      "step": 136750
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.896328071960385e-07,
+      "loss": 0.265,
+      "step": 136755
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.892827598650393e-07,
+      "loss": 0.2555,
+      "step": 136760
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.889330346848489e-07,
+      "loss": 0.2616,
+      "step": 136765
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8858363166001646e-07,
+      "loss": 0.2351,
+      "step": 136770
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8823455079507168e-07,
+      "loss": 0.2585,
+      "step": 136775
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8788579209454982e-07,
+      "loss": 0.2486,
+      "step": 136780
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8753735556298057e-07,
+      "loss": 0.2458,
+      "step": 136785
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8718924120488536e-07,
+      "loss": 0.2521,
+      "step": 136790
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8684144902478552e-07,
+      "loss": 0.2684,
+      "step": 136795
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.864939790271969e-07,
+      "loss": 0.2521,
+      "step": 136800
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8614683121663257e-07,
+      "loss": 0.2539,
+      "step": 136805
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8580000559759726e-07,
+      "loss": 0.2546,
+      "step": 136810
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8545350217459846e-07,
+      "loss": 0.2537,
+      "step": 136815
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.851073209521298e-07,
+      "loss": 0.2552,
+      "step": 136820
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8476146193468768e-07,
+      "loss": 0.242,
+      "step": 136825
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8441592512676853e-07,
+      "loss": 0.2568,
+      "step": 136830
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8407071053285208e-07,
+      "loss": 0.2641,
+      "step": 136835
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8372581815742085e-07,
+      "loss": 0.2589,
+      "step": 136840
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8338124800496014e-07,
+      "loss": 0.2387,
+      "step": 136845
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8303700007993585e-07,
+      "loss": 0.2526,
+      "step": 136850
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.8269307438682214e-07,
+      "loss": 0.2391,
+      "step": 136855
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8234947093008214e-07,
+      "loss": 0.2204,
+      "step": 136860
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8200618971417892e-07,
+      "loss": 0.2157,
+      "step": 136865
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8166323074357284e-07,
+      "loss": 0.2753,
+      "step": 136870
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.813205940227103e-07,
+      "loss": 0.2546,
+      "step": 136875
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8097827955604608e-07,
+      "loss": 0.2641,
+      "step": 136880
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8063628734802107e-07,
+      "loss": 0.2514,
+      "step": 136885
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8036292561080804e-07,
+      "loss": 0.222,
+      "step": 136890
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.8002151347952245e-07,
+      "loss": 0.2635,
+      "step": 136895
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7968042361930227e-07,
+      "loss": 0.2581,
+      "step": 136900
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7933965603457447e-07,
+      "loss": 0.2634,
+      "step": 136905
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7899921072976332e-07,
+      "loss": 0.2852,
+      "step": 136910
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7865908770929306e-07,
+      "loss": 0.2385,
+      "step": 136915
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7831928697757404e-07,
+      "loss": 0.2633,
+      "step": 136920
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7797980853901942e-07,
+      "loss": 0.2316,
+      "step": 136925
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.776406523980395e-07,
+      "loss": 0.2528,
+      "step": 136930
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7730181855903916e-07,
+      "loss": 0.264,
+      "step": 136935
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7696330702641482e-07,
+      "loss": 0.2626,
+      "step": 136940
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7662511780456304e-07,
+      "loss": 0.2429,
+      "step": 136945
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.762872508978719e-07,
+      "loss": 0.256,
+      "step": 136950
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7594970631073515e-07,
+      "loss": 0.2664,
+      "step": 136955
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.756124840475326e-07,
+      "loss": 0.238,
+      "step": 136960
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7527558411263856e-07,
+      "loss": 0.232,
+      "step": 136965
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7493900651043282e-07,
+      "loss": 0.2588,
+      "step": 136970
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7460275124528413e-07,
+      "loss": 0.2515,
+      "step": 136975
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7426681832155844e-07,
+      "loss": 0.2641,
+      "step": 136980
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7393120774361338e-07,
+      "loss": 0.2433,
+      "step": 136985
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7359591951581489e-07,
+      "loss": 0.259,
+      "step": 136990
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7326095364251226e-07,
+      "loss": 0.2382,
+      "step": 136995
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7292631012805204e-07,
+      "loss": 0.2548,
+      "step": 137000
+    },
+    {
+      "epoch": 4.82,
+      "eval_loss": 0.24844510853290558,
+      "eval_runtime": 10.5423,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 137000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7259198897678352e-07,
+      "loss": 0.2178,
+      "step": 137005
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.722579901930449e-07,
+      "loss": 0.2575,
+      "step": 137010
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7192431378117714e-07,
+      "loss": 0.2518,
+      "step": 137015
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7159095974550731e-07,
+      "loss": 0.2554,
+      "step": 137020
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7125792809036533e-07,
+      "loss": 0.2502,
+      "step": 137025
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7092521882007827e-07,
+      "loss": 0.2334,
+      "step": 137030
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7059283193896492e-07,
+      "loss": 0.2553,
+      "step": 137035
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7026076745134123e-07,
+      "loss": 0.2533,
+      "step": 137040
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.699290253615149e-07,
+      "loss": 0.2623,
+      "step": 137045
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6959760567379912e-07,
+      "loss": 0.2666,
+      "step": 137050
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6926650839249325e-07,
+      "loss": 0.245,
+      "step": 137055
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6893573352189663e-07,
+      "loss": 0.2505,
+      "step": 137060
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.686052810663058e-07,
+      "loss": 0.2386,
+      "step": 137065
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.682751510300118e-07,
+      "loss": 0.2312,
+      "step": 137070
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6794534341730006e-07,
+      "loss": 0.2437,
+      "step": 137075
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.676158582324533e-07,
+      "loss": 0.2555,
+      "step": 137080
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.672866954797514e-07,
+      "loss": 0.2692,
+      "step": 137085
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6695785516346318e-07,
+      "loss": 0.2638,
+      "step": 137090
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6662933728786302e-07,
+      "loss": 0.2389,
+      "step": 137095
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6630114185721413e-07,
+      "loss": 0.2585,
+      "step": 137100
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.659732688757798e-07,
+      "loss": 0.2204,
+      "step": 137105
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6564571834781495e-07,
+      "loss": 0.2584,
+      "step": 137110
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6531849027757728e-07,
+      "loss": 0.2436,
+      "step": 137115
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6499158466930786e-07,
+      "loss": 0.2404,
+      "step": 137120
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6466500152725882e-07,
+      "loss": 0.225,
+      "step": 137125
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6433874085566848e-07,
+      "loss": 0.2519,
+      "step": 137130
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.640128026587695e-07,
+      "loss": 0.2528,
+      "step": 137135
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.6368718694079744e-07,
+      "loss": 0.2454,
+      "step": 137140
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.633618937059822e-07,
+      "loss": 0.2469,
+      "step": 137145
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6303692295854268e-07,
+      "loss": 0.2677,
+      "step": 137150
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6271227470270046e-07,
+      "loss": 0.2444,
+      "step": 137155
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6238794894267163e-07,
+      "loss": 0.2543,
+      "step": 137160
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6206394568266393e-07,
+      "loss": 0.2486,
+      "step": 137165
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6174026492689065e-07,
+      "loss": 0.2558,
+      "step": 137170
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6141690667954846e-07,
+      "loss": 0.2419,
+      "step": 137175
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6109387094483953e-07,
+      "loss": 0.259,
+      "step": 137180
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6077115772695773e-07,
+      "loss": 0.247,
+      "step": 137185
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6044876703009414e-07,
+      "loss": 0.2441,
+      "step": 137190
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.6012669885843158e-07,
+      "loss": 0.2417,
+      "step": 137195
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5980495321615273e-07,
+      "loss": 0.2485,
+      "step": 137200
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5948353010743765e-07,
+      "loss": 0.2444,
+      "step": 137205
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.59162429536458e-07,
+      "loss": 0.2523,
+      "step": 137210
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5884165150738538e-07,
+      "loss": 0.2692,
+      "step": 137215
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5852119602438043e-07,
+      "loss": 0.2524,
+      "step": 137220
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5820106309160642e-07,
+      "loss": 0.2647,
+      "step": 137225
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.578812527132212e-07,
+      "loss": 0.247,
+      "step": 137230
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5756176489337693e-07,
+      "loss": 0.2237,
+      "step": 137235
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5724259963622035e-07,
+      "loss": 0.2317,
+      "step": 137240
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5692375694589813e-07,
+      "loss": 0.2366,
+      "step": 137245
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.566052368265486e-07,
+      "loss": 0.261,
+      "step": 137250
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.562870392823046e-07,
+      "loss": 0.2254,
+      "step": 137255
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5596916431730447e-07,
+      "loss": 0.2487,
+      "step": 137260
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.556516119356699e-07,
+      "loss": 0.2459,
+      "step": 137265
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.553343821415254e-07,
+      "loss": 0.2599,
+      "step": 137270
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5501747493899266e-07,
+      "loss": 0.2809,
+      "step": 137275
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5470089033218504e-07,
+      "loss": 0.2483,
+      "step": 137280
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.543846283252104e-07,
+      "loss": 0.2593,
+      "step": 137285
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5406868892217928e-07,
+      "loss": 0.2499,
+      "step": 137290
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5375307212719125e-07,
+      "loss": 0.2366,
+      "step": 137295
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5343777794434577e-07,
+      "loss": 0.2279,
+      "step": 137300
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.531228063777368e-07,
+      "loss": 0.2305,
+      "step": 137305
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5280815743144993e-07,
+      "loss": 0.2283,
+      "step": 137310
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5249383110957915e-07,
+      "loss": 0.2718,
+      "step": 137315
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5217982741619897e-07,
+      "loss": 0.2493,
+      "step": 137320
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5186614635538664e-07,
+      "loss": 0.2549,
+      "step": 137325
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5155278793121953e-07,
+      "loss": 0.2453,
+      "step": 137330
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5123975214776098e-07,
+      "loss": 0.2457,
+      "step": 137335
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5092703900908e-07,
+      "loss": 0.2334,
+      "step": 137340
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5061464851923445e-07,
+      "loss": 0.2459,
+      "step": 137345
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.503025806822822e-07,
+      "loss": 0.251,
+      "step": 137350
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4999083550227277e-07,
+      "loss": 0.2453,
+      "step": 137355
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.496794129832585e-07,
+      "loss": 0.2514,
+      "step": 137360
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4936831312927778e-07,
+      "loss": 0.2441,
+      "step": 137365
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4905753594437465e-07,
+      "loss": 0.2444,
+      "step": 137370
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.487470814325792e-07,
+      "loss": 0.2529,
+      "step": 137375
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.484369495979271e-07,
+      "loss": 0.2412,
+      "step": 137380
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4812714044444286e-07,
+      "loss": 0.2791,
+      "step": 137385
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4781765397614834e-07,
+      "loss": 0.2492,
+      "step": 137390
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4750849019706526e-07,
+      "loss": 0.2469,
+      "step": 137395
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4719964911120433e-07,
+      "loss": 0.2257,
+      "step": 137400
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.46891130722579e-07,
+      "loss": 0.2551,
+      "step": 137405
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.465829350351944e-07,
+      "loss": 0.2612,
+      "step": 137410
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.462750620530501e-07,
+      "loss": 0.253,
+      "step": 137415
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4596751178014567e-07,
+      "loss": 0.2282,
+      "step": 137420
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.4566028422046962e-07,
+      "loss": 0.265,
+      "step": 137425
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4535337937801875e-07,
+      "loss": 0.2416,
+      "step": 137430
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4504679725677594e-07,
+      "loss": 0.2468,
+      "step": 137435
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4474053786071862e-07,
+      "loss": 0.2627,
+      "step": 137440
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4443460119382413e-07,
+      "loss": 0.2615,
+      "step": 137445
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4412898726006708e-07,
+      "loss": 0.2619,
+      "step": 137450
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4382369606341372e-07,
+      "loss": 0.2418,
+      "step": 137455
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4351872760783036e-07,
+      "loss": 0.2381,
+      "step": 137460
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4321408189727214e-07,
+      "loss": 0.2456,
+      "step": 137465
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4290975893570258e-07,
+      "loss": 0.2547,
+      "step": 137470
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.426057587270657e-07,
+      "loss": 0.2227,
+      "step": 137475
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.423020812753112e-07,
+      "loss": 0.2465,
+      "step": 137480
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4199872658438308e-07,
+      "loss": 0.2424,
+      "step": 137485
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4169569465821708e-07,
+      "loss": 0.2548,
+      "step": 137490
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4139298550075174e-07,
+      "loss": 0.2368,
+      "step": 137495
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4109059911591726e-07,
+      "loss": 0.2584,
+      "step": 137500
+    },
+    {
+      "epoch": 4.84,
+      "eval_loss": 0.24842242896556854,
+      "eval_runtime": 10.5421,
+      "eval_samples_per_second": 9.486,
+      "eval_steps_per_second": 9.486,
+      "step": 137500
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4078853550763827e-07,
+      "loss": 0.2424,
+      "step": 137505
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.404867946798366e-07,
+      "loss": 0.2437,
+      "step": 137510
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.4018537663643138e-07,
+      "loss": 0.2562,
+      "step": 137515
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3988428138133614e-07,
+      "loss": 0.2518,
+      "step": 137520
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3958350891845884e-07,
+      "loss": 0.2405,
+      "step": 137525
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3928305925170748e-07,
+      "loss": 0.265,
+      "step": 137530
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3898293238498173e-07,
+      "loss": 0.2407,
+      "step": 137535
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3868312832217566e-07,
+      "loss": 0.243,
+      "step": 137540
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3838364706718898e-07,
+      "loss": 0.2689,
+      "step": 137545
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3808448862390467e-07,
+      "loss": 0.2632,
+      "step": 137550
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.377856529962085e-07,
+      "loss": 0.2625,
+      "step": 137555
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3748714018797792e-07,
+      "loss": 0.2655,
+      "step": 137560
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3718895020309598e-07,
+      "loss": 0.2619,
+      "step": 137565
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3689108304542896e-07,
+      "loss": 0.2728,
+      "step": 137570
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3659353871884606e-07,
+      "loss": 0.2463,
+      "step": 137575
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3629631722721082e-07,
+      "loss": 0.2852,
+      "step": 137580
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3599941857438125e-07,
+      "loss": 0.2604,
+      "step": 137585
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.357028427642154e-07,
+      "loss": 0.2438,
+      "step": 137590
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.354065898005602e-07,
+      "loss": 0.2587,
+      "step": 137595
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3511065968726255e-07,
+      "loss": 0.2569,
+      "step": 137600
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3481505242816661e-07,
+      "loss": 0.2462,
+      "step": 137605
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3451976802711376e-07,
+      "loss": 0.2453,
+      "step": 137610
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3422480648793146e-07,
+      "loss": 0.2474,
+      "step": 137615
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3393016781445278e-07,
+      "loss": 0.26,
+      "step": 137620
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.336358520105052e-07,
+      "loss": 0.2556,
+      "step": 137625
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3334185907990792e-07,
+      "loss": 0.2798,
+      "step": 137630
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.330481890264801e-07,
+      "loss": 0.2664,
+      "step": 137635
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3275484185403254e-07,
+      "loss": 0.2458,
+      "step": 137640
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.324618175663761e-07,
+      "loss": 0.2679,
+      "step": 137645
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.321691161673133e-07,
+      "loss": 0.2491,
+      "step": 137650
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3187673766064667e-07,
+      "loss": 0.2237,
+      "step": 137655
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3158468205017038e-07,
+      "loss": 0.2304,
+      "step": 137660
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3129294933967862e-07,
+      "loss": 0.2609,
+      "step": 137665
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3100153953296e-07,
+      "loss": 0.2479,
+      "step": 137670
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3071045263379765e-07,
+      "loss": 0.2493,
+      "step": 137675
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.304196886459691e-07,
+      "loss": 0.2465,
+      "step": 137680
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.3012924757325184e-07,
+      "loss": 0.2568,
+      "step": 137685
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2983912941941789e-07,
+      "loss": 0.2331,
+      "step": 137690
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.295493341882309e-07,
+      "loss": 0.2528,
+      "step": 137695
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.292598618834573e-07,
+      "loss": 0.2509,
+      "step": 137700
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.289707125088524e-07,
+      "loss": 0.2602,
+      "step": 137705
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2868188606817156e-07,
+      "loss": 0.2426,
+      "step": 137710
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2839338256516454e-07,
+      "loss": 0.2452,
+      "step": 137715
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.281052020035811e-07,
+      "loss": 0.2606,
+      "step": 137720
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2781734438715998e-07,
+      "loss": 0.2324,
+      "step": 137725
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.275298097196398e-07,
+      "loss": 0.2512,
+      "step": 137730
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2724259800475092e-07,
+      "loss": 0.2296,
+      "step": 137735
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2695570924622924e-07,
+      "loss": 0.247,
+      "step": 137740
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2666914344779402e-07,
+      "loss": 0.2445,
+      "step": 137745
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2638290061316727e-07,
+      "loss": 0.2427,
+      "step": 137750
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2609698074606546e-07,
+      "loss": 0.2335,
+      "step": 137755
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.258113838502023e-07,
+      "loss": 0.2391,
+      "step": 137760
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2552610992928594e-07,
+      "loss": 0.2573,
+      "step": 137765
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2524115898701893e-07,
+      "loss": 0.2355,
+      "step": 137770
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.249565310271039e-07,
+      "loss": 0.2606,
+      "step": 137775
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2467222605323514e-07,
+      "loss": 0.2592,
+      "step": 137780
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.243882440691041e-07,
+      "loss": 0.2403,
+      "step": 137785
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2410458507839673e-07,
+      "loss": 0.2583,
+      "step": 137790
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.238212490847962e-07,
+      "loss": 0.2581,
+      "step": 137795
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2353823609198567e-07,
+      "loss": 0.2451,
+      "step": 137800
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.232555461036372e-07,
+      "loss": 0.2673,
+      "step": 137805
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.229731791234201e-07,
+      "loss": 0.2485,
+      "step": 137810
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2269113515500085e-07,
+      "loss": 0.2637,
+      "step": 137815
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.224094142020432e-07,
+      "loss": 0.2276,
+      "step": 137820
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2212801626820536e-07,
+      "loss": 0.254,
+      "step": 137825
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2184694135713714e-07,
+      "loss": 0.2214,
+      "step": 137830
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2156618947249397e-07,
+      "loss": 0.2268,
+      "step": 137835
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.212857606179174e-07,
+      "loss": 0.2677,
+      "step": 137840
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2100565479704896e-07,
+      "loss": 0.2739,
+      "step": 137845
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2072587201352736e-07,
+      "loss": 0.2272,
+      "step": 137850
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2044641227098308e-07,
+      "loss": 0.2485,
+      "step": 137855
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2016727557304652e-07,
+      "loss": 0.2608,
+      "step": 137860
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1988846192334257e-07,
+      "loss": 0.2513,
+      "step": 137865
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1960997132548779e-07,
+      "loss": 0.2354,
+      "step": 137870
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1933180378310427e-07,
+      "loss": 0.2553,
+      "step": 137875
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1905395929979746e-07,
+      "loss": 0.2457,
+      "step": 137880
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1877643787917837e-07,
+      "loss": 0.2558,
+      "step": 137885
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1849923952484965e-07,
+      "loss": 0.2364,
+      "step": 137890
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1822236424041123e-07,
+      "loss": 0.2577,
+      "step": 137895
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1794581202945742e-07,
+      "loss": 0.2297,
+      "step": 137900
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1766958289557983e-07,
+      "loss": 0.2403,
+      "step": 137905
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1739367684236446e-07,
+      "loss": 0.2338,
+      "step": 137910
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1711809387339457e-07,
+      "loss": 0.2553,
+      "step": 137915
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1684283399224505e-07,
+      "loss": 0.2175,
+      "step": 137920
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.165678972024964e-07,
+      "loss": 0.2418,
+      "step": 137925
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.162932835077124e-07,
+      "loss": 0.2471,
+      "step": 137930
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1601899291146245e-07,
+      "loss": 0.2449,
+      "step": 137935
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.157450254173048e-07,
+      "loss": 0.2599,
+      "step": 137940
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.154713810288005e-07,
+      "loss": 0.2583,
+      "step": 137945
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1519805974949948e-07,
+      "loss": 0.2593,
+      "step": 137950
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1492506158295447e-07,
+      "loss": 0.2464,
+      "step": 137955
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1465238653270427e-07,
+      "loss": 0.2589,
+      "step": 137960
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1438003460229607e-07,
+      "loss": 0.2413,
+      "step": 137965
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1410800579526037e-07,
+      "loss": 0.2471,
+      "step": 137970
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1383630011513325e-07,
+      "loss": 0.2523,
+      "step": 137975
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1356491756543964e-07,
+      "loss": 0.2421,
+      "step": 137980
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1329385814970727e-07,
+      "loss": 0.2348,
+      "step": 137985
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1302312187145003e-07,
+      "loss": 0.2566,
+      "step": 137990
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1275270873419008e-07,
+      "loss": 0.2439,
+      "step": 137995
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1248261874143296e-07,
+      "loss": 0.2405,
+      "step": 138000
+    },
+    {
+      "epoch": 4.86,
+      "eval_loss": 0.24844183027744293,
+      "eval_runtime": 10.5317,
+      "eval_samples_per_second": 9.495,
+      "eval_steps_per_second": 9.495,
+      "step": 138000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1221285189668696e-07,
+      "loss": 0.2696,
+      "step": 138005
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1194340820345484e-07,
+      "loss": 0.2456,
+      "step": 138010
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1167428766523658e-07,
+      "loss": 0.2656,
+      "step": 138015
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.114054902855266e-07,
+      "loss": 0.216,
+      "step": 138020
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1113701606781379e-07,
+      "loss": 0.2607,
+      "step": 138025
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1086886501558424e-07,
+      "loss": 0.2399,
+      "step": 138030
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1060103713232128e-07,
+      "loss": 0.252,
+      "step": 138035
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1033353242149713e-07,
+      "loss": 0.2688,
+      "step": 138040
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1006635088659234e-07,
+      "loss": 0.2486,
+      "step": 138045
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.097994925310708e-07,
+      "loss": 0.2501,
+      "step": 138050
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0953295735840196e-07,
+      "loss": 0.2465,
+      "step": 138055
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0926674537204417e-07,
+      "loss": 0.2444,
+      "step": 138060
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.09000856575453e-07,
+      "loss": 0.2591,
+      "step": 138065
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0873529097208124e-07,
+      "loss": 0.2573,
+      "step": 138070
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.084700485653789e-07,
+      "loss": 0.2582,
+      "step": 138075
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0820512935879046e-07,
+      "loss": 0.2638,
+      "step": 138080
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0794053335575205e-07,
+      "loss": 0.2711,
+      "step": 138085
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0767626055969982e-07,
+      "loss": 0.2571,
+      "step": 138090
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.074123109740699e-07,
+      "loss": 0.2652,
+      "step": 138095
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0714868460228456e-07,
+      "loss": 0.2603,
+      "step": 138100
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0688538144776883e-07,
+      "loss": 0.2508,
+      "step": 138105
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0662240151393942e-07,
+      "loss": 0.2563,
+      "step": 138110
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0635974480421585e-07,
+      "loss": 0.2324,
+      "step": 138115
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0609741132200369e-07,
+      "loss": 0.2727,
+      "step": 138120
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0583540107071132e-07,
+      "loss": 0.2503,
+      "step": 138125
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0557371405374162e-07,
+      "loss": 0.2551,
+      "step": 138130
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0531235027448904e-07,
+      "loss": 0.2414,
+      "step": 138135
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0505130973635092e-07,
+      "loss": 0.2511,
+      "step": 138140
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0479059244271616e-07,
+      "loss": 0.2478,
+      "step": 138145
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0453019839696543e-07,
+      "loss": 0.24,
+      "step": 138150
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0427012760248489e-07,
+      "loss": 0.2648,
+      "step": 138155
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.040103800626524e-07,
+      "loss": 0.2579,
+      "step": 138160
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0375095578083471e-07,
+      "loss": 0.242,
+      "step": 138165
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.034918547604069e-07,
+      "loss": 0.2422,
+      "step": 138170
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.032330770047274e-07,
+      "loss": 0.2489,
+      "step": 138175
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.029746225171574e-07,
+      "loss": 0.242,
+      "step": 138180
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0271649130105809e-07,
+      "loss": 0.2356,
+      "step": 138185
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0245868335977405e-07,
+      "loss": 0.2374,
+      "step": 138190
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0220119869665812e-07,
+      "loss": 0.2428,
+      "step": 138195
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0194403731504932e-07,
+      "loss": 0.2554,
+      "step": 138200
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.016871992182894e-07,
+      "loss": 0.2524,
+      "step": 138205
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.014306844097146e-07,
+      "loss": 0.2522,
+      "step": 138210
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0117449289265001e-07,
+      "loss": 0.2532,
+      "step": 138215
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0091862467042912e-07,
+      "loss": 0.2641,
+      "step": 138220
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0066307974636868e-07,
+      "loss": 0.2501,
+      "step": 138225
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0040785812379106e-07,
+      "loss": 0.2482,
+      "step": 138230
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.001529598060047e-07,
+      "loss": 0.2367,
+      "step": 138235
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.989838479632641e-08,
+      "loss": 0.2477,
+      "step": 138240
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.964413309805353e-08,
+      "loss": 0.2497,
+      "step": 138245
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.939020471449456e-08,
+      "loss": 0.2469,
+      "step": 138250
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.913659964894406e-08,
+      "loss": 0.2403,
+      "step": 138255
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.888331790469107e-08,
+      "loss": 0.2633,
+      "step": 138260
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.863035948503019e-08,
+      "loss": 0.255,
+      "step": 138265
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.837772439324488e-08,
+      "loss": 0.2596,
+      "step": 138270
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.812541263261033e-08,
+      "loss": 0.2755,
+      "step": 138275
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.787342420640999e-08,
+      "loss": 0.2565,
+      "step": 138280
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.762175911791072e-08,
+      "loss": 0.2332,
+      "step": 138285
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.73704173703821e-08,
+      "loss": 0.2405,
+      "step": 138290
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.711939896708544e-08,
+      "loss": 0.2411,
+      "step": 138295
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.686870391128478e-08,
+      "loss": 0.233,
+      "step": 138300
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.661833220623029e-08,
+      "loss": 0.2342,
+      "step": 138305
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.636828385517493e-08,
+      "loss": 0.2569,
+      "step": 138310
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.611855886136334e-08,
+      "loss": 0.2464,
+      "step": 138315
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.58691572280429e-08,
+      "loss": 0.2324,
+      "step": 138320
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.562007895844716e-08,
+      "loss": 0.261,
+      "step": 138325
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.537132405581239e-08,
+      "loss": 0.243,
+      "step": 138330
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.512289252336937e-08,
+      "loss": 0.2475,
+      "step": 138335
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.487478436434327e-08,
+      "loss": 0.2221,
+      "step": 138340
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.462699958195376e-08,
+      "loss": 0.2452,
+      "step": 138345
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.437953817941769e-08,
+      "loss": 0.2392,
+      "step": 138350
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.413240015995195e-08,
+      "loss": 0.2473,
+      "step": 138355
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.388558552676508e-08,
+      "loss": 0.2585,
+      "step": 138360
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.363909428305728e-08,
+      "loss": 0.2457,
+      "step": 138365
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.339292643203157e-08,
+      "loss": 0.2499,
+      "step": 138370
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.314708197688537e-08,
+      "loss": 0.2475,
+      "step": 138375
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.290156092081059e-08,
+      "loss": 0.2387,
+      "step": 138380
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.265636326699634e-08,
+      "loss": 0.2408,
+      "step": 138385
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.24114890186234e-08,
+      "loss": 0.2603,
+      "step": 138390
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.21669381788698e-08,
+      "loss": 0.2541,
+      "step": 138395
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.19227107509163e-08,
+      "loss": 0.2484,
+      "step": 138400
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.167880673793261e-08,
+      "loss": 0.2568,
+      "step": 138405
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.143522614308287e-08,
+      "loss": 0.251,
+      "step": 138410
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.11919689695312e-08,
+      "loss": 0.2694,
+      "step": 138415
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.094903522043618e-08,
+      "loss": 0.234,
+      "step": 138420
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.070642489895087e-08,
+      "loss": 0.2431,
+      "step": 138425
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.046413800822828e-08,
+      "loss": 0.257,
+      "step": 138430
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.022217455141314e-08,
+      "loss": 0.2599,
+      "step": 138435
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.998053453164734e-08,
+      "loss": 0.2709,
+      "step": 138440
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.97392179520673e-08,
+      "loss": 0.2447,
+      "step": 138445
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.94982248158066e-08,
+      "loss": 0.2296,
+      "step": 138450
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.925755512599609e-08,
+      "loss": 0.2401,
+      "step": 138455
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.901720888575826e-08,
+      "loss": 0.2526,
+      "step": 138460
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.877718609821562e-08,
+      "loss": 0.2359,
+      "step": 138465
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.85374867664851e-08,
+      "loss": 0.2719,
+      "step": 138470
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.829811089367811e-08,
+      "loss": 0.2384,
+      "step": 138475
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.805905848290053e-08,
+      "loss": 0.2477,
+      "step": 138480
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.782032953725817e-08,
+      "loss": 0.2318,
+      "step": 138485
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.758192405985133e-08,
+      "loss": 0.2362,
+      "step": 138490
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.734384205377755e-08,
+      "loss": 0.2613,
+      "step": 138495
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.710608352212324e-08,
+      "loss": 0.2318,
+      "step": 138500
+    },
+    {
+      "epoch": 4.87,
+      "eval_loss": 0.24843639135360718,
+      "eval_runtime": 10.5531,
+      "eval_samples_per_second": 9.476,
+      "eval_steps_per_second": 9.476,
+      "step": 138500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.686864846798037e-08,
+      "loss": 0.2708,
+      "step": 138505
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.663153689442982e-08,
+      "loss": 0.2577,
+      "step": 138510
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.639474880454968e-08,
+      "loss": 0.2459,
+      "step": 138515
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.615828420141526e-08,
+      "loss": 0.2296,
+      "step": 138520
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.592214308809632e-08,
+      "loss": 0.2533,
+      "step": 138525
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.568632546765987e-08,
+      "loss": 0.2483,
+      "step": 138530
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.545083134316456e-08,
+      "loss": 0.2506,
+      "step": 138535
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.521566071767462e-08,
+      "loss": 0.2508,
+      "step": 138540
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.49808135942376e-08,
+      "loss": 0.2545,
+      "step": 138545
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.474628997590661e-08,
+      "loss": 0.2407,
+      "step": 138550
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.451208986572367e-08,
+      "loss": 0.2555,
+      "step": 138555
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 8.42782132667308e-08,
+      "loss": 0.258,
+      "step": 138560
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.404466018196722e-08,
+      "loss": 0.2316,
+      "step": 138565
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.381143061446383e-08,
+      "loss": 0.2609,
+      "step": 138570
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.35785245672488e-08,
+      "loss": 0.2565,
+      "step": 138575
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.334594204334467e-08,
+      "loss": 0.2244,
+      "step": 138580
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.311368304577128e-08,
+      "loss": 0.2621,
+      "step": 138585
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.28817475775484e-08,
+      "loss": 0.2418,
+      "step": 138590
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.265013564168478e-08,
+      "loss": 0.2513,
+      "step": 138595
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.241884724118631e-08,
+      "loss": 0.2521,
+      "step": 138600
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.218788237905895e-08,
+      "loss": 0.2319,
+      "step": 138605
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.195724105830305e-08,
+      "loss": 0.2504,
+      "step": 138610
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.172692328190512e-08,
+      "loss": 0.2492,
+      "step": 138615
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.149692905286555e-08,
+      "loss": 0.2471,
+      "step": 138620
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.126725837416527e-08,
+      "loss": 0.2494,
+      "step": 138625
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.103791124878801e-08,
+      "loss": 0.2642,
+      "step": 138630
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.080888767971196e-08,
+      "loss": 0.2405,
+      "step": 138635
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.058018766990972e-08,
+      "loss": 0.2446,
+      "step": 138640
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.035181122234837e-08,
+      "loss": 0.244,
+      "step": 138645
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 8.012375834000053e-08,
+      "loss": 0.261,
+      "step": 138650
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.989602902581938e-08,
+      "loss": 0.2475,
+      "step": 138655
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.966862328276647e-08,
+      "loss": 0.2544,
+      "step": 138660
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.944154111379498e-08,
+      "loss": 0.2701,
+      "step": 138665
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.921478252184977e-08,
+      "loss": 0.2404,
+      "step": 138670
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.898834750987849e-08,
+      "loss": 0.2496,
+      "step": 138675
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.876223608081767e-08,
+      "loss": 0.2536,
+      "step": 138680
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.853644823760941e-08,
+      "loss": 0.2337,
+      "step": 138685
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.831098398317916e-08,
+      "loss": 0.2566,
+      "step": 138690
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.80858433204551e-08,
+      "loss": 0.2438,
+      "step": 138695
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.786102625236269e-08,
+      "loss": 0.2492,
+      "step": 138700
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.763653278182182e-08,
+      "loss": 0.2476,
+      "step": 138705
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.74123629117468e-08,
+      "loss": 0.2509,
+      "step": 138710
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.718851664504645e-08,
+      "loss": 0.2566,
+      "step": 138715
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.696499398462953e-08,
+      "loss": 0.2335,
+      "step": 138720
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.67417949333965e-08,
+      "loss": 0.2551,
+      "step": 138725
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.651891949424784e-08,
+      "loss": 0.2579,
+      "step": 138730
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.629636767007287e-08,
+      "loss": 0.2632,
+      "step": 138735
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.607413946376651e-08,
+      "loss": 0.2468,
+      "step": 138740
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.585223487821259e-08,
+      "loss": 0.2377,
+      "step": 138745
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.563065391628932e-08,
+      "loss": 0.2611,
+      "step": 138750
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.540939658087776e-08,
+      "loss": 0.2642,
+      "step": 138755
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.518846287484782e-08,
+      "loss": 0.2432,
+      "step": 138760
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.496785280107222e-08,
+      "loss": 0.2575,
+      "step": 138765
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.474756636240976e-08,
+      "loss": 0.2462,
+      "step": 138770
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.452760356172483e-08,
+      "loss": 0.2464,
+      "step": 138775
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.430796440187349e-08,
+      "loss": 0.2735,
+      "step": 138780
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.408864888570621e-08,
+      "loss": 0.2807,
+      "step": 138785
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.386965701607074e-08,
+      "loss": 0.2442,
+      "step": 138790
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.3650988795812e-08,
+      "loss": 0.2407,
+      "step": 138795
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.343264422776663e-08,
+      "loss": 0.2234,
+      "step": 138800
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.321462331477124e-08,
+      "loss": 0.2519,
+      "step": 138805
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.299692605965691e-08,
+      "loss": 0.2525,
+      "step": 138810
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.277955246524915e-08,
+      "loss": 0.2468,
+      "step": 138815
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.256250253437347e-08,
+      "loss": 0.2615,
+      "step": 138820
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.234577626984151e-08,
+      "loss": 0.2465,
+      "step": 138825
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.212937367447603e-08,
+      "loss": 0.2319,
+      "step": 138830
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.191329475108032e-08,
+      "loss": 0.2629,
+      "step": 138835
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.169753950246605e-08,
+      "loss": 0.2428,
+      "step": 138840
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 7.14821079314254e-08,
+      "loss": 0.2525,
+      "step": 138845
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.126700004076448e-08,
+      "loss": 0.234,
+      "step": 138850
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.10522158332727e-08,
+      "loss": 0.2653,
+      "step": 138855
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.083775531173675e-08,
+      "loss": 0.2505,
+      "step": 138860
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.062361847894605e-08,
+      "loss": 0.2414,
+      "step": 138865
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.040980533767894e-08,
+      "loss": 0.2582,
+      "step": 138870
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 7.019631589070818e-08,
+      "loss": 0.2623,
+      "step": 138875
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.998315014081213e-08,
+      "loss": 0.2364,
+      "step": 138880
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.977030809075247e-08,
+      "loss": 0.2384,
+      "step": 138885
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.955778974329919e-08,
+      "loss": 0.2242,
+      "step": 138890
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.934559510120564e-08,
+      "loss": 0.235,
+      "step": 138895
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.913372416723074e-08,
+      "loss": 0.2355,
+      "step": 138900
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.892217694412505e-08,
+      "loss": 0.2448,
+      "step": 138905
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.871095343463362e-08,
+      "loss": 0.2716,
+      "step": 138910
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.850005364150147e-08,
+      "loss": 0.2524,
+      "step": 138915
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.828947756746252e-08,
+      "loss": 0.2654,
+      "step": 138920
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.807922521525622e-08,
+      "loss": 0.2407,
+      "step": 138925
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.786929658761099e-08,
+      "loss": 0.2612,
+      "step": 138930
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.765969168724962e-08,
+      "loss": 0.2588,
+      "step": 138935
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.745041051689772e-08,
+      "loss": 0.2408,
+      "step": 138940
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.724145307927254e-08,
+      "loss": 0.2578,
+      "step": 138945
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.703281937708306e-08,
+      "loss": 0.2552,
+      "step": 138950
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.682450941304374e-08,
+      "loss": 0.2701,
+      "step": 138955
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.661652318985523e-08,
+      "loss": 0.2682,
+      "step": 138960
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.640886071021813e-08,
+      "loss": 0.2446,
+      "step": 138965
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.620152197683305e-08,
+      "loss": 0.2478,
+      "step": 138970
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.599450699238674e-08,
+      "loss": 0.241,
+      "step": 138975
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.578781575956872e-08,
+      "loss": 0.2433,
+      "step": 138980
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.558144828106572e-08,
+      "loss": 0.2669,
+      "step": 138985
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.537540455955616e-08,
+      "loss": 0.2576,
+      "step": 138990
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.516968459771289e-08,
+      "loss": 0.2703,
+      "step": 138995
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.496428839820878e-08,
+      "loss": 0.2401,
+      "step": 139000
+    },
+    {
+      "epoch": 4.89,
+      "eval_loss": 0.24838781356811523,
+      "eval_runtime": 10.5455,
+      "eval_samples_per_second": 9.483,
+      "eval_steps_per_second": 9.483,
+      "step": 139000
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.475921596370838e-08,
+      "loss": 0.2372,
+      "step": 139005
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.455446729688175e-08,
+      "loss": 0.2433,
+      "step": 139010
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.435004240037957e-08,
+      "loss": 0.2662,
+      "step": 139015
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.414594127685802e-08,
+      "loss": 0.282,
+      "step": 139020
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.39421639289678e-08,
+      "loss": 0.2483,
+      "step": 139025
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.373871035935675e-08,
+      "loss": 0.2787,
+      "step": 139030
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.353558057066445e-08,
+      "loss": 0.2755,
+      "step": 139035
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.333277456552766e-08,
+      "loss": 0.2572,
+      "step": 139040
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.313029234658042e-08,
+      "loss": 0.242,
+      "step": 139045
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.292813391645391e-08,
+      "loss": 0.2458,
+      "step": 139050
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.272629927777108e-08,
+      "loss": 0.243,
+      "step": 139055
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.252478843315479e-08,
+      "loss": 0.2483,
+      "step": 139060
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.232360138521686e-08,
+      "loss": 0.2451,
+      "step": 139065
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.212273813657465e-08,
+      "loss": 0.2657,
+      "step": 139070
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.19221986898344e-08,
+      "loss": 0.2732,
+      "step": 139075
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.17219830475968e-08,
+      "loss": 0.2429,
+      "step": 139080
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.152209121246532e-08,
+      "loss": 0.2576,
+      "step": 139085
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.132252318703513e-08,
+      "loss": 0.2451,
+      "step": 139090
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.11232789738958e-08,
+      "loss": 0.244,
+      "step": 139095
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.092435857563695e-08,
+      "loss": 0.2442,
+      "step": 139100
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.072576199483982e-08,
+      "loss": 0.2445,
+      "step": 139105
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.052748923408291e-08,
+      "loss": 0.249,
+      "step": 139110
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.032954029593917e-08,
+      "loss": 0.2412,
+      "step": 139115
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 6.013191518298156e-08,
+      "loss": 0.2595,
+      "step": 139120
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.993461389777467e-08,
+      "loss": 0.2366,
+      "step": 139125
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 5.973763644288033e-08,
+      "loss": 0.2396,
+      "step": 139130
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.954098282085763e-08,
+      "loss": 0.2491,
+      "step": 139135
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.934465303425729e-08,
+      "loss": 0.2339,
+      "step": 139140
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.914864708563283e-08,
+      "loss": 0.2546,
+      "step": 139145
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.895296497752389e-08,
+      "loss": 0.2587,
+      "step": 139150
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.875760671247565e-08,
+      "loss": 0.252,
+      "step": 139155
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.85625722930222e-08,
+      "loss": 0.2302,
+      "step": 139160
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.8367861721697617e-08,
+      "loss": 0.2826,
+      "step": 139165
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.817347500102765e-08,
+      "loss": 0.2645,
+      "step": 139170
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.797941213353808e-08,
+      "loss": 0.2565,
+      "step": 139175
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.77856731217491e-08,
+      "loss": 0.2418,
+      "step": 139180
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.759225796817536e-08,
+      "loss": 0.25,
+      "step": 139185
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.7399166675328745e-08,
+      "loss": 0.258,
+      "step": 139190
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.7206399245715584e-08,
+      "loss": 0.2421,
+      "step": 139195
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.7013955681839426e-08,
+      "loss": 0.2445,
+      "step": 139200
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.682183598620106e-08,
+      "loss": 0.2674,
+      "step": 139205
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.6630040161290145e-08,
+      "loss": 0.2495,
+      "step": 139210
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.643856820960192e-08,
+      "loss": 0.2737,
+      "step": 139215
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.62474201336205e-08,
+      "loss": 0.262,
+      "step": 139220
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.6056595935827236e-08,
+      "loss": 0.244,
+      "step": 139225
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.586609561869793e-08,
+      "loss": 0.2663,
+      "step": 139230
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.567591918471116e-08,
+      "loss": 0.254,
+      "step": 139235
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.5486066636334374e-08,
+      "loss": 0.274,
+      "step": 139240
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.529653797602952e-08,
+      "loss": 0.2628,
+      "step": 139245
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.5107333206261266e-08,
+      "loss": 0.2555,
+      "step": 139250
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4918452329486004e-08,
+      "loss": 0.2574,
+      "step": 139255
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4729895348154534e-08,
+      "loss": 0.2486,
+      "step": 139260
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.454166226471491e-08,
+      "loss": 0.236,
+      "step": 139265
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.435375308161239e-08,
+      "loss": 0.2445,
+      "step": 139270
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.4166167801286697e-08,
+      "loss": 0.2466,
+      "step": 139275
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.397890642617476e-08,
+      "loss": 0.2489,
+      "step": 139280
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.3791968958705195e-08,
+      "loss": 0.2564,
+      "step": 139285
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.360535540130662e-08,
+      "loss": 0.2539,
+      "step": 139290
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.34190657564021e-08,
+      "loss": 0.2812,
+      "step": 139295
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.3233100026411906e-08,
+      "loss": 0.2725,
+      "step": 139300
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.3047458213748015e-08,
+      "loss": 0.2347,
+      "step": 139305
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.286214032082237e-08,
+      "loss": 0.2598,
+      "step": 139310
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.26771463500414e-08,
+      "loss": 0.2549,
+      "step": 139315
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.249247630380594e-08,
+      "loss": 0.2379,
+      "step": 139320
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.230813018451686e-08,
+      "loss": 0.261,
+      "step": 139325
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.212410799456391e-08,
+      "loss": 0.2443,
+      "step": 139330
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.1940409736339626e-08,
+      "loss": 0.2517,
+      "step": 139335
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.175703541222543e-08,
+      "loss": 0.277,
+      "step": 139340
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.1573985024605534e-08,
+      "loss": 0.2466,
+      "step": 139345
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.13912585758558e-08,
+      "loss": 0.2624,
+      "step": 139350
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.120885606834935e-08,
+      "loss": 0.2759,
+      "step": 139355
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.1026777504453724e-08,
+      "loss": 0.271,
+      "step": 139360
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.08450228865337e-08,
+      "loss": 0.2514,
+      "step": 139365
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.0663592216948496e-08,
+      "loss": 0.2702,
+      "step": 139370
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.0482485498054566e-08,
+      "loss": 0.2748,
+      "step": 139375
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.03017027322028e-08,
+      "loss": 0.2497,
+      "step": 139380
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 5.0121243921741337e-08,
+      "loss": 0.2428,
+      "step": 139385
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.9941109069012726e-08,
+      "loss": 0.2449,
+      "step": 139390
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.976129817635677e-08,
+      "loss": 0.2438,
+      "step": 139395
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.958181124610772e-08,
+      "loss": 0.2722,
+      "step": 139400
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.9402648280594265e-08,
+      "loss": 0.2562,
+      "step": 139405
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.922380928214787e-08,
+      "loss": 0.2715,
+      "step": 139410
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.9045294253083354e-08,
+      "loss": 0.2519,
+      "step": 139415
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.886710319572385e-08,
+      "loss": 0.2516,
+      "step": 139420
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.868923611238141e-08,
+      "loss": 0.269,
+      "step": 139425
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.851169300536529e-08,
+      "loss": 0.2386,
+      "step": 139430
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.833447387698198e-08,
+      "loss": 0.2752,
+      "step": 139435
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.815757872952964e-08,
+      "loss": 0.2407,
+      "step": 139440
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.798100756530921e-08,
+      "loss": 0.2601,
+      "step": 139445
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.78047603866133e-08,
+      "loss": 0.2495,
+      "step": 139450
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.762883719572619e-08,
+      "loss": 0.2573,
+      "step": 139455
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.745323799493495e-08,
+      "loss": 0.2502,
+      "step": 139460
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.72779627865183e-08,
+      "loss": 0.2501,
+      "step": 139465
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.7103011572752206e-08,
+      "loss": 0.2381,
+      "step": 139470
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.692838435590985e-08,
+      "loss": 0.2508,
+      "step": 139475
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.675408113825885e-08,
+      "loss": 0.2582,
+      "step": 139480
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.6580101922058527e-08,
+      "loss": 0.251,
+      "step": 139485
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.640644670957372e-08,
+      "loss": 0.2454,
+      "step": 139490
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.623311550305542e-08,
+      "loss": 0.2563,
+      "step": 139495
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.6060108304754604e-08,
+      "loss": 0.2511,
+      "step": 139500
+    },
+    {
+      "epoch": 4.91,
+      "eval_loss": 0.24838365614414215,
+      "eval_runtime": 10.5546,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 9.475,
+      "step": 139500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.5887425116916685e-08,
+      "loss": 0.2367,
+      "step": 139505
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.5715065941787095e-08,
+      "loss": 0.2403,
+      "step": 139510
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.554303078160016e-08,
+      "loss": 0.2501,
+      "step": 139515
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.537131963859298e-08,
+      "loss": 0.2529,
+      "step": 139520
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.519993251499155e-08,
+      "loss": 0.2343,
+      "step": 139525
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.502886941302464e-08,
+      "loss": 0.2539,
+      "step": 139530
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.485813033491271e-08,
+      "loss": 0.2312,
+      "step": 139535
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.468771528287064e-08,
+      "loss": 0.2603,
+      "step": 139540
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.451762425911332e-08,
+      "loss": 0.2469,
+      "step": 139545
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.434785726584734e-08,
+      "loss": 0.2494,
+      "step": 139550
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.417841430527925e-08,
+      "loss": 0.247,
+      "step": 139555
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.40092953796073e-08,
+      "loss": 0.2504,
+      "step": 139560
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.3840500491026946e-08,
+      "loss": 0.2637,
+      "step": 139565
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.3672029641730896e-08,
+      "loss": 0.2409,
+      "step": 139570
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.350388283390905e-08,
+      "loss": 0.2413,
+      "step": 139575
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.3336060069740224e-08,
+      "loss": 0.2618,
+      "step": 139580
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.3168561351406014e-08,
+      "loss": 0.2557,
+      "step": 139585
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.3001386681082444e-08,
+      "loss": 0.2262,
+      "step": 139590
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.2834536060937235e-08,
+      "loss": 0.2644,
+      "step": 139595
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.26680094931381e-08,
+      "loss": 0.2536,
+      "step": 139600
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.250180697984718e-08,
+      "loss": 0.2648,
+      "step": 139605
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.233592852322388e-08,
+      "loss": 0.2665,
+      "step": 139610
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.217037412541924e-08,
+      "loss": 0.2514,
+      "step": 139615
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.200514378858433e-08,
+      "loss": 0.2558,
+      "step": 139620
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.184023751486743e-08,
+      "loss": 0.2542,
+      "step": 139625
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.167565530640571e-08,
+      "loss": 0.252,
+      "step": 139630
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.151139716533636e-08,
+      "loss": 0.2377,
+      "step": 139635
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.134746309379378e-08,
+      "loss": 0.2441,
+      "step": 139640
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.118385309390682e-08,
+      "loss": 0.2431,
+      "step": 139645
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.102056716779601e-08,
+      "loss": 0.2303,
+      "step": 139650
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.0857605317587424e-08,
+      "loss": 0.2495,
+      "step": 139655
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.069496754539326e-08,
+      "loss": 0.2484,
+      "step": 139660
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.0532653853328495e-08,
+      "loss": 0.2516,
+      "step": 139665
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.037066424349423e-08,
+      "loss": 0.2341,
+      "step": 139670
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.020899871799988e-08,
+      "loss": 0.2712,
+      "step": 139675
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 4.0047657278943774e-08,
+      "loss": 0.2612,
+      "step": 139680
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.988663992841868e-08,
+      "loss": 0.2186,
+      "step": 139685
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.972594666851459e-08,
+      "loss": 0.2565,
+      "step": 139690
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.9565577501321505e-08,
+      "loss": 0.253,
+      "step": 139695
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.940553242891831e-08,
+      "loss": 0.25,
+      "step": 139700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.924581145338668e-08,
+      "loss": 0.2526,
+      "step": 139705
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.908641457679718e-08,
+      "loss": 0.2363,
+      "step": 139710
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.892734180122037e-08,
+      "loss": 0.2487,
+      "step": 139715
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.8768593128724054e-08,
+      "loss": 0.2375,
+      "step": 139720
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.861016856136768e-08,
+      "loss": 0.2468,
+      "step": 139725
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.8452068101205165e-08,
+      "loss": 0.2504,
+      "step": 139730
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.82942917502932e-08,
+      "loss": 0.2475,
+      "step": 139735
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.8136839510680145e-08,
+      "loss": 0.2352,
+      "step": 139740
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.7979711384408814e-08,
+      "loss": 0.2396,
+      "step": 139745
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.782290737352201e-08,
+      "loss": 0.2344,
+      "step": 139750
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.7666427480054224e-08,
+      "loss": 0.2612,
+      "step": 139755
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.751027170603438e-08,
+      "loss": 0.2375,
+      "step": 139760
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.73544400534942e-08,
+      "loss": 0.2605,
+      "step": 139765
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.719893252445705e-08,
+      "loss": 0.2591,
+      "step": 139770
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.7043749120938e-08,
+      "loss": 0.2505,
+      "step": 139775
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.688888984495764e-08,
+      "loss": 0.2627,
+      "step": 139780
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.673435469851994e-08,
+      "loss": 0.2744,
+      "step": 139785
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.658014368363993e-08,
+      "loss": 0.2597,
+      "step": 139790
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.642625680231049e-08,
+      "loss": 0.2467,
+      "step": 139795
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.627269405653832e-08,
+      "loss": 0.2613,
+      "step": 139800
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.6119455448310746e-08,
+      "loss": 0.2453,
+      "step": 139805
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.59665409796206e-08,
+      "loss": 0.2469,
+      "step": 139810
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.581395065245519e-08,
+      "loss": 0.2677,
+      "step": 139815
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.566168446879348e-08,
+      "loss": 0.2571,
+      "step": 139820
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.5509742430608894e-08,
+      "loss": 0.2689,
+      "step": 139825
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.535812453988319e-08,
+      "loss": 0.2546,
+      "step": 139830
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.52068307985759e-08,
+      "loss": 0.2396,
+      "step": 139835
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.505586120865767e-08,
+      "loss": 0.2477,
+      "step": 139840
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.4905215772085275e-08,
+      "loss": 0.2458,
+      "step": 139845
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.475489449081826e-08,
+      "loss": 0.2482,
+      "step": 139850
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.460489736680228e-08,
+      "loss": 0.2831,
+      "step": 139855
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.445522440199134e-08,
+      "loss": 0.2551,
+      "step": 139860
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.4305875598325545e-08,
+      "loss": 0.2651,
+      "step": 139865
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.415685095774779e-08,
+      "loss": 0.2481,
+      "step": 139870
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.400815048218708e-08,
+      "loss": 0.2391,
+      "step": 139875
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3859774173577996e-08,
+      "loss": 0.2464,
+      "step": 139880
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.371172203384676e-08,
+      "loss": 0.2561,
+      "step": 139885
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.356399406491684e-08,
+      "loss": 0.264,
+      "step": 139890
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3416590268703364e-08,
+      "loss": 0.2389,
+      "step": 139895
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3269510647121474e-08,
+      "loss": 0.2743,
+      "step": 139900
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.312275520208352e-08,
+      "loss": 0.2377,
+      "step": 139905
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.297632393549077e-08,
+      "loss": 0.2321,
+      "step": 139910
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.283021684924448e-08,
+      "loss": 0.2727,
+      "step": 139915
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.268443394524867e-08,
+      "loss": 0.2393,
+      "step": 139920
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.2538975225387956e-08,
+      "loss": 0.2432,
+      "step": 139925
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.2393840691555266e-08,
+      "loss": 0.2301,
+      "step": 139930
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.22490303456352e-08,
+      "loss": 0.2793,
+      "step": 139935
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.2104544189506816e-08,
+      "loss": 0.251,
+      "step": 139940
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.196038222504638e-08,
+      "loss": 0.2762,
+      "step": 139945
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.181654445412463e-08,
+      "loss": 0.2547,
+      "step": 139950
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.1673030878612286e-08,
+      "loss": 0.2638,
+      "step": 139955
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.1529841500368975e-08,
+      "loss": 0.2362,
+      "step": 139960
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.138697632125709e-08,
+      "loss": 0.2567,
+      "step": 139965
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.1244435343130706e-08,
+      "loss": 0.2716,
+      "step": 139970
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.110221856783835e-08,
+      "loss": 0.2464,
+      "step": 139975
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.096032599723131e-08,
+      "loss": 0.2455,
+      "step": 139980
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.081875763314701e-08,
+      "loss": 0.2536,
+      "step": 139985
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.067751347742842e-08,
+      "loss": 0.2541,
+      "step": 139990
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.053659353190463e-08,
+      "loss": 0.2264,
+      "step": 139995
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.039599779840752e-08,
+      "loss": 0.2472,
+      "step": 140000
+    },
+    {
+      "epoch": 4.93,
+      "eval_loss": 0.2483794391155243,
+      "eval_runtime": 10.5607,
+      "eval_samples_per_second": 9.469,
+      "eval_steps_per_second": 9.469,
+      "step": 140000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.025572627876339e-08,
+      "loss": 0.2355,
+      "step": 140005
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.011577897479301e-08,
+      "loss": 0.2437,
+      "step": 140010
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.997615588831437e-08,
+      "loss": 0.2383,
+      "step": 140015
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.9836857021137147e-08,
+      "loss": 0.2328,
+      "step": 140020
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.969788237507376e-08,
+      "loss": 0.2205,
+      "step": 140025
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.9559231951928334e-08,
+      "loss": 0.2371,
+      "step": 140030
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.942090575349943e-08,
+      "loss": 0.2592,
+      "step": 140035
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.9282903781582828e-08,
+      "loss": 0.2468,
+      "step": 140040
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.9145226037974317e-08,
+      "loss": 0.2362,
+      "step": 140045
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.9007872524455805e-08,
+      "loss": 0.2539,
+      "step": 140050
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.887084324281475e-08,
+      "loss": 0.2358,
+      "step": 140055
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.873413819482751e-08,
+      "loss": 0.2481,
+      "step": 140060
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.8597757382273216e-08,
+      "loss": 0.2433,
+      "step": 140065
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.8461700806917123e-08,
+      "loss": 0.2358,
+      "step": 140070
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.8325968470532816e-08,
+      "loss": 0.2537,
+      "step": 140075
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.8190560374877217e-08,
+      "loss": 0.2366,
+      "step": 140080
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.8055476521710034e-08,
+      "loss": 0.2612,
+      "step": 140085
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7920716912785416e-08,
+      "loss": 0.2616,
+      "step": 140090
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7786281549851965e-08,
+      "loss": 0.2407,
+      "step": 140095
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7652170434658285e-08,
+      "loss": 0.253,
+      "step": 140100
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7518383568941875e-08,
+      "loss": 0.2263,
+      "step": 140105
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7384920954443006e-08,
+      "loss": 0.2594,
+      "step": 140110
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7251782592893627e-08,
+      "loss": 0.2574,
+      "step": 140115
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7118968486020136e-08,
+      "loss": 0.2437,
+      "step": 140120
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.6986478635551705e-08,
+      "loss": 0.2568,
+      "step": 140125
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.6854313043203627e-08,
+      "loss": 0.2295,
+      "step": 140130
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.672247171069675e-08,
+      "loss": 0.2409,
+      "step": 140135
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.6590954639738043e-08,
+      "loss": 0.2298,
+      "step": 140140
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.645976183204002e-08,
+      "loss": 0.2613,
+      "step": 140145
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.6328893289301326e-08,
+      "loss": 0.2364,
+      "step": 140150
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.6198349013226154e-08,
+      "loss": 0.2616,
+      "step": 140155
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.606812900550759e-08,
+      "loss": 0.252,
+      "step": 140160
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.593823326783318e-08,
+      "loss": 0.2611,
+      "step": 140165
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.580866180189323e-08,
+      "loss": 0.2567,
+      "step": 140170
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5679414609369734e-08,
+      "loss": 0.2269,
+      "step": 140175
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.55504916919419e-08,
+      "loss": 0.2433,
+      "step": 140180
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5421893051280622e-08,
+      "loss": 0.2451,
+      "step": 140185
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5293618689056775e-08,
+      "loss": 0.2458,
+      "step": 140190
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.51656686069357e-08,
+      "loss": 0.2379,
+      "step": 140195
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5038042806579953e-08,
+      "loss": 0.2551,
+      "step": 140200
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.4910741289646543e-08,
+      "loss": 0.2582,
+      "step": 140205
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.478376405778693e-08,
+      "loss": 0.243,
+      "step": 140210
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.4657111112652566e-08,
+      "loss": 0.2432,
+      "step": 140215
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.4530782455886582e-08,
+      "loss": 0.2447,
+      "step": 140220
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.4404778089126555e-08,
+      "loss": 0.252,
+      "step": 140225
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.4279098014012847e-08,
+      "loss": 0.2502,
+      "step": 140230
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.41537422321747e-08,
+      "loss": 0.2246,
+      "step": 140235
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.4028710745241378e-08,
+      "loss": 0.2442,
+      "step": 140240
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.39040035548338e-08,
+      "loss": 0.2282,
+      "step": 140245
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3779620662572888e-08,
+      "loss": 0.248,
+      "step": 140250
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.36555620700768e-08,
+      "loss": 0.2496,
+      "step": 140255
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.35318277789498e-08,
+      "loss": 0.2691,
+      "step": 140260
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.3408417790801716e-08,
+      "loss": 0.2883,
+      "step": 140265
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.3285332107236823e-08,
+      "loss": 0.2487,
+      "step": 140270
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.316257072985384e-08,
+      "loss": 0.2364,
+      "step": 140275
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.304013366024038e-08,
+      "loss": 0.2636,
+      "step": 140280
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.29180208999924e-08,
+      "loss": 0.2453,
+      "step": 140285
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2796232450691956e-08,
+      "loss": 0.2455,
+      "step": 140290
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2674768313923898e-08,
+      "loss": 0.2399,
+      "step": 140295
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.255362849126197e-08,
+      "loss": 0.2482,
+      "step": 140300
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2432812984279905e-08,
+      "loss": 0.2399,
+      "step": 140305
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2312321794548673e-08,
+      "loss": 0.2593,
+      "step": 140310
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.219215492362814e-08,
+      "loss": 0.239,
+      "step": 140315
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2072312373083716e-08,
+      "loss": 0.261,
+      "step": 140320
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.1952794144469713e-08,
+      "loss": 0.2639,
+      "step": 140325
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.183360023933767e-08,
+      "loss": 0.2514,
+      "step": 140330
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.1714730659233573e-08,
+      "loss": 0.2684,
+      "step": 140335
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.1596185405703407e-08,
+      "loss": 0.2716,
+      "step": 140340
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.1477964480284827e-08,
+      "loss": 0.2562,
+      "step": 140345
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.1360067884512725e-08,
+      "loss": 0.248,
+      "step": 140350
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.12424956199192e-08,
+      "loss": 0.239,
+      "step": 140355
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.1125247688030812e-08,
+      "loss": 0.2444,
+      "step": 140360
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.1008324090368568e-08,
+      "loss": 0.2539,
+      "step": 140365
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.0891724828453476e-08,
+      "loss": 0.2366,
+      "step": 140370
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.077544990379543e-08,
+      "loss": 0.2539,
+      "step": 140375
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.0659499317907116e-08,
+      "loss": 0.2419,
+      "step": 140380
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.0543873072292887e-08,
+      "loss": 0.2431,
+      "step": 140385
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.0428571168454312e-08,
+      "loss": 0.2521,
+      "step": 140390
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.0313593607887427e-08,
+      "loss": 0.2359,
+      "step": 140395
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.019894039208825e-08,
+      "loss": 0.2645,
+      "step": 140400
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.008461152254171e-08,
+      "loss": 0.2329,
+      "step": 140405
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.9970607000735497e-08,
+      "loss": 0.2538,
+      "step": 140410
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.9856926828148993e-08,
+      "loss": 0.2494,
+      "step": 140415
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.9743571006256014e-08,
+      "loss": 0.2461,
+      "step": 140420
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.963053953653038e-08,
+      "loss": 0.2636,
+      "step": 140425
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.9517832420440363e-08,
+      "loss": 0.25,
+      "step": 140430
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.940544965944868e-08,
+      "loss": 0.2498,
+      "step": 140435
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.929339125501528e-08,
+      "loss": 0.2806,
+      "step": 140440
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.9181657208591775e-08,
+      "loss": 0.2557,
+      "step": 140445
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.9070247521632556e-08,
+      "loss": 0.2491,
+      "step": 140450
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.895916219558369e-08,
+      "loss": 0.2463,
+      "step": 140455
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.8848401231885692e-08,
+      "loss": 0.2464,
+      "step": 140460
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.8737964631979078e-08,
+      "loss": 0.2438,
+      "step": 140465
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.8627852397298808e-08,
+      "loss": 0.2542,
+      "step": 140470
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.851806452926874e-08,
+      "loss": 0.2692,
+      "step": 140475
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.840860102932107e-08,
+      "loss": 0.2289,
+      "step": 140480
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.829946189887133e-08,
+      "loss": 0.2403,
+      "step": 140485
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.8190647139343376e-08,
+      "loss": 0.2488,
+      "step": 140490
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.8082156752141642e-08,
+      "loss": 0.2613,
+      "step": 140495
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.7973990738684443e-08,
+      "loss": 0.2542,
+      "step": 140500
+    },
+    {
+      "epoch": 4.94,
+      "eval_loss": 0.24837003648281097,
+      "eval_runtime": 10.5523,
+      "eval_samples_per_second": 9.477,
+      "eval_steps_per_second": 9.477,
+      "step": 140500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.786614910036788e-08,
+      "loss": 0.2355,
+      "step": 140505
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.7758631838596385e-08,
+      "loss": 0.2509,
+      "step": 140510
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.7651438954763287e-08,
+      "loss": 0.2399,
+      "step": 140515
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.7544570450264697e-08,
+      "loss": 0.2539,
+      "step": 140520
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.7438026326485612e-08,
+      "loss": 0.262,
+      "step": 140525
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.7331806584808263e-08,
+      "loss": 0.234,
+      "step": 140530
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.7225911226614877e-08,
+      "loss": 0.2576,
+      "step": 140535
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.712034025327658e-08,
+      "loss": 0.2385,
+      "step": 140540
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.701509366616727e-08,
+      "loss": 0.2688,
+      "step": 140545
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.69101714666553e-08,
+      "loss": 0.247,
+      "step": 140550
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6805573656095142e-08,
+      "loss": 0.2424,
+      "step": 140555
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6701300235855144e-08,
+      "loss": 0.2788,
+      "step": 140560
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6597351207281452e-08,
+      "loss": 0.2521,
+      "step": 140565
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6493726571725766e-08,
+      "loss": 0.2425,
+      "step": 140570
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6390426330534226e-08,
+      "loss": 0.2437,
+      "step": 140575
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6287450485050205e-08,
+      "loss": 0.2269,
+      "step": 140580
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6184799036608745e-08,
+      "loss": 0.2454,
+      "step": 140585
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6082471986542115e-08,
+      "loss": 0.2661,
+      "step": 140590
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5980469336179805e-08,
+      "loss": 0.2495,
+      "step": 140595
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5878791086845758e-08,
+      "loss": 0.2439,
+      "step": 140600
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5777437239858363e-08,
+      "loss": 0.248,
+      "step": 140605
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5676407796538784e-08,
+      "loss": 0.2583,
+      "step": 140610
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.557570275819431e-08,
+      "loss": 0.2516,
+      "step": 140615
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5475322126135006e-08,
+      "loss": 0.2551,
+      "step": 140620
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5375265901662604e-08,
+      "loss": 0.2522,
+      "step": 140625
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.527553408607607e-08,
+      "loss": 0.2352,
+      "step": 140630
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5176126680671587e-08,
+      "loss": 0.2735,
+      "step": 140635
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5077043686739788e-08,
+      "loss": 0.2506,
+      "step": 140640
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.4978285105565758e-08,
+      "loss": 0.2519,
+      "step": 140645
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.487985093843458e-08,
+      "loss": 0.2482,
+      "step": 140650
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.478174118662301e-08,
+      "loss": 0.2525,
+      "step": 140655
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.4683955851402254e-08,
+      "loss": 0.2385,
+      "step": 140660
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.458649493404629e-08,
+      "loss": 0.2423,
+      "step": 140665
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.4489358435818e-08,
+      "loss": 0.2522,
+      "step": 140670
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.4392546357980263e-08,
+      "loss": 0.2519,
+      "step": 140675
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.429605870178763e-08,
+      "loss": 0.2423,
+      "step": 140680
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.4199895468497426e-08,
+      "loss": 0.2499,
+      "step": 140685
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.4104056659353104e-08,
+      "loss": 0.2446,
+      "step": 140690
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.400854227560089e-08,
+      "loss": 0.2484,
+      "step": 140695
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3913352318481453e-08,
+      "loss": 0.2521,
+      "step": 140700
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3818486789232699e-08,
+      "loss": 0.2474,
+      "step": 140705
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.372394568908142e-08,
+      "loss": 0.2445,
+      "step": 140710
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3629729019259962e-08,
+      "loss": 0.251,
+      "step": 140715
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3535836780989575e-08,
+      "loss": 0.2597,
+      "step": 140720
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3442268975488726e-08,
+      "loss": 0.25,
+      "step": 140725
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3349025603975884e-08,
+      "loss": 0.2393,
+      "step": 140730
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3256106667655644e-08,
+      "loss": 0.2513,
+      "step": 140735
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3163512167740922e-08,
+      "loss": 0.2491,
+      "step": 140740
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.3071242105427983e-08,
+      "loss": 0.2626,
+      "step": 140745
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2979296481921422e-08,
+      "loss": 0.2549,
+      "step": 140750
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2887675298409175e-08,
+      "loss": 0.2489,
+      "step": 140755
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2796378556084732e-08,
+      "loss": 0.2622,
+      "step": 140760
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.270540625613048e-08,
+      "loss": 0.2323,
+      "step": 140765
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2614758399731585e-08,
+      "loss": 0.2562,
+      "step": 140770
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2524434988062106e-08,
+      "loss": 0.2434,
+      "step": 140775
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2434436022293327e-08,
+      "loss": 0.2521,
+      "step": 140780
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2344761503599312e-08,
+      "loss": 0.2525,
+      "step": 140785
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2255411433137464e-08,
+      "loss": 0.2441,
+      "step": 140790
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2166385812073522e-08,
+      "loss": 0.2572,
+      "step": 140795
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.2077684641562114e-08,
+      "loss": 0.2391,
+      "step": 140800
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.1989307922752324e-08,
+      "loss": 0.2428,
+      "step": 140805
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.1901255656796006e-08,
+      "loss": 0.2312,
+      "step": 140810
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.1813527844833916e-08,
+      "loss": 0.2521,
+      "step": 140815
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.1726124488004032e-08,
+      "loss": 0.2576,
+      "step": 140820
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.1639045587441554e-08,
+      "loss": 0.2525,
+      "step": 140825
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.1552291144278915e-08,
+      "loss": 0.2638,
+      "step": 140830
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.1465861159642988e-08,
+      "loss": 0.2554,
+      "step": 140835
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.1379755634655099e-08,
+      "loss": 0.2675,
+      "step": 140840
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.1293974570431021e-08,
+      "loss": 0.2464,
+      "step": 140845
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.120851796808653e-08,
+      "loss": 0.2727,
+      "step": 140850
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.1123385828734622e-08,
+      "loss": 0.2349,
+      "step": 140855
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.103857815347442e-08,
+      "loss": 0.257,
+      "step": 140860
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0954094943410598e-08,
+      "loss": 0.2598,
+      "step": 140865
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.086993619963672e-08,
+      "loss": 0.2654,
+      "step": 140870
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0786101923251912e-08,
+      "loss": 0.2598,
+      "step": 140875
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0702592115338639e-08,
+      "loss": 0.2388,
+      "step": 140880
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.061940677698492e-08,
+      "loss": 0.2646,
+      "step": 140885
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0536545909270446e-08,
+      "loss": 0.2582,
+      "step": 140890
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0454009513269359e-08,
+      "loss": 0.2524,
+      "step": 140895
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0371797590053023e-08,
+      "loss": 0.2725,
+      "step": 140900
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0289910140692804e-08,
+      "loss": 0.226,
+      "step": 140905
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.020834716624619e-08,
+      "loss": 0.2708,
+      "step": 140910
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0127108667778994e-08,
+      "loss": 0.2469,
+      "step": 140915
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0046194646340379e-08,
+      "loss": 0.2362,
+      "step": 140920
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.965605102982278e-09,
+      "loss": 0.2453,
+      "step": 140925
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.885340038751079e-09,
+      "loss": 0.2572,
+      "step": 140930
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.805399454690389e-09,
+      "loss": 0.2479,
+      "step": 140935
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.725783351838268e-09,
+      "loss": 0.2517,
+      "step": 140940
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.646491731227225e-09,
+      "loss": 0.247,
+      "step": 140945
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.567524593886989e-09,
+      "loss": 0.2489,
+      "step": 140950
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.488881940844518e-09,
+      "loss": 0.257,
+      "step": 140955
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.410563773118442e-09,
+      "loss": 0.2491,
+      "step": 140960
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.332570091724612e-09,
+      "loss": 0.2267,
+      "step": 140965
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.254900897681662e-09,
+      "loss": 0.2503,
+      "step": 140970
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.177556191991565e-09,
+      "loss": 0.2487,
+      "step": 140975
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.100535975661851e-09,
+      "loss": 0.2439,
+      "step": 140980
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.023840249694492e-09,
+      "loss": 0.2705,
+      "step": 140985
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.947469015080367e-09,
+      "loss": 0.2442,
+      "step": 140990
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.871422272813124e-09,
+      "loss": 0.2463,
+      "step": 140995
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.795700023883634e-09,
+      "loss": 0.2905,
+      "step": 141000
+    },
+    {
+      "epoch": 4.96,
+      "eval_loss": 0.24837106466293335,
+      "eval_runtime": 10.5397,
+      "eval_samples_per_second": 9.488,
+      "eval_steps_per_second": 9.488,
+      "step": 141000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.720302269268899e-09,
+      "loss": 0.2621,
+      "step": 141005
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.645229009954236e-09,
+      "loss": 0.2356,
+      "step": 141010
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.570480246908318e-09,
+      "loss": 0.253,
+      "step": 141015
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.496055981105367e-09,
+      "loss": 0.2589,
+      "step": 141020
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.421956213514049e-09,
+      "loss": 0.2399,
+      "step": 141025
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.348180945091932e-09,
+      "loss": 0.2517,
+      "step": 141030
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.274730176796586e-09,
+      "loss": 0.2442,
+      "step": 141035
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.201603909585576e-09,
+      "loss": 0.2488,
+      "step": 141040
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.128802144405367e-09,
+      "loss": 0.261,
+      "step": 141045
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 8.056324882202425e-09,
+      "loss": 0.2444,
+      "step": 141050
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.984172123920441e-09,
+      "loss": 0.2682,
+      "step": 141055
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.912343870492001e-09,
+      "loss": 0.2582,
+      "step": 141060
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.84084012284969e-09,
+      "loss": 0.2334,
+      "step": 141065
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.7696608819261e-09,
+      "loss": 0.268,
+      "step": 141070
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.698806148639937e-09,
+      "loss": 0.262,
+      "step": 141075
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.628275923918238e-09,
+      "loss": 0.2562,
+      "step": 141080
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.558070208671385e-09,
+      "loss": 0.2562,
+      "step": 141085
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.488189003809765e-09,
+      "loss": 0.2381,
+      "step": 141090
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.4186323102465315e-09,
+      "loss": 0.2648,
+      "step": 141095
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.3494001288809676e-09,
+      "loss": 0.2472,
+      "step": 141100
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.2804924606095785e-09,
+      "loss": 0.2379,
+      "step": 141105
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.211909306334419e-09,
+      "loss": 0.2699,
+      "step": 141110
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.1436506669381175e-09,
+      "loss": 0.2614,
+      "step": 141115
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 7.0757165433144035e-09,
+      "loss": 0.2625,
+      "step": 141120
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.008106936337577e-09,
+      "loss": 0.2355,
+      "step": 141125
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.940821846893042e-09,
+      "loss": 0.2697,
+      "step": 141130
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.873861275849547e-09,
+      "loss": 0.2454,
+      "step": 141135
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.807225224078617e-09,
+      "loss": 0.2598,
+      "step": 141140
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.7409136924434516e-09,
+      "loss": 0.2537,
+      "step": 141145
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.674926681804472e-09,
+      "loss": 0.2552,
+      "step": 141150
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.609264193024878e-09,
+      "loss": 0.2453,
+      "step": 141155
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.543926226948438e-09,
+      "loss": 0.2443,
+      "step": 141160
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.478912784430025e-09,
+      "loss": 0.241,
+      "step": 141165
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.414223866313407e-09,
+      "loss": 0.2486,
+      "step": 141170
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.349859473434028e-09,
+      "loss": 0.2404,
+      "step": 141175
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.285819606630105e-09,
+      "loss": 0.2474,
+      "step": 141180
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.2221042667343075e-09,
+      "loss": 0.2452,
+      "step": 141185
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.15871345457375e-09,
+      "loss": 0.2747,
+      "step": 141190
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.0956471709699985e-09,
+      "loss": 0.2486,
+      "step": 141195
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 6.032905416741841e-09,
+      "loss": 0.2398,
+      "step": 141200
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.970488192705293e-09,
+      "loss": 0.2424,
+      "step": 141205
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.908395499670815e-09,
+      "loss": 0.2566,
+      "step": 141210
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.846627338443322e-09,
+      "loss": 0.2596,
+      "step": 141215
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.7851837098249486e-09,
+      "loss": 0.2413,
+      "step": 141220
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.724064614615054e-09,
+      "loss": 0.2422,
+      "step": 141225
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.6632700536074505e-09,
+      "loss": 0.2516,
+      "step": 141230
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.602800027587618e-09,
+      "loss": 0.2625,
+      "step": 141235
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.542654537346592e-09,
+      "loss": 0.2523,
+      "step": 141240
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.482833583661529e-09,
+      "loss": 0.2497,
+      "step": 141245
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.423337167309583e-09,
+      "loss": 0.2619,
+      "step": 141250
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.364165289062362e-09,
+      "loss": 0.253,
+      "step": 141255
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.305317949691468e-09,
+      "loss": 0.2374,
+      "step": 141260
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.246795149957407e-09,
+      "loss": 0.2619,
+      "step": 141265
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.188596890620678e-09,
+      "loss": 0.2507,
+      "step": 141270
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.130723172439011e-09,
+      "loss": 0.2566,
+      "step": 141275
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.073173996161806e-09,
+      "loss": 0.2527,
+      "step": 141280
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 5.015949362538464e-09,
+      "loss": 0.2586,
+      "step": 141285
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.959049272310057e-09,
+      "loss": 0.254,
+      "step": 141290
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.90247372621766e-09,
+      "loss": 0.2495,
+      "step": 141295
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.846222724991245e-09,
+      "loss": 0.2651,
+      "step": 141300
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.790296269366334e-09,
+      "loss": 0.2469,
+      "step": 141305
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.734694360067349e-09,
+      "loss": 0.2607,
+      "step": 141310
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.679416997813158e-09,
+      "loss": 0.2666,
+      "step": 141315
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.624464183325406e-09,
+      "loss": 0.2676,
+      "step": 141320
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.569835917317411e-09,
+      "loss": 0.2595,
+      "step": 141325
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.5155322004969415e-09,
+      "loss": 0.2523,
+      "step": 141330
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.461553033571764e-09,
+      "loss": 0.2547,
+      "step": 141335
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.407898417238543e-09,
+      "loss": 0.2426,
+      "step": 141340
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.354568352193944e-09,
+      "loss": 0.25,
+      "step": 141345
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.301562839134632e-09,
+      "loss": 0.2483,
+      "step": 141350
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.248881878748945e-09,
+      "loss": 0.2567,
+      "step": 141355
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.196525471714119e-09,
+      "loss": 0.2535,
+      "step": 141360
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.1444936187184925e-09,
+      "loss": 0.2507,
+      "step": 141365
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.092786320430974e-09,
+      "loss": 0.2381,
+      "step": 141370
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.041403577528802e-09,
+      "loss": 0.2421,
+      "step": 141375
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.9903453906725565e-09,
+      "loss": 0.2582,
+      "step": 141380
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.939611760531148e-09,
+      "loss": 0.2553,
+      "step": 141385
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.889202687759607e-09,
+      "loss": 0.2348,
+      "step": 141390
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.839118173015743e-09,
+      "loss": 0.2817,
+      "step": 141395
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.789358216946259e-09,
+      "loss": 0.2468,
+      "step": 141400
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.7399228201978606e-09,
+      "loss": 0.258,
+      "step": 141405
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.6908119834144772e-09,
+      "loss": 0.2321,
+      "step": 141410
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.642025707231711e-09,
+      "loss": 0.2494,
+      "step": 141415
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.593563992285165e-09,
+      "loss": 0.2506,
+      "step": 141420
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.5454268392048906e-09,
+      "loss": 0.2526,
+      "step": 141425
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.4976142486098374e-09,
+      "loss": 0.2732,
+      "step": 141430
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.450126221127281e-09,
+      "loss": 0.224,
+      "step": 141435
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.40296275737062e-09,
+      "loss": 0.2584,
+      "step": 141440
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.356123857953253e-09,
+      "loss": 0.2479,
+      "step": 141445
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3096095234858016e-09,
+      "loss": 0.2733,
+      "step": 141450
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.2634197545650115e-09,
+      "loss": 0.2542,
+      "step": 141455
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.217554551798729e-09,
+      "loss": 0.2575,
+      "step": 141460
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.1720139157781493e-09,
+      "loss": 0.2401,
+      "step": 141465
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.126797847094465e-09,
+      "loss": 0.2382,
+      "step": 141470
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.0819063463360943e-09,
+      "loss": 0.259,
+      "step": 141475
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.0373394140859048e-09,
+      "loss": 0.2539,
+      "step": 141480
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.9930970509239874e-09,
+      "loss": 0.243,
+      "step": 141485
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.9491792574221077e-09,
+      "loss": 0.2471,
+      "step": 141490
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.9055860341492546e-09,
+      "loss": 0.2529,
+      "step": 141495
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.8623173816744175e-09,
+      "loss": 0.2396,
+      "step": 141500
+    },
+    {
+      "epoch": 4.98,
+      "eval_loss": 0.2483615279197693,
+      "eval_runtime": 10.5543,
+      "eval_samples_per_second": 9.475,
+      "eval_steps_per_second": 9.475,
+      "step": 141500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.8193733005610346e-09,
+      "loss": 0.2468,
+      "step": 141505
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.7767537913614418e-09,
+      "loss": 0.259,
+      "step": 141510
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.734458854633526e-09,
+      "loss": 0.2438,
+      "step": 141515
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.692488490924072e-09,
+      "loss": 0.2723,
+      "step": 141520
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.6508427007798654e-09,
+      "loss": 0.2448,
+      "step": 141525
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.6095214847393635e-09,
+      "loss": 0.2505,
+      "step": 141530
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.568524843341025e-09,
+      "loss": 0.2619,
+      "step": 141535
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.5278527771177562e-09,
+      "loss": 0.2537,
+      "step": 141540
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.4875052865941386e-09,
+      "loss": 0.264,
+      "step": 141545
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.447482372297527e-09,
+      "loss": 0.2323,
+      "step": 141550
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.4077840347441762e-09,
+      "loss": 0.2491,
+      "step": 141555
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.36841027445589e-09,
+      "loss": 0.2279,
+      "step": 141560
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.329361091935045e-09,
+      "loss": 0.2559,
+      "step": 141565
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.290636487697895e-09,
+      "loss": 0.2693,
+      "step": 141570
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.2522364622384885e-09,
+      "loss": 0.2513,
+      "step": 141575
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.2141610160619773e-09,
+      "loss": 0.26,
+      "step": 141580
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.176410149659636e-09,
+      "loss": 0.2768,
+      "step": 141585
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1389838635227366e-09,
+      "loss": 0.2432,
+      "step": 141590
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.101882158137003e-09,
+      "loss": 0.2819,
+      "step": 141595
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.0651050339853815e-09,
+      "loss": 0.2579,
+      "step": 141600
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.028652491542493e-09,
+      "loss": 0.247,
+      "step": 141605
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.9925245312829577e-09,
+      "loss": 0.2499,
+      "step": 141610
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.95672115367862e-09,
+      "loss": 0.2774,
+      "step": 141615
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.921242359190223e-09,
+      "loss": 0.2289,
+      "step": 141620
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.886088148278509e-09,
+      "loss": 0.2339,
+      "step": 141625
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.85125852140422e-09,
+      "loss": 0.251,
+      "step": 141630
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.8167534790169971e-09,
+      "loss": 0.2504,
+      "step": 141635
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.782573021563705e-09,
+      "loss": 0.2926,
+      "step": 141640
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7487171494884325e-09,
+      "loss": 0.242,
+      "step": 141645
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7151858632324935e-09,
+      "loss": 0.2579,
+      "step": 141650
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6819791632288751e-09,
+      "loss": 0.2323,
+      "step": 141655
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6490970499133395e-09,
+      "loss": 0.2269,
+      "step": 141660
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6165395237077718e-09,
+      "loss": 0.2234,
+      "step": 141665
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5843065850368321e-09,
+      "loss": 0.2495,
+      "step": 141670
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5523982343196297e-09,
+      "loss": 0.2542,
+      "step": 141675
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5208144719697227e-09,
+      "loss": 0.2523,
+      "step": 141680
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.4895552983978934e-09,
+      "loss": 0.2379,
+      "step": 141685
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4586207140093733e-09,
+      "loss": 0.2469,
+      "step": 141690
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4280107192066184e-09,
+      "loss": 0.2736,
+      "step": 141695
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3977253143837577e-09,
+      "loss": 0.2477,
+      "step": 141700
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3677644999404716e-09,
+      "loss": 0.2599,
+      "step": 141705
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3381282762597868e-09,
+      "loss": 0.2482,
+      "step": 141710
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3088166437275062e-09,
+      "loss": 0.2591,
+      "step": 141715
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2798296027294321e-09,
+      "loss": 0.264,
+      "step": 141720
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2511671536347136e-09,
+      "loss": 0.2271,
+      "step": 141725
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.2228292968208266e-09,
+      "loss": 0.2488,
+      "step": 141730
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1948160326513692e-09,
+      "loss": 0.2753,
+      "step": 141735
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1671273614927147e-09,
+      "loss": 0.2398,
+      "step": 141740
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.139763283705686e-09,
+      "loss": 0.246,
+      "step": 141745
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1127237996427786e-09,
+      "loss": 0.257,
+      "step": 141750
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0860089096564884e-09,
+      "loss": 0.2507,
+      "step": 141755
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0596186140937602e-09,
+      "loss": 0.2598,
+      "step": 141760
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0335529132987631e-09,
+      "loss": 0.2502,
+      "step": 141765
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.007811807604564e-09,
+      "loss": 0.2493,
+      "step": 141770
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.823952973525563e-10,
+      "loss": 0.2634,
+      "step": 141775
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.573033828674805e-10,
+      "loss": 0.2628,
+      "step": 141780
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.325360644768522e-10,
+      "loss": 0.247,
+      "step": 141785
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.080933425026361e-10,
+      "loss": 0.2553,
+      "step": 141790
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.839752172612459e-10,
+      "loss": 0.2516,
+      "step": 141795
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.601816890663195e-10,
+      "loss": 0.2546,
+      "step": 141800
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.367127582259438e-10,
+      "loss": 0.2592,
+      "step": 141805
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.135684250482056e-10,
+      "loss": 0.2581,
+      "step": 141810
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.907486898300898e-10,
+      "loss": 0.251,
+      "step": 141815
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.682535528685808e-10,
+      "loss": 0.2412,
+      "step": 141820
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.460830144551123e-10,
+      "loss": 0.2584,
+      "step": 141825
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.242370748811178e-10,
+      "loss": 0.2558,
+      "step": 141830
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.027157344241531e-10,
+      "loss": 0.2496,
+      "step": 141835
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.815189933701005e-10,
+      "loss": 0.267,
+      "step": 141840
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.606468519909648e-10,
+      "loss": 0.2709,
+      "step": 141845
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.400993105587505e-10,
+      "loss": 0.2453,
+      "step": 141850
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.198763693399112e-10,
+      "loss": 0.2638,
+      "step": 141855
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.999780285953493e-10,
+      "loss": 0.2354,
+      "step": 141860
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.804042885859673e-10,
+      "loss": 0.2624,
+      "step": 141865
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.611551495643408e-10,
+      "loss": 0.2541,
+      "step": 141870
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.4223061178027e-10,
+      "loss": 0.2372,
+      "step": 141875
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.236306754807796e-10,
+      "loss": 0.2381,
+      "step": 141880
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.053553409045675e-10,
+      "loss": 0.2495,
+      "step": 141885
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.874046082931072e-10,
+      "loss": 0.2311,
+      "step": 141890
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.6977847787677e-10,
+      "loss": 0.2347,
+      "step": 141895
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.524769498859271e-10,
+      "loss": 0.248,
+      "step": 141900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.3550002454539886e-10,
+      "loss": 0.2674,
+      "step": 141905
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.1884770207167855e-10,
+      "loss": 0.2447,
+      "step": 141910
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.025199826868109e-10,
+      "loss": 0.2451,
+      "step": 141915
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.865168665989627e-10,
+      "loss": 0.2551,
+      "step": 141920
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.708383540135252e-10,
+      "loss": 0.2753,
+      "step": 141925
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.5548444514144075e-10,
+      "loss": 0.2452,
+      "step": 141930
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.4045514017699845e-10,
+      "loss": 0.2607,
+      "step": 141935
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.2575043931448723e-10,
+      "loss": 0.2621,
+      "step": 141940
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.113703427481962e-10,
+      "loss": 0.2502,
+      "step": 141945
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.973148506640877e-10,
+      "loss": 0.2524,
+      "step": 141950
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.8358396324257295e-10,
+      "loss": 0.2224,
+      "step": 141955
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.7017768066406323e-10,
+      "loss": 0.2429,
+      "step": 141960
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5709600310341864e-10,
+      "loss": 0.2337,
+      "step": 141965
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.443389307271726e-10,
+      "loss": 0.2236,
+      "step": 141970
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.3190646370185865e-10,
+      "loss": 0.2481,
+      "step": 141975
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.197986021912346e-10,
+      "loss": 0.2354,
+      "step": 141980
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.0801534635073172e-10,
+      "loss": 0.2381,
+      "step": 141985
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.9655669633300567e-10,
+      "loss": 0.2499,
+      "step": 141990
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.8542265228793653e-10,
+      "loss": 0.2834,
+      "step": 141995
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.7461321435985333e-10,
+      "loss": 0.2573,
+      "step": 142000
+    },
+    {
+      "epoch": 5.0,
+      "eval_loss": 0.24838173389434814,
+      "eval_runtime": 10.5549,
+      "eval_samples_per_second": 9.474,
+      "eval_steps_per_second": 9.474,
+      "step": 142000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 142115,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 4.632594192202221e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}