diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6881 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9954250299854608,
+  "eval_steps": 500,
+  "global_step": 4900,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 8.71941089630127,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 1.4235,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 7.11819314956665,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.3017,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.49417781829834,
+      "learning_rate": 6.25e-06,
+      "loss": 1.1621,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 2.3250486850738525,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.0125,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.7382829189300537,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 0.9134,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.43791663646698,
+      "learning_rate": 1.25e-05,
+      "loss": 0.8154,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.2901860475540161,
+      "learning_rate": 1.4583333333333335e-05,
+      "loss": 0.7274,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7073578238487244,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.7086,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.6095296740531921,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.6406,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.5200157165527344,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.6042,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8864995241165161,
+      "learning_rate": 2.2916666666666667e-05,
+      "loss": 0.5778,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 1.4653278589248657,
+      "learning_rate": 2.5e-05,
+      "loss": 0.57,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.39008694887161255,
+      "learning_rate": 2.7083333333333332e-05,
+      "loss": 0.5503,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.46981045603752136,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.5499,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.5259650945663452,
+      "learning_rate": 3.125e-05,
+      "loss": 0.5309,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.36070212721824646,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.5259,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.3584323823451996,
+      "learning_rate": 3.541666666666667e-05,
+      "loss": 0.5045,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.4284314513206482,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.5085,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.30807143449783325,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 0.4763,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.30832841992378235,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4975,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.381212443113327,
+      "learning_rate": 4.375e-05,
+      "loss": 0.4909,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.3347671627998352,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.4605,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3108418881893158,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.486,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3511927127838135,
+      "learning_rate": 5e-05,
+      "loss": 0.4601,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.33370575308799744,
+      "learning_rate": 4.99999412412136e-05,
+      "loss": 0.4867,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.3290724754333496,
+      "learning_rate": 4.99997649651306e-05,
+      "loss": 0.4683,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 0.2863904535770416,
+      "learning_rate": 4.999947117257964e-05,
+      "loss": 0.4595,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3704560399055481,
+      "learning_rate": 4.9999059864941726e-05,
+      "loss": 0.4793,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.3636023700237274,
+      "learning_rate": 4.999853104415031e-05,
+      "loss": 0.4652,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.633187472820282,
+      "learning_rate": 4.999788471269122e-05,
+      "loss": 0.4616,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.39704492688179016,
+      "learning_rate": 4.9997120873602654e-05,
+      "loss": 0.4751,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.30856508016586304,
+      "learning_rate": 4.999623953047522e-05,
+      "loss": 0.4708,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3069395422935486,
+      "learning_rate": 4.999524068745182e-05,
+      "loss": 0.4595,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3214091956615448,
+      "learning_rate": 4.9994124349227736e-05,
+      "loss": 0.4393,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.3112456202507019,
+      "learning_rate": 4.999289052105054e-05,
+      "loss": 0.445,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.2936570942401886,
+      "learning_rate": 4.9991539208720096e-05,
+      "loss": 0.4714,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.34774595499038696,
+      "learning_rate": 4.999007041858851e-05,
+      "loss": 0.4528,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.29614701867103577,
+      "learning_rate": 4.9988484157560136e-05,
+      "loss": 0.4455,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.3168615698814392,
+      "learning_rate": 4.998678043309151e-05,
+      "loss": 0.4336,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.30092811584472656,
+      "learning_rate": 4.998495925319134e-05,
+      "loss": 0.4238,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.37993124127388,
+      "learning_rate": 4.998302062642045e-05,
+      "loss": 0.4376,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.3411352038383484,
+      "learning_rate": 4.998096456189175e-05,
+      "loss": 0.4545,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2533304989337921,
+      "learning_rate": 4.9978791069270184e-05,
+      "loss": 0.4513,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.393837034702301,
+      "learning_rate": 4.9976500158772706e-05,
+      "loss": 0.4467,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.2702462673187256,
+      "learning_rate": 4.9974091841168195e-05,
+      "loss": 0.4361,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.25119566917419434,
+      "learning_rate": 4.9971566127777436e-05,
+      "loss": 0.4543,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2709728181362152,
+      "learning_rate": 4.996892303047306e-05,
+      "loss": 0.4481,
+      "step": 235
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2583755850791931,
+      "learning_rate": 4.9966162561679486e-05,
+      "loss": 0.4211,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.26327213644981384,
+      "learning_rate": 4.9963284734372855e-05,
+      "loss": 0.4355,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.2728334665298462,
+      "learning_rate": 4.9960289562080974e-05,
+      "loss": 0.4407,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.24257560074329376,
+      "learning_rate": 4.995717705888327e-05,
+      "loss": 0.4245,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.25513607263565063,
+      "learning_rate": 4.995394723941068e-05,
+      "loss": 0.445,
+      "step": 260
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.25588515400886536,
+      "learning_rate": 4.9950600118845644e-05,
+      "loss": 0.4205,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.278248131275177,
+      "learning_rate": 4.994713571292196e-05,
+      "loss": 0.4283,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.27247855067253113,
+      "learning_rate": 4.994355403792479e-05,
+      "loss": 0.4311,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.2843756079673767,
+      "learning_rate": 4.993985511069051e-05,
+      "loss": 0.4408,
+      "step": 280
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2906840443611145,
+      "learning_rate": 4.993603894860669e-05,
+      "loss": 0.4419,
+      "step": 285
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.24140453338623047,
+      "learning_rate": 4.993210556961197e-05,
+      "loss": 0.406,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2896282374858856,
+      "learning_rate": 4.9928054992195985e-05,
+      "loss": 0.433,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.2979041635990143,
+      "learning_rate": 4.992388723539931e-05,
+      "loss": 0.4072,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.25824418663978577,
+      "learning_rate": 4.991960231881333e-05,
+      "loss": 0.4293,
+      "step": 305
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2912902534008026,
+      "learning_rate": 4.991520026258015e-05,
+      "loss": 0.4231,
+      "step": 310
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2505576014518738,
+      "learning_rate": 4.9910681087392554e-05,
+      "loss": 0.42,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.258882075548172,
+      "learning_rate": 4.9906044814493824e-05,
+      "loss": 0.436,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.3020581305027008,
+      "learning_rate": 4.9901291465677706e-05,
+      "loss": 0.4177,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.29218652844429016,
+      "learning_rate": 4.9896421063288286e-05,
+      "loss": 0.426,
+      "step": 330
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.3000968098640442,
+      "learning_rate": 4.989143363021987e-05,
+      "loss": 0.4318,
+      "step": 335
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2585335969924927,
+      "learning_rate": 4.98863291899169e-05,
+      "loss": 0.4115,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.25833752751350403,
+      "learning_rate": 4.988110776637383e-05,
+      "loss": 0.4362,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.2597838342189789,
+      "learning_rate": 4.987576938413504e-05,
+      "loss": 0.4282,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.28554433584213257,
+      "learning_rate": 4.9870314068294655e-05,
+      "loss": 0.4176,
+      "step": 355
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2506715953350067,
+      "learning_rate": 4.9864741844496506e-05,
+      "loss": 0.4033,
+      "step": 360
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.30882102251052856,
+      "learning_rate": 4.9859052738933966e-05,
+      "loss": 0.4227,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.30396586656570435,
+      "learning_rate": 4.985324677834983e-05,
+      "loss": 0.4254,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.25673285126686096,
+      "learning_rate": 4.9847323990036174e-05,
+      "loss": 0.4219,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.2796240448951721,
+      "learning_rate": 4.984128440183429e-05,
+      "loss": 0.4172,
+      "step": 380
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.24497225880622864,
+      "learning_rate": 4.983512804213447e-05,
+      "loss": 0.4256,
+      "step": 385
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.2815968990325928,
+      "learning_rate": 4.9828854939875945e-05,
+      "loss": 0.4341,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.23544123768806458,
+      "learning_rate": 4.9822465124546705e-05,
+      "loss": 0.4125,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.26519954204559326,
+      "learning_rate": 4.981595862618336e-05,
+      "loss": 0.4075,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.251598596572876,
+      "learning_rate": 4.9809335475371045e-05,
+      "loss": 0.4117,
+      "step": 405
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2405155748128891,
+      "learning_rate": 4.9802595703243205e-05,
+      "loss": 0.4181,
+      "step": 410
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.29237255454063416,
+      "learning_rate": 4.979573934148152e-05,
+      "loss": 0.4224,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.2751956582069397,
+      "learning_rate": 4.97887664223157e-05,
+      "loss": 0.4286,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.252384215593338,
+      "learning_rate": 4.978167697852338e-05,
+      "loss": 0.4113,
+      "step": 425
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.24101611971855164,
+      "learning_rate": 4.977447104342992e-05,
+      "loss": 0.4001,
+      "step": 430
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.23108260333538055,
+      "learning_rate": 4.976714865090827e-05,
+      "loss": 0.4327,
+      "step": 435
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.2622400224208832,
+      "learning_rate": 4.975970983537884e-05,
+      "loss": 0.4138,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.24792075157165527,
+      "learning_rate": 4.975215463180929e-05,
+      "loss": 0.4024,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.25087806582450867,
+      "learning_rate": 4.974448307571437e-05,
+      "loss": 0.4157,
+      "step": 450
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2531994879245758,
+      "learning_rate": 4.9736695203155805e-05,
+      "loss": 0.4069,
+      "step": 455
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.27899864315986633,
+      "learning_rate": 4.9728791050742064e-05,
+      "loss": 0.4073,
+      "step": 460
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.22666263580322266,
+      "learning_rate": 4.972077065562821e-05,
+      "loss": 0.4209,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2369895726442337,
+      "learning_rate": 4.971263405551576e-05,
+      "loss": 0.3888,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.2376381754875183,
+      "learning_rate": 4.9704381288652436e-05,
+      "loss": 0.4151,
+      "step": 475
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.36553555727005005,
+      "learning_rate": 4.9696012393832034e-05,
+      "loss": 0.4055,
+      "step": 480
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3605659008026123,
+      "learning_rate": 4.9687527410394275e-05,
+      "loss": 0.4153,
+      "step": 485
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.3582921028137207,
+      "learning_rate": 4.967892637822451e-05,
+      "loss": 0.4135,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.25980615615844727,
+      "learning_rate": 4.967020933775366e-05,
+      "loss": 0.4147,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.23904970288276672,
+      "learning_rate": 4.966137632995793e-05,
+      "loss": 0.4033,
+      "step": 500
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.27073633670806885,
+      "learning_rate": 4.965242739635867e-05,
+      "loss": 0.4192,
+      "step": 505
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.22205090522766113,
+      "learning_rate": 4.964336257902217e-05,
+      "loss": 0.4135,
+      "step": 510
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.24786050617694855,
+      "learning_rate": 4.963418192055942e-05,
+      "loss": 0.4028,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.23719647526741028,
+      "learning_rate": 4.962488546412598e-05,
+      "loss": 0.4158,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.2376853972673416,
+      "learning_rate": 4.9615473253421727e-05,
+      "loss": 0.4019,
+      "step": 525
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.25200483202934265,
+      "learning_rate": 4.960594533269067e-05,
+      "loss": 0.4186,
+      "step": 530
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.24270349740982056,
+      "learning_rate": 4.9596301746720734e-05,
+      "loss": 0.4079,
+      "step": 535
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.2608851492404938,
+      "learning_rate": 4.958654254084355e-05,
+      "loss": 0.3855,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.24482597410678864,
+      "learning_rate": 4.9576667760934256e-05,
+      "loss": 0.3935,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.23347611725330353,
+      "learning_rate": 4.956667745341124e-05,
+      "loss": 0.4106,
+      "step": 550
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2340419739484787,
+      "learning_rate": 4.955657166523597e-05,
+      "loss": 0.3967,
+      "step": 555
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2722868323326111,
+      "learning_rate": 4.9546350443912783e-05,
+      "loss": 0.3947,
+      "step": 560
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2323121577501297,
+      "learning_rate": 4.953601383748857e-05,
+      "loss": 0.4156,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.24989305436611176,
+      "learning_rate": 4.952556189455266e-05,
+      "loss": 0.4046,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.2533356547355652,
+      "learning_rate": 4.9514994664236535e-05,
+      "loss": 0.3925,
+      "step": 575
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24604739248752594,
+      "learning_rate": 4.9504312196213596e-05,
+      "loss": 0.3964,
+      "step": 580
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24576042592525482,
+      "learning_rate": 4.9493514540698963e-05,
+      "loss": 0.4068,
+      "step": 585
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.24704518914222717,
+      "learning_rate": 4.9482601748449204e-05,
+      "loss": 0.3944,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.22531753778457642,
+      "learning_rate": 4.94715738707621e-05,
+      "loss": 0.4064,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.23824095726013184,
+      "learning_rate": 4.9460430959476456e-05,
+      "loss": 0.4014,
+      "step": 600
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.2367417812347412,
+      "learning_rate": 4.944917306697175e-05,
+      "loss": 0.4005,
+      "step": 605
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.24747785925865173,
+      "learning_rate": 4.9437800246168025e-05,
+      "loss": 0.4303,
+      "step": 610
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.27296656370162964,
+      "learning_rate": 4.942631255052551e-05,
+      "loss": 0.4111,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.23974792659282684,
+      "learning_rate": 4.9414710034044464e-05,
+      "loss": 0.4011,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.24948929250240326,
+      "learning_rate": 4.940299275126486e-05,
+      "loss": 0.3777,
+      "step": 625
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.265428364276886,
+      "learning_rate": 4.9391160757266164e-05,
+      "loss": 0.3868,
+      "step": 630
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.27857980132102966,
+      "learning_rate": 4.937921410766707e-05,
+      "loss": 0.3795,
+      "step": 635
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.2549702823162079,
+      "learning_rate": 4.936715285862523e-05,
+      "loss": 0.4148,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.24650795757770538,
+      "learning_rate": 4.9354977066836986e-05,
+      "loss": 0.4051,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.24199923872947693,
+      "learning_rate": 4.934268678953711e-05,
+      "loss": 0.3873,
+      "step": 650
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.23479871451854706,
+      "learning_rate": 4.933028208449856e-05,
+      "loss": 0.4089,
+      "step": 655
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.24073335528373718,
+      "learning_rate": 4.931776301003217e-05,
+      "loss": 0.3883,
+      "step": 660
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2442849725484848,
+      "learning_rate": 4.930512962498638e-05,
+      "loss": 0.394,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.2270922213792801,
+      "learning_rate": 4.929238198874698e-05,
+      "loss": 0.3912,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.22913384437561035,
+      "learning_rate": 4.9279520161236824e-05,
+      "loss": 0.3949,
+      "step": 675
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.22802837193012238,
+      "learning_rate": 4.9266544202915545e-05,
+      "loss": 0.3972,
+      "step": 680
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.24811463057994843,
+      "learning_rate": 4.9253454174779265e-05,
+      "loss": 0.388,
+      "step": 685
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.22244548797607422,
+      "learning_rate": 4.924025013836031e-05,
+      "loss": 0.4024,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.21797122061252594,
+      "learning_rate": 4.922693215572695e-05,
+      "loss": 0.3935,
+      "step": 695
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.23180623352527618,
+      "learning_rate": 4.921350028948305e-05,
+      "loss": 0.3884,
+      "step": 700
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.2671414315700531,
+      "learning_rate": 4.9199954602767826e-05,
+      "loss": 0.3827,
+      "step": 705
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.25717946887016296,
+      "learning_rate": 4.9186295159255537e-05,
+      "loss": 0.3964,
+      "step": 710
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.24643802642822266,
+      "learning_rate": 4.9172522023155154e-05,
+      "loss": 0.3777,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.24494287371635437,
+      "learning_rate": 4.91586352592101e-05,
+      "loss": 0.3675,
+      "step": 720
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.24090999364852905,
+      "learning_rate": 4.914463493269794e-05,
+      "loss": 0.3843,
+      "step": 725
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2768496572971344,
+      "learning_rate": 4.913052110943004e-05,
+      "loss": 0.3875,
+      "step": 730
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.24465887248516083,
+      "learning_rate": 4.9116293855751285e-05,
+      "loss": 0.3946,
+      "step": 735
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.24821196496486664,
+      "learning_rate": 4.9101953238539775e-05,
+      "loss": 0.3876,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.2489999234676361,
+      "learning_rate": 4.9087499325206494e-05,
+      "loss": 0.3751,
+      "step": 745
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.22408539056777954,
+      "learning_rate": 4.907293218369499e-05,
+      "loss": 0.3877,
+      "step": 750
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.26210272312164307,
+      "learning_rate": 4.905825188248106e-05,
+      "loss": 0.3775,
+      "step": 755
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.23799563944339752,
+      "learning_rate": 4.9043458490572444e-05,
+      "loss": 0.3757,
+      "step": 760
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.24405913054943085,
+      "learning_rate": 4.90285520775085e-05,
+      "loss": 0.3951,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 8.514335632324219,
+      "learning_rate": 4.901353271335982e-05,
+      "loss": 0.3722,
+      "step": 770
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3607272803783417,
+      "learning_rate": 4.899840046872799e-05,
+      "loss": 0.3785,
+      "step": 775
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.27763640880584717,
+      "learning_rate": 4.898315541474519e-05,
+      "loss": 0.4015,
+      "step": 780
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.23536086082458496,
+      "learning_rate": 4.896779762307389e-05,
+      "loss": 0.3634,
+      "step": 785
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.2394334375858307,
+      "learning_rate": 4.895232716590651e-05,
+      "loss": 0.3848,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.33459916710853577,
+      "learning_rate": 4.893674411596507e-05,
+      "loss": 0.3895,
+      "step": 795
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2422620952129364,
+      "learning_rate": 4.892104854650085e-05,
+      "loss": 0.3766,
+      "step": 800
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2283576875925064,
+      "learning_rate": 4.890524053129408e-05,
+      "loss": 0.3765,
+      "step": 805
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2283744066953659,
+      "learning_rate": 4.888932014465352e-05,
+      "loss": 0.3696,
+      "step": 810
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.2649267613887787,
+      "learning_rate": 4.887328746141619e-05,
+      "loss": 0.3955,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.242611825466156,
+      "learning_rate": 4.885714255694698e-05,
+      "loss": 0.3949,
+      "step": 820
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.22519369423389435,
+      "learning_rate": 4.884088550713827e-05,
+      "loss": 0.378,
+      "step": 825
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2294245809316635,
+      "learning_rate": 4.8824516388409625e-05,
+      "loss": 0.371,
+      "step": 830
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.24208737909793854,
+      "learning_rate": 4.880803527770742e-05,
+      "loss": 0.3807,
+      "step": 835
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.23562513291835785,
+      "learning_rate": 4.879144225250445e-05,
+      "loss": 0.3932,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.2297345995903015,
+      "learning_rate": 4.877473739079961e-05,
+      "loss": 0.3743,
+      "step": 845
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.22091899812221527,
+      "learning_rate": 4.8757920771117475e-05,
+      "loss": 0.377,
+      "step": 850
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.24047815799713135,
+      "learning_rate": 4.874099247250798e-05,
+      "loss": 0.3635,
+      "step": 855
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.22934621572494507,
+      "learning_rate": 4.872395257454605e-05,
+      "loss": 0.3705,
+      "step": 860
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.2382223904132843,
+      "learning_rate": 4.8706801157331154e-05,
+      "loss": 0.3821,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.23687830567359924,
+      "learning_rate": 4.8689538301487025e-05,
+      "loss": 0.3889,
+      "step": 870
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.4864930510520935,
+      "learning_rate": 4.867216408816122e-05,
+      "loss": 0.3816,
+      "step": 875
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5121108293533325,
+      "learning_rate": 4.865467859902475e-05,
+      "loss": 0.3772,
+      "step": 880
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.33512434363365173,
+      "learning_rate": 4.86370819162717e-05,
+      "loss": 0.3711,
+      "step": 885
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.26488932967185974,
+      "learning_rate": 4.8619374122618854e-05,
+      "loss": 0.3742,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.3071129322052002,
+      "learning_rate": 4.860155530130529e-05,
+      "loss": 0.3822,
+      "step": 895
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2589118480682373,
+      "learning_rate": 4.858362553609199e-05,
+      "loss": 0.3937,
+      "step": 900
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.2737272381782532,
+      "learning_rate": 4.856558491126146e-05,
+      "loss": 0.3947,
+      "step": 905
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40217074751853943,
+      "learning_rate": 4.8547433511617305e-05,
+      "loss": 0.396,
+      "step": 910
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.24473513662815094,
+      "learning_rate": 4.8529171422483876e-05,
+      "loss": 0.3743,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.24511292576789856,
+      "learning_rate": 4.851079872970582e-05,
+      "loss": 0.3782,
+      "step": 920
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2283199429512024,
+      "learning_rate": 4.849231551964771e-05,
+      "loss": 0.3693,
+      "step": 925
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.21812038123607635,
+      "learning_rate": 4.847372187919362e-05,
+      "loss": 0.3802,
+      "step": 930
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2254110723733902,
+      "learning_rate": 4.845501789574674e-05,
+      "loss": 0.3855,
+      "step": 935
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.22982986271381378,
+      "learning_rate": 4.843620365722894e-05,
+      "loss": 0.3663,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.24178490042686462,
+      "learning_rate": 4.8417279252080363e-05,
+      "loss": 0.3786,
+      "step": 945
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2579748034477234,
+      "learning_rate": 4.8398244769259e-05,
+      "loss": 0.3637,
+      "step": 950
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.2304283082485199,
+      "learning_rate": 4.837910029824031e-05,
+      "loss": 0.3839,
+      "step": 955
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.23236927390098572,
+      "learning_rate": 4.835984592901678e-05,
+      "loss": 0.3881,
+      "step": 960
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.23549768328666687,
+      "learning_rate": 4.834048175209746e-05,
+      "loss": 0.3782,
+      "step": 965
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.21959024667739868,
+      "learning_rate": 4.8321007858507594e-05,
+      "loss": 0.3781,
+      "step": 970
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.23583754897117615,
+      "learning_rate": 4.830142433978818e-05,
+      "loss": 0.3602,
+      "step": 975
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.24004381895065308,
+      "learning_rate": 4.828173128799551e-05,
+      "loss": 0.3735,
+      "step": 980
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.21957090497016907,
+      "learning_rate": 4.826192879570078e-05,
+      "loss": 0.376,
+      "step": 985
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.25009414553642273,
+      "learning_rate": 4.824201695598963e-05,
+      "loss": 0.3729,
+      "step": 990
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.26149284839630127,
+      "learning_rate": 4.822199586246168e-05,
+      "loss": 0.3863,
+      "step": 995
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.22410422563552856,
+      "learning_rate": 4.8201865609230156e-05,
+      "loss": 0.3717,
+      "step": 1000
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2567145824432373,
+      "learning_rate": 4.81816262909214e-05,
+      "loss": 0.375,
+      "step": 1005
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.22770246863365173,
+      "learning_rate": 4.8161278002674424e-05,
+      "loss": 0.3607,
+      "step": 1010
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.2278997153043747,
+      "learning_rate": 4.81408208401405e-05,
+      "loss": 0.3631,
+      "step": 1015
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.24125157296657562,
+      "learning_rate": 4.8120254899482665e-05,
+      "loss": 0.3579,
+      "step": 1020
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.2390599250793457,
+      "learning_rate": 4.809958027737529e-05,
+      "loss": 0.3605,
+      "step": 1025
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.25808480381965637,
+      "learning_rate": 4.8078797071003644e-05,
+      "loss": 0.3783,
+      "step": 1030
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.504199206829071,
+      "learning_rate": 4.80579053780634e-05,
+      "loss": 0.3515,
+      "step": 1035
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.235836461186409,
+      "learning_rate": 4.803690529676019e-05,
+      "loss": 0.3747,
+      "step": 1040
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.24636542797088623,
+      "learning_rate": 4.801579692580918e-05,
+      "loss": 0.3935,
+      "step": 1045
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.2520650625228882,
+      "learning_rate": 4.7994580364434524e-05,
+      "loss": 0.355,
+      "step": 1050
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.22997605800628662,
+      "learning_rate": 4.7973255712368994e-05,
+      "loss": 0.3708,
+      "step": 1055
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.24863748252391815,
+      "learning_rate": 4.795182306985343e-05,
+      "loss": 0.3637,
+      "step": 1060
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.25943225622177124,
+      "learning_rate": 4.793028253763633e-05,
+      "loss": 0.3798,
+      "step": 1065
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2547088861465454,
+      "learning_rate": 4.790863421697333e-05,
+      "loss": 0.3727,
+      "step": 1070
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.25044122338294983,
+      "learning_rate": 4.788687820962675e-05,
+      "loss": 0.3609,
+      "step": 1075
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.24127812683582306,
+      "learning_rate": 4.786501461786512e-05,
+      "loss": 0.3649,
+      "step": 1080
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.2445056438446045,
+      "learning_rate": 4.7843043544462696e-05,
+      "loss": 0.375,
+      "step": 1085
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.23667505383491516,
+      "learning_rate": 4.7820965092698967e-05,
+      "loss": 0.3776,
+      "step": 1090
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.23887944221496582,
+      "learning_rate": 4.779877936635816e-05,
+      "loss": 0.3724,
+      "step": 1095
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.2539786100387573,
+      "learning_rate": 4.7776486469728796e-05,
+      "loss": 0.362,
+      "step": 1100
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.23806200921535492,
+      "learning_rate": 4.7754086507603155e-05,
+      "loss": 0.3466,
+      "step": 1105
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.24354040622711182,
+      "learning_rate": 4.773157958527681e-05,
+      "loss": 0.3641,
+      "step": 1110
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.23725096881389618,
+      "learning_rate": 4.770896580854811e-05,
+      "loss": 0.3667,
+      "step": 1115
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.2345529943704605,
+      "learning_rate": 4.768624528371771e-05,
+      "loss": 0.3678,
+      "step": 1120
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.24982014298439026,
+      "learning_rate": 4.766341811758803e-05,
+      "loss": 0.3765,
+      "step": 1125
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23971329629421234,
+      "learning_rate": 4.764048441746282e-05,
+      "loss": 0.3823,
+      "step": 1130
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.23192477226257324,
+      "learning_rate": 4.7617444291146565e-05,
+      "loss": 0.3551,
+      "step": 1135
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.22103410959243774,
+      "learning_rate": 4.759429784694407e-05,
+      "loss": 0.369,
+      "step": 1140
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2329845428466797,
+      "learning_rate": 4.757104519365989e-05,
+      "loss": 0.3745,
+      "step": 1145
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.23781472444534302,
+      "learning_rate": 4.754768644059784e-05,
+      "loss": 0.3653,
+      "step": 1150
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.2504659593105316,
+      "learning_rate": 4.752422169756048e-05,
+      "loss": 0.3721,
+      "step": 1155
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.22282807528972626,
+      "learning_rate": 4.750065107484859e-05,
+      "loss": 0.3555,
+      "step": 1160
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.23634228110313416,
+      "learning_rate": 4.7476974683260674e-05,
+      "loss": 0.3664,
+      "step": 1165
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2371123880147934,
+      "learning_rate": 4.74531926340924e-05,
+      "loss": 0.3631,
+      "step": 1170
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.23620998859405518,
+      "learning_rate": 4.742930503913614e-05,
+      "loss": 0.379,
+      "step": 1175
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.25195619463920593,
+      "learning_rate": 4.740531201068036e-05,
+      "loss": 0.3675,
+      "step": 1180
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2376963049173355,
+      "learning_rate": 4.738121366150916e-05,
+      "loss": 0.3756,
+      "step": 1185
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.2429811954498291,
+      "learning_rate": 4.7357010104901725e-05,
+      "loss": 0.3697,
+      "step": 1190
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2390989065170288,
+      "learning_rate": 4.7332701454631786e-05,
+      "loss": 0.3692,
+      "step": 1195
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.233895406126976,
+      "learning_rate": 4.730828782496709e-05,
+      "loss": 0.3445,
+      "step": 1200
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.24079565703868866,
+      "learning_rate": 4.728376933066885e-05,
+      "loss": 0.3572,
+      "step": 1205
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2393987774848938,
+      "learning_rate": 4.725914608699122e-05,
+      "loss": 0.3581,
+      "step": 1210
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.2563488483428955,
+      "learning_rate": 4.723441820968076e-05,
+      "loss": 0.3752,
+      "step": 1215
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.23606589436531067,
+      "learning_rate": 4.7209585814975876e-05,
+      "loss": 0.3571,
+      "step": 1220
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2621292769908905,
+      "learning_rate": 4.7184649019606275e-05,
+      "loss": 0.3924,
+      "step": 1225
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.23915182054042816,
+      "learning_rate": 4.715960794079243e-05,
+      "loss": 0.3764,
+      "step": 1230
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.6031570434570312,
+      "learning_rate": 4.7134462696245e-05,
+      "loss": 0.3756,
+      "step": 1235
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.2603359520435333,
+      "learning_rate": 4.710921340416431e-05,
+      "loss": 0.3516,
+      "step": 1240
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23302561044692993,
+      "learning_rate": 4.708386018323979e-05,
+      "loss": 0.3667,
+      "step": 1245
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23199358582496643,
+      "learning_rate": 4.70584031526494e-05,
+      "loss": 0.3738,
+      "step": 1250
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.23459462821483612,
+      "learning_rate": 4.7032842432059075e-05,
+      "loss": 0.3601,
+      "step": 1255
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.22799892723560333,
+      "learning_rate": 4.700717814162218e-05,
+      "loss": 0.3578,
+      "step": 1260
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2482994943857193,
+      "learning_rate": 4.698141040197889e-05,
+      "loss": 0.3745,
+      "step": 1265
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.2286883145570755,
+      "learning_rate": 4.6955539334255716e-05,
+      "loss": 0.3661,
+      "step": 1270
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.22291447222232819,
+      "learning_rate": 4.6929565060064864e-05,
+      "loss": 0.3555,
+      "step": 1275
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.22780221700668335,
+      "learning_rate": 4.690348770150366e-05,
+      "loss": 0.3669,
+      "step": 1280
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.23842570185661316,
+      "learning_rate": 4.6877307381154044e-05,
+      "loss": 0.3675,
+      "step": 1285
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2277093231678009,
+      "learning_rate": 4.685102422208191e-05,
+      "loss": 0.366,
+      "step": 1290
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2254444658756256,
+      "learning_rate": 4.682463834783658e-05,
+      "loss": 0.3477,
+      "step": 1295
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2116675227880478,
+      "learning_rate": 4.6798149882450216e-05,
+      "loss": 0.3586,
+      "step": 1300
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.24347162246704102,
+      "learning_rate": 4.677155895043722e-05,
+      "loss": 0.3509,
+      "step": 1305
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.2656736373901367,
+      "learning_rate": 4.674486567679367e-05,
+      "loss": 0.3679,
+      "step": 1310
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.23632311820983887,
+      "learning_rate": 4.6718070186996706e-05,
+      "loss": 0.3548,
+      "step": 1315
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2460298240184784,
+      "learning_rate": 4.669117260700397e-05,
+      "loss": 0.3504,
+      "step": 1320
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.28527209162712097,
+      "learning_rate": 4.666417306325299e-05,
+      "loss": 0.3699,
+      "step": 1325
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2586132287979126,
+      "learning_rate": 4.663707168266061e-05,
+      "loss": 0.3645,
+      "step": 1330
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.2613135278224945,
+      "learning_rate": 4.660986859262236e-05,
+      "loss": 0.3701,
+      "step": 1335
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.27791205048561096,
+      "learning_rate": 4.658256392101189e-05,
+      "loss": 0.3597,
+      "step": 1340
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.27237483859062195,
+      "learning_rate": 4.6555157796180335e-05,
+      "loss": 0.3688,
+      "step": 1345
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.26933595538139343,
+      "learning_rate": 4.652765034695576e-05,
+      "loss": 0.3596,
+      "step": 1350
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.2452528327703476,
+      "learning_rate": 4.6500041702642525e-05,
+      "loss": 0.3641,
+      "step": 1355
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.24474672973155975,
+      "learning_rate": 4.647233199302064e-05,
+      "loss": 0.3664,
+      "step": 1360
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.25770512223243713,
+      "learning_rate": 4.6444521348345216e-05,
+      "loss": 0.3692,
+      "step": 1365
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.26786795258522034,
+      "learning_rate": 4.641660989934585e-05,
+      "loss": 0.3665,
+      "step": 1370
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.23145703971385956,
+      "learning_rate": 4.6388597777225964e-05,
+      "loss": 0.3543,
+      "step": 1375
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2546435296535492,
+      "learning_rate": 4.6360485113662216e-05,
+      "loss": 0.3654,
+      "step": 1380
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.22902755439281464,
+      "learning_rate": 4.6332272040803895e-05,
+      "loss": 0.3609,
+      "step": 1385
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.3179545998573303,
+      "learning_rate": 4.630395869127226e-05,
+      "loss": 0.359,
+      "step": 1390
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.25595393776893616,
+      "learning_rate": 4.627554519815999e-05,
+      "loss": 0.3557,
+      "step": 1395
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.25582295656204224,
+      "learning_rate": 4.624703169503043e-05,
+      "loss": 0.3605,
+      "step": 1400
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.26308387517929077,
+      "learning_rate": 4.621841831591711e-05,
+      "loss": 0.3564,
+      "step": 1405
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.2370084524154663,
+      "learning_rate": 4.618970519532303e-05,
+      "loss": 0.3502,
+      "step": 1410
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.24112296104431152,
+      "learning_rate": 4.616089246822003e-05,
+      "loss": 0.3728,
+      "step": 1415
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.2341877967119217,
+      "learning_rate": 4.613198027004818e-05,
+      "loss": 0.3597,
+      "step": 1420
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.248354971408844,
+      "learning_rate": 4.6102968736715134e-05,
+      "loss": 0.3429,
+      "step": 1425
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.23436321318149567,
+      "learning_rate": 4.60738580045955e-05,
+      "loss": 0.3607,
+      "step": 1430
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.24360601603984833,
+      "learning_rate": 4.6044648210530175e-05,
+      "loss": 0.3635,
+      "step": 1435
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2809852361679077,
+      "learning_rate": 4.6015339491825716e-05,
+      "loss": 0.3723,
+      "step": 1440
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.2518574297428131,
+      "learning_rate": 4.5985931986253716e-05,
+      "loss": 0.3672,
+      "step": 1445
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.25123685598373413,
+      "learning_rate": 4.595642583205011e-05,
+      "loss": 0.3789,
+      "step": 1450
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.24918679893016815,
+      "learning_rate": 4.592682116791457e-05,
+      "loss": 0.3592,
+      "step": 1455
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.25612717866897583,
+      "learning_rate": 4.589711813300983e-05,
+      "loss": 0.3583,
+      "step": 1460
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.21745339035987854,
+      "learning_rate": 4.586731686696102e-05,
+      "loss": 0.3446,
+      "step": 1465
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.24862965941429138,
+      "learning_rate": 4.583741750985505e-05,
+      "loss": 0.3477,
+      "step": 1470
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2478131204843521,
+      "learning_rate": 4.580742020223992e-05,
+      "loss": 0.3638,
+      "step": 1475
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.22280053794384003,
+      "learning_rate": 4.577732508512404e-05,
+      "loss": 0.3656,
+      "step": 1480
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.2296367883682251,
+      "learning_rate": 4.574713229997563e-05,
+      "loss": 0.3469,
+      "step": 1485
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2436257153749466,
+      "learning_rate": 4.5716841988722004e-05,
+      "loss": 0.3439,
+      "step": 1490
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.23801317811012268,
+      "learning_rate": 4.568645429374891e-05,
+      "loss": 0.3647,
+      "step": 1495
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.23976649343967438,
+      "learning_rate": 4.5655969357899874e-05,
+      "loss": 0.3624,
+      "step": 1500
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.24574269354343414,
+      "learning_rate": 4.562538732447553e-05,
+      "loss": 0.3504,
+      "step": 1505
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.2505897283554077,
+      "learning_rate": 4.559470833723292e-05,
+      "loss": 0.352,
+      "step": 1510
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.2543080151081085,
+      "learning_rate": 4.556393254038486e-05,
+      "loss": 0.3659,
+      "step": 1515
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.25658947229385376,
+      "learning_rate": 4.5533060078599226e-05,
+      "loss": 0.3628,
+      "step": 1520
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.25442373752593994,
+      "learning_rate": 4.550209109699829e-05,
+      "loss": 0.3475,
+      "step": 1525
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.27169328927993774,
+      "learning_rate": 4.5471025741158024e-05,
+      "loss": 0.3404,
+      "step": 1530
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.22989720106124878,
+      "learning_rate": 4.5439864157107434e-05,
+      "loss": 0.3621,
+      "step": 1535
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.24393929541110992,
+      "learning_rate": 4.540860649132789e-05,
+      "loss": 0.3524,
+      "step": 1540
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2499324530363083,
+      "learning_rate": 4.537725289075239e-05,
+      "loss": 0.3561,
+      "step": 1545
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2389133721590042,
+      "learning_rate": 4.534580350276487e-05,
+      "loss": 0.3591,
+      "step": 1550
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.2484143227338791,
+      "learning_rate": 4.531425847519958e-05,
+      "loss": 0.3563,
+      "step": 1555
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.22839954495429993,
+      "learning_rate": 4.528261795634032e-05,
+      "loss": 0.3338,
+      "step": 1560
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.24448874592781067,
+      "learning_rate": 4.525088209491977e-05,
+      "loss": 0.3677,
+      "step": 1565
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2378901243209839,
+      "learning_rate": 4.5219051040118775e-05,
+      "loss": 0.3487,
+      "step": 1570
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.25053712725639343,
+      "learning_rate": 4.5187124941565685e-05,
+      "loss": 0.3555,
+      "step": 1575
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.28136828541755676,
+      "learning_rate": 4.51551039493356e-05,
+      "loss": 0.3635,
+      "step": 1580
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.2337874472141266,
+      "learning_rate": 4.5122988213949676e-05,
+      "loss": 0.3529,
+      "step": 1585
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.23640981316566467,
+      "learning_rate": 4.509077788637446e-05,
+      "loss": 0.3491,
+      "step": 1590
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.25626295804977417,
+      "learning_rate": 4.505847311802112e-05,
+      "loss": 0.3529,
+      "step": 1595
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.24295687675476074,
+      "learning_rate": 4.502607406074479e-05,
+      "loss": 0.3381,
+      "step": 1600
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.23309361934661865,
+      "learning_rate": 4.499358086684381e-05,
+      "loss": 0.3541,
+      "step": 1605
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.23449920117855072,
+      "learning_rate": 4.496099368905902e-05,
+      "loss": 0.3464,
+      "step": 1610
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2483125478029251,
+      "learning_rate": 4.4928312680573064e-05,
+      "loss": 0.3624,
+      "step": 1615
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24859212338924408,
+      "learning_rate": 4.489553799500966e-05,
+      "loss": 0.357,
+      "step": 1620
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.24081198871135712,
+      "learning_rate": 4.4862669786432865e-05,
+      "loss": 0.376,
+      "step": 1625
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.2605394124984741,
+      "learning_rate": 4.4829708209346365e-05,
+      "loss": 0.3483,
+      "step": 1630
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.24203234910964966,
+      "learning_rate": 4.4796653418692745e-05,
+      "loss": 0.3503,
+      "step": 1635
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2464054822921753,
+      "learning_rate": 4.4763505569852745e-05,
+      "loss": 0.3665,
+      "step": 1640
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.23673394322395325,
+      "learning_rate": 4.473026481864456e-05,
+      "loss": 0.3483,
+      "step": 1645
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.22426186501979828,
+      "learning_rate": 4.469693132132308e-05,
+      "loss": 0.3551,
+      "step": 1650
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.2271643429994583,
+      "learning_rate": 4.4663505234579185e-05,
+      "loss": 0.3601,
+      "step": 1655
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2544379234313965,
+      "learning_rate": 4.4629986715538976e-05,
+      "loss": 0.3543,
+      "step": 1660
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.24445408582687378,
+      "learning_rate": 4.459637592176304e-05,
+      "loss": 0.3497,
+      "step": 1665
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.25206229090690613,
+      "learning_rate": 4.456267301124575e-05,
+      "loss": 0.3513,
+      "step": 1670
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.23434026539325714,
+      "learning_rate": 4.452887814241446e-05,
+      "loss": 0.3425,
+      "step": 1675
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.48449090123176575,
+      "learning_rate": 4.449499147412882e-05,
+      "loss": 0.3361,
+      "step": 1680
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2515006959438324,
+      "learning_rate": 4.4461013165679986e-05,
+      "loss": 0.3347,
+      "step": 1685
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.24977871775627136,
+      "learning_rate": 4.44269433767899e-05,
+      "loss": 0.337,
+      "step": 1690
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.2558392584323883,
+      "learning_rate": 4.43927822676105e-05,
+      "loss": 0.334,
+      "step": 1695
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.26668402552604675,
+      "learning_rate": 4.435852999872302e-05,
+      "loss": 0.3528,
+      "step": 1700
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.22888560593128204,
+      "learning_rate": 4.432418673113721e-05,
+      "loss": 0.3307,
+      "step": 1705
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.25077155232429504,
+      "learning_rate": 4.428975262629055e-05,
+      "loss": 0.3563,
+      "step": 1710
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2320801019668579,
+      "learning_rate": 4.425522784604755e-05,
+      "loss": 0.3426,
+      "step": 1715
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.24620367586612701,
+      "learning_rate": 4.4220612552698947e-05,
+      "loss": 0.3556,
+      "step": 1720
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.2687610685825348,
+      "learning_rate": 4.418590690896094e-05,
+      "loss": 0.3551,
+      "step": 1725
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.24751456081867218,
+      "learning_rate": 4.415111107797445e-05,
+      "loss": 0.3445,
+      "step": 1730
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.2729167640209198,
+      "learning_rate": 4.411622522330435e-05,
+      "loss": 0.3577,
+      "step": 1735
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.23679660260677338,
+      "learning_rate": 4.4081249508938684e-05,
+      "loss": 0.3487,
+      "step": 1740
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24365286529064178,
+      "learning_rate": 4.404618409928788e-05,
+      "loss": 0.3611,
+      "step": 1745
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24482664465904236,
+      "learning_rate": 4.4011029159184015e-05,
+      "loss": 0.3424,
+      "step": 1750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.24546444416046143,
+      "learning_rate": 4.3975784853880005e-05,
+      "loss": 0.357,
+      "step": 1755
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.24618016183376312,
+      "learning_rate": 4.39404513490489e-05,
+      "loss": 0.3534,
+      "step": 1760
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.2552330493927002,
+      "learning_rate": 4.390502881078296e-05,
+      "loss": 0.3502,
+      "step": 1765
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.24205201864242554,
+      "learning_rate": 4.386951740559305e-05,
+      "loss": 0.3523,
+      "step": 1770
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.27865052223205566,
+      "learning_rate": 4.3833917300407716e-05,
+      "loss": 0.3466,
+      "step": 1775
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.234114408493042,
+      "learning_rate": 4.379822866257248e-05,
+      "loss": 0.3446,
+      "step": 1780
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2365516871213913,
+      "learning_rate": 4.376245165984903e-05,
+      "loss": 0.3455,
+      "step": 1785
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2584002614021301,
+      "learning_rate": 4.372658646041443e-05,
+      "loss": 0.327,
+      "step": 1790
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.24103760719299316,
+      "learning_rate": 4.3690633232860316e-05,
+      "loss": 0.3495,
+      "step": 1795
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.2564971446990967,
+      "learning_rate": 4.365459214619214e-05,
+      "loss": 0.3485,
+      "step": 1800
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2242477685213089,
+      "learning_rate": 4.3618463369828335e-05,
+      "loss": 0.341,
+      "step": 1805
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.23354335129261017,
+      "learning_rate": 4.358224707359955e-05,
+      "loss": 0.3483,
+      "step": 1810
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.22686982154846191,
+      "learning_rate": 4.354594342774784e-05,
+      "loss": 0.3464,
+      "step": 1815
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2452050894498825,
+      "learning_rate": 4.350955260292584e-05,
+      "loss": 0.3412,
+      "step": 1820
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.2559107542037964,
+      "learning_rate": 4.347307477019602e-05,
+      "loss": 0.3513,
+      "step": 1825
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.24746520817279816,
+      "learning_rate": 4.343651010102984e-05,
+      "loss": 0.3522,
+      "step": 1830
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.24204814434051514,
+      "learning_rate": 4.3399858767306936e-05,
+      "loss": 0.3574,
+      "step": 1835
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2356475591659546,
+      "learning_rate": 4.336312094131434e-05,
+      "loss": 0.3434,
+      "step": 1840
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2569087743759155,
+      "learning_rate": 4.332629679574566e-05,
+      "loss": 0.3369,
+      "step": 1845
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.25907132029533386,
+      "learning_rate": 4.3289386503700266e-05,
+      "loss": 0.3573,
+      "step": 1850
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.24386312067508698,
+      "learning_rate": 4.325239023868247e-05,
+      "loss": 0.3473,
+      "step": 1855
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2457239180803299,
+      "learning_rate": 4.321530817460073e-05,
+      "loss": 0.336,
+      "step": 1860
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2373226284980774,
+      "learning_rate": 4.317814048576682e-05,
+      "loss": 0.3546,
+      "step": 1865
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.2573254704475403,
+      "learning_rate": 4.3140887346894974e-05,
+      "loss": 0.3513,
+      "step": 1870
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.254350870847702,
+      "learning_rate": 4.3103548933101154e-05,
+      "loss": 0.3494,
+      "step": 1875
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.26855236291885376,
+      "learning_rate": 4.3066125419902156e-05,
+      "loss": 0.3538,
+      "step": 1880
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.22530697286128998,
+      "learning_rate": 4.302861698321478e-05,
+      "loss": 0.3433,
+      "step": 1885
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.2630355656147003,
+      "learning_rate": 4.299102379935505e-05,
+      "loss": 0.3395,
+      "step": 1890
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.24788448214530945,
+      "learning_rate": 4.2953346045037345e-05,
+      "loss": 0.353,
+      "step": 1895
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.24751460552215576,
+      "learning_rate": 4.291558389737362e-05,
+      "loss": 0.345,
+      "step": 1900
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.24289636313915253,
+      "learning_rate": 4.2877737533872485e-05,
+      "loss": 0.3327,
+      "step": 1905
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.25640085339546204,
+      "learning_rate": 4.283980713243847e-05,
+      "loss": 0.3563,
+      "step": 1910
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.24024629592895508,
+      "learning_rate": 4.2801792871371116e-05,
+      "loss": 0.358,
+      "step": 1915
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2351466864347458,
+      "learning_rate": 4.2763694929364166e-05,
+      "loss": 0.3384,
+      "step": 1920
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2508438229560852,
+      "learning_rate": 4.272551348550474e-05,
+      "loss": 0.3393,
+      "step": 1925
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.236288383603096,
+      "learning_rate": 4.268724871927245e-05,
+      "loss": 0.3532,
+      "step": 1930
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.25894463062286377,
+      "learning_rate": 4.264890081053859e-05,
+      "loss": 0.3486,
+      "step": 1935
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.24554409086704254,
+      "learning_rate": 4.261046993956531e-05,
+      "loss": 0.3315,
+      "step": 1940
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.2480149120092392,
+      "learning_rate": 4.25719562870047e-05,
+      "loss": 0.3385,
+      "step": 1945
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.24384349584579468,
+      "learning_rate": 4.253336003389799e-05,
+      "loss": 0.3369,
+      "step": 1950
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2618124485015869,
+      "learning_rate": 4.249468136167472e-05,
+      "loss": 0.3344,
+      "step": 1955
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.25003039836883545,
+      "learning_rate": 4.245592045215182e-05,
+      "loss": 0.3419,
+      "step": 1960
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.2513079345226288,
+      "learning_rate": 4.241707748753283e-05,
+      "loss": 0.3352,
+      "step": 1965
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.23028899729251862,
+      "learning_rate": 4.237815265040696e-05,
+      "loss": 0.3264,
+      "step": 1970
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.25453129410743713,
+      "learning_rate": 4.2339146123748324e-05,
+      "loss": 0.3413,
+      "step": 1975
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.27212798595428467,
+      "learning_rate": 4.2300058090915004e-05,
+      "loss": 0.3399,
+      "step": 1980
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.24868816137313843,
+      "learning_rate": 4.226088873564825e-05,
+      "loss": 0.3325,
+      "step": 1985
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.2333601415157318,
+      "learning_rate": 4.222163824207155e-05,
+      "loss": 0.3489,
+      "step": 1990
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.24276578426361084,
+      "learning_rate": 4.218230679468981e-05,
+      "loss": 0.3312,
+      "step": 1995
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.24993957579135895,
+      "learning_rate": 4.214289457838849e-05,
+      "loss": 0.3406,
+      "step": 2000
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.24900434911251068,
+      "learning_rate": 4.2103401778432694e-05,
+      "loss": 0.3377,
+      "step": 2005
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.24076253175735474,
+      "learning_rate": 4.206382858046636e-05,
+      "loss": 0.3473,
+      "step": 2010
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.2292889505624771,
+      "learning_rate": 4.202417517051132e-05,
+      "loss": 0.3395,
+      "step": 2015
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.24988548457622528,
+      "learning_rate": 4.1984441734966486e-05,
+      "loss": 0.3437,
+      "step": 2020
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.242272287607193,
+      "learning_rate": 4.1944628460606916e-05,
+      "loss": 0.3381,
+      "step": 2025
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.24164338409900665,
+      "learning_rate": 4.1904735534583006e-05,
+      "loss": 0.3438,
+      "step": 2030
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.24440313875675201,
+      "learning_rate": 4.1864763144419536e-05,
+      "loss": 0.3395,
+      "step": 2035
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.23709137737751007,
+      "learning_rate": 4.182471147801485e-05,
+      "loss": 0.3507,
+      "step": 2040
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.23476146161556244,
+      "learning_rate": 4.1784580723639923e-05,
+      "loss": 0.3175,
+      "step": 2045
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.24348315596580505,
+      "learning_rate": 4.1744371069937505e-05,
+      "loss": 0.3295,
+      "step": 2050
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.23205485939979553,
+      "learning_rate": 4.1704082705921246e-05,
+      "loss": 0.3328,
+      "step": 2055
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.2688595652580261,
+      "learning_rate": 4.166371582097476e-05,
+      "loss": 0.3233,
+      "step": 2060
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.25747400522232056,
+      "learning_rate": 4.162327060485078e-05,
+      "loss": 0.3368,
+      "step": 2065
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.242996484041214,
+      "learning_rate": 4.158274724767028e-05,
+      "loss": 0.3391,
+      "step": 2070
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.23208199441432953,
+      "learning_rate": 4.154214593992149e-05,
+      "loss": 0.3316,
+      "step": 2075
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2287871092557907,
+      "learning_rate": 4.1501466872459104e-05,
+      "loss": 0.3332,
+      "step": 2080
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.27117621898651123,
+      "learning_rate": 4.146071023650335e-05,
+      "loss": 0.3393,
+      "step": 2085
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2395385354757309,
+      "learning_rate": 4.141987622363904e-05,
+      "loss": 0.3367,
+      "step": 2090
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.2599872052669525,
+      "learning_rate": 4.1378965025814754e-05,
+      "loss": 0.3519,
+      "step": 2095
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.28649839758872986,
+      "learning_rate": 4.133797683534188e-05,
+      "loss": 0.3268,
+      "step": 2100
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.23685117065906525,
+      "learning_rate": 4.129691184489371e-05,
+      "loss": 0.3339,
+      "step": 2105
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.2584911584854126,
+      "learning_rate": 4.1255770247504576e-05,
+      "loss": 0.3352,
+      "step": 2110
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.25693652033805847,
+      "learning_rate": 4.12145522365689e-05,
+      "loss": 0.3371,
+      "step": 2115
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.24498964846134186,
+      "learning_rate": 4.117325800584031e-05,
+      "loss": 0.3353,
+      "step": 2120
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.24351784586906433,
+      "learning_rate": 4.113188774943071e-05,
+      "loss": 0.3285,
+      "step": 2125
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.2461797147989273,
+      "learning_rate": 4.10904416618094e-05,
+      "loss": 0.3483,
+      "step": 2130
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.23336690664291382,
+      "learning_rate": 4.10489199378021e-05,
+      "loss": 0.338,
+      "step": 2135
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.269683837890625,
+      "learning_rate": 4.100732277259012e-05,
+      "loss": 0.3329,
+      "step": 2140
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.24568964540958405,
+      "learning_rate": 4.0965650361709363e-05,
+      "loss": 0.3387,
+      "step": 2145
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2428756207227707,
+      "learning_rate": 4.092390290104946e-05,
+      "loss": 0.3479,
+      "step": 2150
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2435704469680786,
+      "learning_rate": 4.088208058685282e-05,
+      "loss": 0.3268,
+      "step": 2155
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2513998746871948,
+      "learning_rate": 4.0840183615713714e-05,
+      "loss": 0.334,
+      "step": 2160
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2560335099697113,
+      "learning_rate": 4.0798212184577354e-05,
+      "loss": 0.3422,
+      "step": 2165
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.2384011596441269,
+      "learning_rate": 4.075616649073898e-05,
+      "loss": 0.3463,
+      "step": 2170
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2399209588766098,
+      "learning_rate": 4.071404673184289e-05,
+      "loss": 0.3325,
+      "step": 2175
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.26694726943969727,
+      "learning_rate": 4.067185310588158e-05,
+      "loss": 0.3439,
+      "step": 2180
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.26630017161369324,
+      "learning_rate": 4.0629585811194726e-05,
+      "loss": 0.3414,
+      "step": 2185
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.24448637664318085,
+      "learning_rate": 4.058724504646834e-05,
+      "loss": 0.3394,
+      "step": 2190
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.2364281266927719,
+      "learning_rate": 4.0544831010733774e-05,
+      "loss": 0.322,
+      "step": 2195
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2470795065164566,
+      "learning_rate": 4.05023439033668e-05,
+      "loss": 0.3288,
+      "step": 2200
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.24532996118068695,
+      "learning_rate": 4.04597839240867e-05,
+      "loss": 0.342,
+      "step": 2205
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2632855176925659,
+      "learning_rate": 4.041715127295529e-05,
+      "loss": 0.3371,
+      "step": 2210
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.22981517016887665,
+      "learning_rate": 4.0374446150376e-05,
+      "loss": 0.3346,
+      "step": 2215
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.2959480583667755,
+      "learning_rate": 4.033166875709291e-05,
+      "loss": 0.3478,
+      "step": 2220
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.24138358235359192,
+      "learning_rate": 4.028881929418985e-05,
+      "loss": 0.3375,
+      "step": 2225
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.24090544879436493,
+      "learning_rate": 4.024589796308941e-05,
+      "loss": 0.3367,
+      "step": 2230
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.24531295895576477,
+      "learning_rate": 4.020290496555202e-05,
+      "loss": 0.3188,
+      "step": 2235
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2573501169681549,
+      "learning_rate": 4.015984050367499e-05,
+      "loss": 0.3348,
+      "step": 2240
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.2843419909477234,
+      "learning_rate": 4.0116704779891545e-05,
+      "loss": 0.3297,
+      "step": 2245
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.25673505663871765,
+      "learning_rate": 4.0073497996969936e-05,
+      "loss": 0.3298,
+      "step": 2250
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.26149725914001465,
+      "learning_rate": 4.003022035801238e-05,
+      "loss": 0.3209,
+      "step": 2255
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.23931410908699036,
+      "learning_rate": 3.9986872066454226e-05,
+      "loss": 0.3267,
+      "step": 2260
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2629973292350769,
+      "learning_rate": 3.9943453326062907e-05,
+      "loss": 0.3343,
+      "step": 2265
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.2728996276855469,
+      "learning_rate": 3.9899964340937025e-05,
+      "loss": 0.3394,
+      "step": 2270
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2435467392206192,
+      "learning_rate": 3.9856405315505374e-05,
+      "loss": 0.3177,
+      "step": 2275
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2617558240890503,
+      "learning_rate": 3.9812776454525984e-05,
+      "loss": 0.331,
+      "step": 2280
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.24701017141342163,
+      "learning_rate": 3.976907796308519e-05,
+      "loss": 0.3327,
+      "step": 2285
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.2749021053314209,
+      "learning_rate": 3.9725310046596595e-05,
+      "loss": 0.3374,
+      "step": 2290
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.23305217921733856,
+      "learning_rate": 3.9681472910800196e-05,
+      "loss": 0.3352,
+      "step": 2295
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.29269129037857056,
+      "learning_rate": 3.963756676176132e-05,
+      "loss": 0.3338,
+      "step": 2300
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.23365184664726257,
+      "learning_rate": 3.959359180586975e-05,
+      "loss": 0.328,
+      "step": 2305
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.27199557423591614,
+      "learning_rate": 3.954954824983867e-05,
+      "loss": 0.3454,
+      "step": 2310
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.2569097876548767,
+      "learning_rate": 3.950543630070378e-05,
+      "loss": 0.341,
+      "step": 2315
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.25776511430740356,
+      "learning_rate": 3.946125616582221e-05,
+      "loss": 0.3385,
+      "step": 2320
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.24016167223453522,
+      "learning_rate": 3.941700805287168e-05,
+      "loss": 0.334,
+      "step": 2325
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.2434518188238144,
+      "learning_rate": 3.937269216984941e-05,
+      "loss": 0.3425,
+      "step": 2330
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.26765549182891846,
+      "learning_rate": 3.9328308725071204e-05,
+      "loss": 0.3344,
+      "step": 2335
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.24450050294399261,
+      "learning_rate": 3.928385792717045e-05,
+      "loss": 0.3129,
+      "step": 2340
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.28204238414764404,
+      "learning_rate": 3.923933998509713e-05,
+      "loss": 0.328,
+      "step": 2345
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.24494262039661407,
+      "learning_rate": 3.9194755108116895e-05,
+      "loss": 0.3404,
+      "step": 2350
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2702252268791199,
+      "learning_rate": 3.915010350580998e-05,
+      "loss": 0.3274,
+      "step": 2355
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2707866132259369,
+      "learning_rate": 3.9105385388070305e-05,
+      "loss": 0.332,
+      "step": 2360
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.2505589723587036,
+      "learning_rate": 3.906060096510446e-05,
+      "loss": 0.3354,
+      "step": 2365
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2387889325618744,
+      "learning_rate": 3.901575044743072e-05,
+      "loss": 0.3224,
+      "step": 2370
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2803099751472473,
+      "learning_rate": 3.8970834045878024e-05,
+      "loss": 0.33,
+      "step": 2375
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.23512738943099976,
+      "learning_rate": 3.892585197158505e-05,
+      "loss": 0.3325,
+      "step": 2380
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.25273269414901733,
+      "learning_rate": 3.8880804435999164e-05,
+      "loss": 0.3256,
+      "step": 2385
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.24171994626522064,
+      "learning_rate": 3.883569165087544e-05,
+      "loss": 0.3306,
+      "step": 2390
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.29272180795669556,
+      "learning_rate": 3.8790513828275683e-05,
+      "loss": 0.3213,
+      "step": 2395
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.22815151512622833,
+      "learning_rate": 3.874527118056741e-05,
+      "loss": 0.3239,
+      "step": 2400
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.25038009881973267,
+      "learning_rate": 3.8699963920422874e-05,
+      "loss": 0.3316,
+      "step": 2405
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.27102407813072205,
+      "learning_rate": 3.8654592260818045e-05,
+      "loss": 0.325,
+      "step": 2410
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.2518659234046936,
+      "learning_rate": 3.860915641503161e-05,
+      "loss": 0.3181,
+      "step": 2415
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.24750158190727234,
+      "learning_rate": 3.856365659664399e-05,
+      "loss": 0.3416,
+      "step": 2420
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2569617033004761,
+      "learning_rate": 3.851809301953631e-05,
+      "loss": 0.3365,
+      "step": 2425
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.2528016269207001,
+      "learning_rate": 3.8472465897889394e-05,
+      "loss": 0.3351,
+      "step": 2430
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.25869566202163696,
+      "learning_rate": 3.842677544618279e-05,
+      "loss": 0.3232,
+      "step": 2435
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.26310843229293823,
+      "learning_rate": 3.8381021879193766e-05,
+      "loss": 0.3202,
+      "step": 2440
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.22816044092178345,
+      "learning_rate": 3.833520541199622e-05,
+      "loss": 0.3276,
+      "step": 2445
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24431347846984863,
+      "learning_rate": 3.828932625995977e-05,
+      "loss": 0.3206,
+      "step": 2450
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24391967058181763,
+      "learning_rate": 3.824338463874866e-05,
+      "loss": 0.3217,
+      "step": 2455
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24665364623069763,
+      "learning_rate": 3.819738076432081e-05,
+      "loss": 0.3106,
+      "step": 2460
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.27670589089393616,
+      "learning_rate": 3.815131485292678e-05,
+      "loss": 0.3185,
+      "step": 2465
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.2397053837776184,
+      "learning_rate": 3.8105187121108723e-05,
+      "loss": 0.2967,
+      "step": 2470
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.24300773441791534,
+      "learning_rate": 3.805899778569939e-05,
+      "loss": 0.3019,
+      "step": 2475
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.2437438815832138,
+      "learning_rate": 3.801274706382115e-05,
+      "loss": 0.307,
+      "step": 2480
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.2918024957180023,
+      "learning_rate": 3.7966435172884885e-05,
+      "loss": 0.3048,
+      "step": 2485
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.24549159407615662,
+      "learning_rate": 3.7920062330589044e-05,
+      "loss": 0.3089,
+      "step": 2490
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.26322951912879944,
+      "learning_rate": 3.787362875491858e-05,
+      "loss": 0.3029,
+      "step": 2495
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.24100036919116974,
+      "learning_rate": 3.782713466414395e-05,
+      "loss": 0.3065,
+      "step": 2500
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.2686925232410431,
+      "learning_rate": 3.778058027682004e-05,
+      "loss": 0.2978,
+      "step": 2505
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.26946887373924255,
+      "learning_rate": 3.77339658117852e-05,
+      "loss": 0.303,
+      "step": 2510
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.2917940616607666,
+      "learning_rate": 3.7687291488160195e-05,
+      "loss": 0.3045,
+      "step": 2515
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.2630554735660553,
+      "learning_rate": 3.764055752534714e-05,
+      "loss": 0.3062,
+      "step": 2520
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.2723802924156189,
+      "learning_rate": 3.759376414302852e-05,
+      "loss": 0.3068,
+      "step": 2525
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.25582003593444824,
+      "learning_rate": 3.754691156116612e-05,
+      "loss": 0.2995,
+      "step": 2530
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.2598457932472229,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.3025,
+      "step": 2535
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.2770859897136688,
+      "learning_rate": 3.745302968004749e-05,
+      "loss": 0.3104,
+      "step": 2540
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.28139764070510864,
+      "learning_rate": 3.7406000822102085e-05,
+      "loss": 0.2987,
+      "step": 2545
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.24686959385871887,
+      "learning_rate": 3.73589136472325e-05,
+      "loss": 0.2982,
+      "step": 2550
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.24469734728336334,
+      "learning_rate": 3.7311768376781536e-05,
+      "loss": 0.2884,
+      "step": 2555
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.2741526961326599,
+      "learning_rate": 3.726456523236511e-05,
+      "loss": 0.3088,
+      "step": 2560
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.24935713410377502,
+      "learning_rate": 3.721730443587118e-05,
+      "loss": 0.319,
+      "step": 2565
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.2700643539428711,
+      "learning_rate": 3.716998620945871e-05,
+      "loss": 0.3014,
+      "step": 2570
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.27301180362701416,
+      "learning_rate": 3.712261077555663e-05,
+      "loss": 0.2993,
+      "step": 2575
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.29451754689216614,
+      "learning_rate": 3.707517835686278e-05,
+      "loss": 0.3202,
+      "step": 2580
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.24200481176376343,
+      "learning_rate": 3.7027689176342855e-05,
+      "loss": 0.3103,
+      "step": 2585
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.28864267468452454,
+      "learning_rate": 3.69801434572294e-05,
+      "loss": 0.3069,
+      "step": 2590
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.26540395617485046,
+      "learning_rate": 3.693254142302071e-05,
+      "loss": 0.2953,
+      "step": 2595
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.24319253861904144,
+      "learning_rate": 3.68848832974798e-05,
+      "loss": 0.3035,
+      "step": 2600
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.27891623973846436,
+      "learning_rate": 3.683716930463335e-05,
+      "loss": 0.3118,
+      "step": 2605
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.26043465733528137,
+      "learning_rate": 3.67893996687707e-05,
+      "loss": 0.3031,
+      "step": 2610
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.24342221021652222,
+      "learning_rate": 3.6741574614442683e-05,
+      "loss": 0.3072,
+      "step": 2615
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2626497447490692,
+      "learning_rate": 3.669369436646069e-05,
+      "loss": 0.2972,
+      "step": 2620
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2995108664035797,
+      "learning_rate": 3.6645759149895526e-05,
+      "loss": 0.2956,
+      "step": 2625
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2610829770565033,
+      "learning_rate": 3.659776919007642e-05,
+      "loss": 0.292,
+      "step": 2630
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.2650500237941742,
+      "learning_rate": 3.6549724712589905e-05,
+      "loss": 0.3038,
+      "step": 2635
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.254011869430542,
+      "learning_rate": 3.6501625943278805e-05,
+      "loss": 0.2968,
+      "step": 2640
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.26473093032836914,
+      "learning_rate": 3.645347310824114e-05,
+      "loss": 0.3032,
+      "step": 2645
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.2749622166156769,
+      "learning_rate": 3.6405266433829075e-05,
+      "loss": 0.2985,
+      "step": 2650
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.25224971771240234,
+      "learning_rate": 3.635700614664788e-05,
+      "loss": 0.3101,
+      "step": 2655
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.2885672450065613,
+      "learning_rate": 3.6308692473554814e-05,
+      "loss": 0.3013,
+      "step": 2660
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.2557135820388794,
+      "learning_rate": 3.62603256416581e-05,
+      "loss": 0.3122,
+      "step": 2665
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.25804078578948975,
+      "learning_rate": 3.621190587831586e-05,
+      "loss": 0.2944,
+      "step": 2670
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.2645909786224365,
+      "learning_rate": 3.6163433411135004e-05,
+      "loss": 0.3114,
+      "step": 2675
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.28380557894706726,
+      "learning_rate": 3.61149084679702e-05,
+      "loss": 0.2808,
+      "step": 2680
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.276014119386673,
+      "learning_rate": 3.60663312769228e-05,
+      "loss": 0.2971,
+      "step": 2685
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.2478584200143814,
+      "learning_rate": 3.601770206633974e-05,
+      "loss": 0.3049,
+      "step": 2690
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.26906976103782654,
+      "learning_rate": 3.596902106481249e-05,
+      "loss": 0.2988,
+      "step": 2695
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.3004421591758728,
+      "learning_rate": 3.592028850117598e-05,
+      "loss": 0.3063,
+      "step": 2700
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.24905380606651306,
+      "learning_rate": 3.5871504604507514e-05,
+      "loss": 0.3028,
+      "step": 2705
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.2927190065383911,
+      "learning_rate": 3.5822669604125684e-05,
+      "loss": 0.3095,
+      "step": 2710
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.2823159098625183,
+      "learning_rate": 3.5773783729589335e-05,
+      "loss": 0.3066,
+      "step": 2715
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.2610376179218292,
+      "learning_rate": 3.572484721069643e-05,
+      "loss": 0.2799,
+      "step": 2720
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.2829107642173767,
+      "learning_rate": 3.567586027748302e-05,
+      "loss": 0.3046,
+      "step": 2725
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.2743787169456482,
+      "learning_rate": 3.562682316022211e-05,
+      "loss": 0.299,
+      "step": 2730
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.25424623489379883,
+      "learning_rate": 3.557773608942262e-05,
+      "loss": 0.3038,
+      "step": 2735
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.2697458863258362,
+      "learning_rate": 3.5528599295828305e-05,
+      "loss": 0.288,
+      "step": 2740
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.27604496479034424,
+      "learning_rate": 3.547941301041661e-05,
+      "loss": 0.3043,
+      "step": 2745
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.2915935814380646,
+      "learning_rate": 3.5430177464397654e-05,
+      "loss": 0.315,
+      "step": 2750
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.2570568919181824,
+      "learning_rate": 3.5380892889213125e-05,
+      "loss": 0.2963,
+      "step": 2755
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.2617090046405792,
+      "learning_rate": 3.533155951653516e-05,
+      "loss": 0.3074,
+      "step": 2760
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.29622194170951843,
+      "learning_rate": 3.5282177578265296e-05,
+      "loss": 0.3163,
+      "step": 2765
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.2855910658836365,
+      "learning_rate": 3.5232747306533344e-05,
+      "loss": 0.3083,
+      "step": 2770
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.2611843943595886,
+      "learning_rate": 3.5183268933696315e-05,
+      "loss": 0.2975,
+      "step": 2775
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.27095702290534973,
+      "learning_rate": 3.513374269233737e-05,
+      "loss": 0.2964,
+      "step": 2780
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.27189502120018005,
+      "learning_rate": 3.508416881526464e-05,
+      "loss": 0.29,
+      "step": 2785
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.2780579924583435,
+      "learning_rate": 3.503454753551019e-05,
+      "loss": 0.311,
+      "step": 2790
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.25616490840911865,
+      "learning_rate": 3.4984879086328914e-05,
+      "loss": 0.2872,
+      "step": 2795
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.2975301742553711,
+      "learning_rate": 3.493516370119745e-05,
+      "loss": 0.2848,
+      "step": 2800
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.26427510380744934,
+      "learning_rate": 3.488540161381304e-05,
+      "loss": 0.2972,
+      "step": 2805
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.2540915310382843,
+      "learning_rate": 3.483559305809248e-05,
+      "loss": 0.3071,
+      "step": 2810
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.2644979953765869,
+      "learning_rate": 3.478573826817099e-05,
+      "loss": 0.2995,
+      "step": 2815
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.26824507117271423,
+      "learning_rate": 3.473583747840112e-05,
+      "loss": 0.3093,
+      "step": 2820
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.25197768211364746,
+      "learning_rate": 3.4685890923351665e-05,
+      "loss": 0.2987,
+      "step": 2825
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.271640419960022,
+      "learning_rate": 3.4635898837806535e-05,
+      "loss": 0.3068,
+      "step": 2830
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.30842480063438416,
+      "learning_rate": 3.4585861456763686e-05,
+      "loss": 0.289,
+      "step": 2835
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.29954609274864197,
+      "learning_rate": 3.4535779015433975e-05,
+      "loss": 0.3062,
+      "step": 2840
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.25205594301223755,
+      "learning_rate": 3.4485651749240066e-05,
+      "loss": 0.2952,
+      "step": 2845
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.28727492690086365,
+      "learning_rate": 3.443547989381536e-05,
+      "loss": 0.2973,
+      "step": 2850
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.2634449303150177,
+      "learning_rate": 3.4385263685002846e-05,
+      "loss": 0.2976,
+      "step": 2855
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.2632748782634735,
+      "learning_rate": 3.4335003358853996e-05,
+      "loss": 0.3032,
+      "step": 2860
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.3057663142681122,
+      "learning_rate": 3.428469915162767e-05,
+      "loss": 0.2987,
+      "step": 2865
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.2766965627670288,
+      "learning_rate": 3.423435129978901e-05,
+      "loss": 0.2995,
+      "step": 2870
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.27240169048309326,
+      "learning_rate": 3.4183960040008294e-05,
+      "loss": 0.31,
+      "step": 2875
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.2594124674797058,
+      "learning_rate": 3.413352560915988e-05,
+      "loss": 0.2862,
+      "step": 2880
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.3315170109272003,
+      "learning_rate": 3.408304824432104e-05,
+      "loss": 0.2945,
+      "step": 2885
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.29100900888442993,
+      "learning_rate": 3.403252818277086e-05,
+      "loss": 0.3037,
+      "step": 2890
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.2730395495891571,
+      "learning_rate": 3.3981965661989146e-05,
+      "loss": 0.3069,
+      "step": 2895
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.3000994026660919,
+      "learning_rate": 3.393136091965529e-05,
+      "loss": 0.3097,
+      "step": 2900
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.26236626505851746,
+      "learning_rate": 3.3880714193647155e-05,
+      "loss": 0.2941,
+      "step": 2905
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.27992677688598633,
+      "learning_rate": 3.383002572203995e-05,
+      "loss": 0.2901,
+      "step": 2910
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.2957090735435486,
+      "learning_rate": 3.377929574310512e-05,
+      "loss": 0.3025,
+      "step": 2915
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.2695033848285675,
+      "learning_rate": 3.372852449530922e-05,
+      "loss": 0.3033,
+      "step": 2920
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.28816473484039307,
+      "learning_rate": 3.3677712217312805e-05,
+      "loss": 0.2997,
+      "step": 2925
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.26239269971847534,
+      "learning_rate": 3.36268591479693e-05,
+      "loss": 0.3009,
+      "step": 2930
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.2978915572166443,
+      "learning_rate": 3.357596552632388e-05,
+      "loss": 0.3104,
+      "step": 2935
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.24776782095432281,
+      "learning_rate": 3.352503159161234e-05,
+      "loss": 0.2902,
+      "step": 2940
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.25982436537742615,
+      "learning_rate": 3.347405758325996e-05,
+      "loss": 0.288,
+      "step": 2945
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.28331804275512695,
+      "learning_rate": 3.342304374088043e-05,
+      "loss": 0.2905,
+      "step": 2950
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.288766086101532,
+      "learning_rate": 3.3371990304274656e-05,
+      "loss": 0.2891,
+      "step": 2955
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2807261049747467,
+      "learning_rate": 3.332089751342968e-05,
+      "loss": 0.3077,
+      "step": 2960
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2729346752166748,
+      "learning_rate": 3.326976560851753e-05,
+      "loss": 0.2877,
+      "step": 2965
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2683100402355194,
+      "learning_rate": 3.32185948298941e-05,
+      "loss": 0.3009,
+      "step": 2970
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2973572313785553,
+      "learning_rate": 3.316738541809803e-05,
+      "loss": 0.3016,
+      "step": 2975
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.2905617952346802,
+      "learning_rate": 3.311613761384953e-05,
+      "loss": 0.2986,
+      "step": 2980
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.29995477199554443,
+      "learning_rate": 3.3064851658049314e-05,
+      "loss": 0.3051,
+      "step": 2985
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.2770727872848511,
+      "learning_rate": 3.301352779177743e-05,
+      "loss": 0.2879,
+      "step": 2990
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.31071165204048157,
+      "learning_rate": 3.2962166256292113e-05,
+      "loss": 0.2834,
+      "step": 2995
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.2740139067173004,
+      "learning_rate": 3.291076729302869e-05,
+      "loss": 0.3002,
+      "step": 3000
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.2945749759674072,
+      "learning_rate": 3.285933114359842e-05,
+      "loss": 0.3024,
+      "step": 3005
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.2676278054714203,
+      "learning_rate": 3.280785804978735e-05,
+      "loss": 0.3056,
+      "step": 3010
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.25451141595840454,
+      "learning_rate": 3.2756348253555215e-05,
+      "loss": 0.2962,
+      "step": 3015
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.26468273997306824,
+      "learning_rate": 3.270480199703425e-05,
+      "loss": 0.3063,
+      "step": 3020
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.26697760820388794,
+      "learning_rate": 3.2653219522528106e-05,
+      "loss": 0.2836,
+      "step": 3025
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.28052154183387756,
+      "learning_rate": 3.260160107251067e-05,
+      "loss": 0.3027,
+      "step": 3030
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.2898489832878113,
+      "learning_rate": 3.2549946889624926e-05,
+      "loss": 0.2968,
+      "step": 3035
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.29555365443229675,
+      "learning_rate": 3.249825721668185e-05,
+      "loss": 0.2911,
+      "step": 3040
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.2801489531993866,
+      "learning_rate": 3.244653229665925e-05,
+      "loss": 0.3057,
+      "step": 3045
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.26459407806396484,
+      "learning_rate": 3.23947723727006e-05,
+      "loss": 0.3059,
+      "step": 3050
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.311621755361557,
+      "learning_rate": 3.2342977688113915e-05,
+      "loss": 0.3057,
+      "step": 3055
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.2606475353240967,
+      "learning_rate": 3.2291148486370626e-05,
+      "loss": 0.2905,
+      "step": 3060
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.29442301392555237,
+      "learning_rate": 3.223928501110441e-05,
+      "loss": 0.2962,
+      "step": 3065
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.26171961426734924,
+      "learning_rate": 3.218738750611008e-05,
+      "loss": 0.2835,
+      "step": 3070
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.27043333649635315,
+      "learning_rate": 3.2135456215342356e-05,
+      "loss": 0.2957,
+      "step": 3075
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3261232376098633,
+      "learning_rate": 3.2083491382914824e-05,
+      "loss": 0.2853,
+      "step": 3080
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.2558610141277313,
+      "learning_rate": 3.2031493253098724e-05,
+      "loss": 0.2842,
+      "step": 3085
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.2991327941417694,
+      "learning_rate": 3.197946207032181e-05,
+      "loss": 0.3079,
+      "step": 3090
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.31375688314437866,
+      "learning_rate": 3.1927398079167226e-05,
+      "loss": 0.3053,
+      "step": 3095
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.2674601078033447,
+      "learning_rate": 3.1875301524372306e-05,
+      "loss": 0.2937,
+      "step": 3100
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.27789610624313354,
+      "learning_rate": 3.1823172650827496e-05,
+      "loss": 0.2926,
+      "step": 3105
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.28713828325271606,
+      "learning_rate": 3.177101170357513e-05,
+      "loss": 0.2989,
+      "step": 3110
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.2908244729042053,
+      "learning_rate": 3.1718818927808335e-05,
+      "loss": 0.2911,
+      "step": 3115
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.2980084717273712,
+      "learning_rate": 3.166659456886983e-05,
+      "loss": 0.2933,
+      "step": 3120
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.2810494005680084,
+      "learning_rate": 3.1614338872250836e-05,
+      "loss": 0.2973,
+      "step": 3125
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.29591745138168335,
+      "learning_rate": 3.156205208358984e-05,
+      "loss": 0.2922,
+      "step": 3130
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.2694539725780487,
+      "learning_rate": 3.150973444867151e-05,
+      "loss": 0.2868,
+      "step": 3135
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.29078954458236694,
+      "learning_rate": 3.145738621342549e-05,
+      "loss": 0.2935,
+      "step": 3140
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.24549968540668488,
+      "learning_rate": 3.14050076239253e-05,
+      "loss": 0.2838,
+      "step": 3145
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.2833496630191803,
+      "learning_rate": 3.1352598926387124e-05,
+      "loss": 0.2993,
+      "step": 3150
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.29394856095314026,
+      "learning_rate": 3.130016036716867e-05,
+      "loss": 0.2924,
+      "step": 3155
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.4183152914047241,
+      "learning_rate": 3.124769219276803e-05,
+      "loss": 0.3023,
+      "step": 3160
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.269716739654541,
+      "learning_rate": 3.11951946498225e-05,
+      "loss": 0.2817,
+      "step": 3165
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.28056713938713074,
+      "learning_rate": 3.114266798510744e-05,
+      "loss": 0.2947,
+      "step": 3170
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.2674812376499176,
+      "learning_rate": 3.1090112445535104e-05,
+      "loss": 0.2985,
+      "step": 3175
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.2738688290119171,
+      "learning_rate": 3.103752827815345e-05,
+      "loss": 0.3015,
+      "step": 3180
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.2820100784301758,
+      "learning_rate": 3.0984915730145046e-05,
+      "loss": 0.2895,
+      "step": 3185
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.28290191292762756,
+      "learning_rate": 3.093227504882584e-05,
+      "loss": 0.2848,
+      "step": 3190
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.2668739855289459,
+      "learning_rate": 3.087960648164403e-05,
+      "loss": 0.2917,
+      "step": 3195
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.2666045129299164,
+      "learning_rate": 3.0826910276178916e-05,
+      "loss": 0.2987,
+      "step": 3200
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.27977439761161804,
+      "learning_rate": 3.07741866801397e-05,
+      "loss": 0.2926,
+      "step": 3205
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.27020302414894104,
+      "learning_rate": 3.072143594136433e-05,
+      "loss": 0.3003,
+      "step": 3210
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.2469637244939804,
+      "learning_rate": 3.0668658307818385e-05,
+      "loss": 0.2842,
+      "step": 3215
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.28846561908721924,
+      "learning_rate": 3.061585402759381e-05,
+      "loss": 0.298,
+      "step": 3220
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.2849927544593811,
+      "learning_rate": 3.056302334890786e-05,
+      "loss": 0.2925,
+      "step": 3225
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.28161710500717163,
+      "learning_rate": 3.051016652010185e-05,
+      "loss": 0.2967,
+      "step": 3230
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.2685248851776123,
+      "learning_rate": 3.045728378964004e-05,
+      "loss": 0.2872,
+      "step": 3235
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.270854651927948,
+      "learning_rate": 3.0404375406108427e-05,
+      "loss": 0.2827,
+      "step": 3240
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.29226136207580566,
+      "learning_rate": 3.0351441618213604e-05,
+      "loss": 0.3056,
+      "step": 3245
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.27794334292411804,
+      "learning_rate": 3.029848267478158e-05,
+      "loss": 0.2884,
+      "step": 3250
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.2685510516166687,
+      "learning_rate": 3.024549882475662e-05,
+      "loss": 0.2922,
+      "step": 3255
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.2610897421836853,
+      "learning_rate": 3.0192490317200058e-05,
+      "loss": 0.2917,
+      "step": 3260
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.2856181859970093,
+      "learning_rate": 3.013945740128914e-05,
+      "loss": 0.2885,
+      "step": 3265
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.26383763551712036,
+      "learning_rate": 3.008640032631585e-05,
+      "loss": 0.3069,
+      "step": 3270
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.28392094373703003,
+      "learning_rate": 3.003331934168573e-05,
+      "loss": 0.2916,
+      "step": 3275
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.2767746150493622,
+      "learning_rate": 2.9980214696916726e-05,
+      "loss": 0.2977,
+      "step": 3280
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3030785918235779,
+      "learning_rate": 2.9927086641637997e-05,
+      "loss": 0.3008,
+      "step": 3285
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.29651913046836853,
+      "learning_rate": 2.9873935425588734e-05,
+      "loss": 0.2983,
+      "step": 3290
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.31585001945495605,
+      "learning_rate": 2.9820761298617028e-05,
+      "loss": 0.2928,
+      "step": 3295
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.2689068615436554,
+      "learning_rate": 2.9767564510678652e-05,
+      "loss": 0.2923,
+      "step": 3300
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.27636662125587463,
+      "learning_rate": 2.9714345311835895e-05,
+      "loss": 0.2983,
+      "step": 3305
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.274234414100647,
+      "learning_rate": 2.9661103952256396e-05,
+      "loss": 0.2958,
+      "step": 3310
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.294877827167511,
+      "learning_rate": 2.9607840682211985e-05,
+      "loss": 0.294,
+      "step": 3315
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.2964394688606262,
+      "learning_rate": 2.9554555752077457e-05,
+      "loss": 0.2986,
+      "step": 3320
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.2825262248516083,
+      "learning_rate": 2.950124941232944e-05,
+      "loss": 0.2926,
+      "step": 3325
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.2611103653907776,
+      "learning_rate": 2.9447921913545197e-05,
+      "loss": 0.2917,
+      "step": 3330
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.2674500048160553,
+      "learning_rate": 2.9394573506401467e-05,
+      "loss": 0.3018,
+      "step": 3335
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.2672838270664215,
+      "learning_rate": 2.9341204441673266e-05,
+      "loss": 0.2927,
+      "step": 3340
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3334977328777313,
+      "learning_rate": 2.9287814970232686e-05,
+      "loss": 0.2905,
+      "step": 3345
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.2567777633666992,
+      "learning_rate": 2.9234405343047804e-05,
+      "loss": 0.2905,
+      "step": 3350
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.28230491280555725,
+      "learning_rate": 2.9180975811181392e-05,
+      "loss": 0.2948,
+      "step": 3355
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.32607340812683105,
+      "learning_rate": 2.9127526625789808e-05,
+      "loss": 0.2919,
+      "step": 3360
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.2736309766769409,
+      "learning_rate": 2.9074058038121794e-05,
+      "loss": 0.3001,
+      "step": 3365
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3076474666595459,
+      "learning_rate": 2.9020570299517297e-05,
+      "loss": 0.2869,
+      "step": 3370
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.2726801037788391,
+      "learning_rate": 2.8967063661406285e-05,
+      "loss": 0.2921,
+      "step": 3375
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.7224235534667969,
+      "learning_rate": 2.891353837530757e-05,
+      "loss": 0.3094,
+      "step": 3380
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.27749213576316833,
+      "learning_rate": 2.885999469282761e-05,
+      "loss": 0.2868,
+      "step": 3385
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.31395354866981506,
+      "learning_rate": 2.880643286565936e-05,
+      "loss": 0.3055,
+      "step": 3390
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.276438444852829,
+      "learning_rate": 2.875285314558106e-05,
+      "loss": 0.2913,
+      "step": 3395
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.32008302211761475,
+      "learning_rate": 2.8699255784455043e-05,
+      "loss": 0.2878,
+      "step": 3400
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.2967100143432617,
+      "learning_rate": 2.8645641034226583e-05,
+      "loss": 0.2918,
+      "step": 3405
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.28055939078330994,
+      "learning_rate": 2.8592009146922704e-05,
+      "loss": 0.2887,
+      "step": 3410
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.2662840485572815,
+      "learning_rate": 2.853836037465097e-05,
+      "loss": 0.2848,
+      "step": 3415
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.2847791612148285,
+      "learning_rate": 2.848469496959832e-05,
+      "loss": 0.2967,
+      "step": 3420
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.2742089331150055,
+      "learning_rate": 2.8431013184029876e-05,
+      "loss": 0.2887,
+      "step": 3425
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.2804618775844574,
+      "learning_rate": 2.8377315270287764e-05,
+      "loss": 0.2881,
+      "step": 3430
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.27763351798057556,
+      "learning_rate": 2.8323601480789928e-05,
+      "loss": 0.2884,
+      "step": 3435
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.2841647267341614,
+      "learning_rate": 2.8269872068028925e-05,
+      "loss": 0.283,
+      "step": 3440
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.28074926137924194,
+      "learning_rate": 2.821612728457078e-05,
+      "loss": 0.2813,
+      "step": 3445
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3072793781757355,
+      "learning_rate": 2.816236738305374e-05,
+      "loss": 0.2927,
+      "step": 3450
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.26491352915763855,
+      "learning_rate": 2.8108592616187133e-05,
+      "loss": 0.2884,
+      "step": 3455
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.28425270318984985,
+      "learning_rate": 2.805480323675016e-05,
+      "loss": 0.2925,
+      "step": 3460
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3058440089225769,
+      "learning_rate": 2.800099949759072e-05,
+      "loss": 0.2867,
+      "step": 3465
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.2989778518676758,
+      "learning_rate": 2.79471816516242e-05,
+      "loss": 0.2926,
+      "step": 3470
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.2917656898498535,
+      "learning_rate": 2.789334995183231e-05,
+      "loss": 0.2989,
+      "step": 3475
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.27660423517227173,
+      "learning_rate": 2.7839504651261872e-05,
+      "loss": 0.2854,
+      "step": 3480
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.3180350661277771,
+      "learning_rate": 2.7785646003023656e-05,
+      "loss": 0.2917,
+      "step": 3485
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.2923102080821991,
+      "learning_rate": 2.773177426029116e-05,
+      "loss": 0.2871,
+      "step": 3490
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.2785269320011139,
+      "learning_rate": 2.767788967629944e-05,
+      "loss": 0.2989,
+      "step": 3495
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.2686062157154083,
+      "learning_rate": 2.762399250434392e-05,
+      "loss": 0.2842,
+      "step": 3500
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.27798107266426086,
+      "learning_rate": 2.75700829977792e-05,
+      "loss": 0.2839,
+      "step": 3505
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.28423336148262024,
+      "learning_rate": 2.7516161410017844e-05,
+      "loss": 0.279,
+      "step": 3510
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.2867024838924408,
+      "learning_rate": 2.7462227994529217e-05,
+      "loss": 0.2829,
+      "step": 3515
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3039509654045105,
+      "learning_rate": 2.7408283004838293e-05,
+      "loss": 0.2857,
+      "step": 3520
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.28565874695777893,
+      "learning_rate": 2.735432669452443e-05,
+      "loss": 0.2909,
+      "step": 3525
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2665278911590576,
+      "learning_rate": 2.730035931722022e-05,
+      "loss": 0.2918,
+      "step": 3530
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.3140422999858856,
+      "learning_rate": 2.7246381126610272e-05,
+      "loss": 0.287,
+      "step": 3535
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.2795071005821228,
+      "learning_rate": 2.7192392376430016e-05,
+      "loss": 0.2976,
+      "step": 3540
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.40941300988197327,
+      "learning_rate": 2.7138393320464533e-05,
+      "loss": 0.2853,
+      "step": 3545
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3148713707923889,
+      "learning_rate": 2.7084384212547337e-05,
+      "loss": 0.2962,
+      "step": 3550
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.2854318618774414,
+      "learning_rate": 2.7030365306559207e-05,
+      "loss": 0.2892,
+      "step": 3555
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.26501521468162537,
+      "learning_rate": 2.697633685642697e-05,
+      "loss": 0.2875,
+      "step": 3560
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.2624688148498535,
+      "learning_rate": 2.6922299116122317e-05,
+      "loss": 0.2925,
+      "step": 3565
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3039272427558899,
+      "learning_rate": 2.686825233966061e-05,
+      "loss": 0.2892,
+      "step": 3570
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.28600260615348816,
+      "learning_rate": 2.6814196781099686e-05,
+      "loss": 0.2927,
+      "step": 3575
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.28527864813804626,
+      "learning_rate": 2.676013269453867e-05,
+      "loss": 0.281,
+      "step": 3580
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.28271040320396423,
+      "learning_rate": 2.6706060334116777e-05,
+      "loss": 0.2907,
+      "step": 3585
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.30454739928245544,
+      "learning_rate": 2.66519799540121e-05,
+      "loss": 0.2782,
+      "step": 3590
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.2760806679725647,
+      "learning_rate": 2.6597891808440446e-05,
+      "loss": 0.3052,
+      "step": 3595
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.3208744525909424,
+      "learning_rate": 2.654379615165411e-05,
+      "loss": 0.2937,
+      "step": 3600
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.28671467304229736,
+      "learning_rate": 2.6489693237940705e-05,
+      "loss": 0.2922,
+      "step": 3605
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.2875402867794037,
+      "learning_rate": 2.6435583321621966e-05,
+      "loss": 0.2946,
+      "step": 3610
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.26397180557250977,
+      "learning_rate": 2.6381466657052522e-05,
+      "loss": 0.294,
+      "step": 3615
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.3011622130870819,
+      "learning_rate": 2.632734349861874e-05,
+      "loss": 0.2923,
+      "step": 3620
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2682401239871979,
+      "learning_rate": 2.6273214100737516e-05,
+      "loss": 0.2841,
+      "step": 3625
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2877505421638489,
+      "learning_rate": 2.6219078717855043e-05,
+      "loss": 0.3014,
+      "step": 3630
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.2974471151828766,
+      "learning_rate": 2.6164937604445705e-05,
+      "loss": 0.2906,
+      "step": 3635
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.27440640330314636,
+      "learning_rate": 2.611079101501078e-05,
+      "loss": 0.288,
+      "step": 3640
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.27096638083457947,
+      "learning_rate": 2.6056639204077292e-05,
+      "loss": 0.2798,
+      "step": 3645
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.29108545184135437,
+      "learning_rate": 2.6002482426196816e-05,
+      "loss": 0.2846,
+      "step": 3650
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.3168250024318695,
+      "learning_rate": 2.5948320935944286e-05,
+      "loss": 0.297,
+      "step": 3655
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.279838889837265,
+      "learning_rate": 2.5894154987916764e-05,
+      "loss": 0.2769,
+      "step": 3660
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.30819419026374817,
+      "learning_rate": 2.5839984836732294e-05,
+      "loss": 0.2881,
+      "step": 3665
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.32967063784599304,
+      "learning_rate": 2.5785810737028653e-05,
+      "loss": 0.2817,
+      "step": 3670
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.2700733542442322,
+      "learning_rate": 2.573163294346219e-05,
+      "loss": 0.3013,
+      "step": 3675
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.28947699069976807,
+      "learning_rate": 2.5677451710706625e-05,
+      "loss": 0.2949,
+      "step": 3680
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.27911829948425293,
+      "learning_rate": 2.5623267293451826e-05,
+      "loss": 0.2847,
+      "step": 3685
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.26226121187210083,
+      "learning_rate": 2.556907994640264e-05,
+      "loss": 0.2986,
+      "step": 3690
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.29214856028556824,
+      "learning_rate": 2.5514889924277702e-05,
+      "loss": 0.2926,
+      "step": 3695
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.27132898569107056,
+      "learning_rate": 2.5460697481808194e-05,
+      "loss": 0.2768,
+      "step": 3700
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.34020254015922546,
+      "learning_rate": 2.540650287373669e-05,
+      "loss": 0.2865,
+      "step": 3705
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.2973337471485138,
+      "learning_rate": 2.5352306354815953e-05,
+      "loss": 0.2981,
+      "step": 3710
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.2942398488521576,
+      "learning_rate": 2.5298108179807696e-05,
+      "loss": 0.2908,
+      "step": 3715
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.3027085065841675,
+      "learning_rate": 2.5243908603481453e-05,
+      "loss": 0.3029,
+      "step": 3720
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.4202698767185211,
+      "learning_rate": 2.5189707880613334e-05,
+      "loss": 0.2893,
+      "step": 3725
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.3037981390953064,
+      "learning_rate": 2.5135506265984822e-05,
+      "loss": 0.3042,
+      "step": 3730
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.2791476845741272,
+      "learning_rate": 2.5081304014381614e-05,
+      "loss": 0.2815,
+      "step": 3735
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.28679245710372925,
+      "learning_rate": 2.5027101380592387e-05,
+      "loss": 0.2891,
+      "step": 3740
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.29510611295700073,
+      "learning_rate": 2.4972898619407622e-05,
+      "loss": 0.281,
+      "step": 3745
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.33581578731536865,
+      "learning_rate": 2.4918695985618392e-05,
+      "loss": 0.2877,
+      "step": 3750
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.30332496762275696,
+      "learning_rate": 2.4864493734015187e-05,
+      "loss": 0.2783,
+      "step": 3755
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.26364871859550476,
+      "learning_rate": 2.4810292119386675e-05,
+      "loss": 0.2825,
+      "step": 3760
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.2966884672641754,
+      "learning_rate": 2.475609139651855e-05,
+      "loss": 0.29,
+      "step": 3765
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.31036069989204407,
+      "learning_rate": 2.470189182019231e-05,
+      "loss": 0.2853,
+      "step": 3770
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.2837959825992584,
+      "learning_rate": 2.4647693645184053e-05,
+      "loss": 0.2882,
+      "step": 3775
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.6355114579200745,
+      "learning_rate": 2.459349712626331e-05,
+      "loss": 0.2871,
+      "step": 3780
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.3249611258506775,
+      "learning_rate": 2.4539302518191802e-05,
+      "loss": 0.2902,
+      "step": 3785
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.31686100363731384,
+      "learning_rate": 2.4485110075722304e-05,
+      "loss": 0.2823,
+      "step": 3790
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.26546356081962585,
+      "learning_rate": 2.4430920053597356e-05,
+      "loss": 0.276,
+      "step": 3795
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.279491662979126,
+      "learning_rate": 2.4376732706548183e-05,
+      "loss": 0.2692,
+      "step": 3800
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.2788805663585663,
+      "learning_rate": 2.432254828929338e-05,
+      "loss": 0.2877,
+      "step": 3805
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.2937188744544983,
+      "learning_rate": 2.426836705653782e-05,
+      "loss": 0.2948,
+      "step": 3810
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.2882876694202423,
+      "learning_rate": 2.4214189262971352e-05,
+      "loss": 0.2816,
+      "step": 3815
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.3089386522769928,
+      "learning_rate": 2.416001516326771e-05,
+      "loss": 0.2805,
+      "step": 3820
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.2976425290107727,
+      "learning_rate": 2.410584501208324e-05,
+      "loss": 0.2816,
+      "step": 3825
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.2887030243873596,
+      "learning_rate": 2.405167906405572e-05,
+      "loss": 0.2827,
+      "step": 3830
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.3007504940032959,
+      "learning_rate": 2.3997517573803187e-05,
+      "loss": 0.2679,
+      "step": 3835
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.2942444980144501,
+      "learning_rate": 2.394336079592271e-05,
+      "loss": 0.2812,
+      "step": 3840
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.2711913585662842,
+      "learning_rate": 2.3889208984989227e-05,
+      "loss": 0.2845,
+      "step": 3845
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3077562749385834,
+      "learning_rate": 2.383506239555429e-05,
+      "loss": 0.2886,
+      "step": 3850
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.2941572964191437,
+      "learning_rate": 2.378092128214496e-05,
+      "loss": 0.293,
+      "step": 3855
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.29964739084243774,
+      "learning_rate": 2.3726785899262493e-05,
+      "loss": 0.2784,
+      "step": 3860
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.270431786775589,
+      "learning_rate": 2.3672656501381272e-05,
+      "loss": 0.278,
+      "step": 3865
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.33198636770248413,
+      "learning_rate": 2.3618533342947487e-05,
+      "loss": 0.288,
+      "step": 3870
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.28286775946617126,
+      "learning_rate": 2.3564416678378043e-05,
+      "loss": 0.2968,
+      "step": 3875
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.3287088871002197,
+      "learning_rate": 2.35103067620593e-05,
+      "loss": 0.283,
+      "step": 3880
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.3398906886577606,
+      "learning_rate": 2.34562038483459e-05,
+      "loss": 0.2807,
+      "step": 3885
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.2893100082874298,
+      "learning_rate": 2.3402108191559563e-05,
+      "loss": 0.3067,
+      "step": 3890
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.28878894448280334,
+      "learning_rate": 2.33480200459879e-05,
+      "loss": 0.2703,
+      "step": 3895
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.3113791346549988,
+      "learning_rate": 2.329393966588323e-05,
+      "loss": 0.2771,
+      "step": 3900
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.31787440180778503,
+      "learning_rate": 2.323986730546133e-05,
+      "loss": 0.2907,
+      "step": 3905
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.24957780539989471,
+      "learning_rate": 2.318580321890032e-05,
+      "loss": 0.2748,
+      "step": 3910
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.305794894695282,
+      "learning_rate": 2.3131747660339394e-05,
+      "loss": 0.2841,
+      "step": 3915
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.2979389727115631,
+      "learning_rate": 2.3077700883877692e-05,
+      "loss": 0.2822,
+      "step": 3920
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.260028213262558,
+      "learning_rate": 2.3023663143573033e-05,
+      "loss": 0.2877,
+      "step": 3925
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.3042019307613373,
+      "learning_rate": 2.2969634693440796e-05,
+      "loss": 0.2734,
+      "step": 3930
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.28115126490592957,
+      "learning_rate": 2.2915615787452666e-05,
+      "loss": 0.2857,
+      "step": 3935
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.2826026678085327,
+      "learning_rate": 2.2861606679535473e-05,
+      "loss": 0.2828,
+      "step": 3940
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.28725960850715637,
+      "learning_rate": 2.2807607623569986e-05,
+      "loss": 0.2954,
+      "step": 3945
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.27994778752326965,
+      "learning_rate": 2.275361887338973e-05,
+      "loss": 0.2904,
+      "step": 3950
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.2928130626678467,
+      "learning_rate": 2.2699640682779782e-05,
+      "loss": 0.2899,
+      "step": 3955
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.33391568064689636,
+      "learning_rate": 2.264567330547557e-05,
+      "loss": 0.2848,
+      "step": 3960
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.25743329524993896,
+      "learning_rate": 2.2591716995161716e-05,
+      "loss": 0.2857,
+      "step": 3965
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.30197808146476746,
+      "learning_rate": 2.2537772005470782e-05,
+      "loss": 0.2921,
+      "step": 3970
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.27792906761169434,
+      "learning_rate": 2.248383858998217e-05,
+      "loss": 0.2721,
+      "step": 3975
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.2906716763973236,
+      "learning_rate": 2.242991700222081e-05,
+      "loss": 0.2863,
+      "step": 3980
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3254633843898773,
+      "learning_rate": 2.2376007495656084e-05,
+      "loss": 0.2822,
+      "step": 3985
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.2850257158279419,
+      "learning_rate": 2.2322110323700568e-05,
+      "loss": 0.2799,
+      "step": 3990
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.29622894525527954,
+      "learning_rate": 2.226822573970885e-05,
+      "loss": 0.2785,
+      "step": 3995
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.32647231221199036,
+      "learning_rate": 2.221435399697635e-05,
+      "loss": 0.2895,
+      "step": 4000
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.2990321218967438,
+      "learning_rate": 2.2160495348738123e-05,
+      "loss": 0.2763,
+      "step": 4005
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.30945920944213867,
+      "learning_rate": 2.2106650048167697e-05,
+      "loss": 0.2714,
+      "step": 4010
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.31236645579338074,
+      "learning_rate": 2.20528183483758e-05,
+      "loss": 0.2941,
+      "step": 4015
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.2706669867038727,
+      "learning_rate": 2.199900050240929e-05,
+      "loss": 0.2703,
+      "step": 4020
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.2748274505138397,
+      "learning_rate": 2.1945196763249842e-05,
+      "loss": 0.2619,
+      "step": 4025
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.3171229362487793,
+      "learning_rate": 2.189140738381288e-05,
+      "loss": 0.2791,
+      "step": 4030
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.3274431526660919,
+      "learning_rate": 2.1837632616946266e-05,
+      "loss": 0.2874,
+      "step": 4035
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.29980897903442383,
+      "learning_rate": 2.1783872715429228e-05,
+      "loss": 0.2752,
+      "step": 4040
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.29202738404273987,
+      "learning_rate": 2.1730127931971078e-05,
+      "loss": 0.2774,
+      "step": 4045
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.2927723526954651,
+      "learning_rate": 2.1676398519210084e-05,
+      "loss": 0.2881,
+      "step": 4050
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.30047136545181274,
+      "learning_rate": 2.1622684729712242e-05,
+      "loss": 0.2768,
+      "step": 4055
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.30534085631370544,
+      "learning_rate": 2.1568986815970126e-05,
+      "loss": 0.2931,
+      "step": 4060
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.30663934350013733,
+      "learning_rate": 2.151530503040169e-05,
+      "loss": 0.2904,
+      "step": 4065
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.3229426443576813,
+      "learning_rate": 2.1461639625349027e-05,
+      "loss": 0.2792,
+      "step": 4070
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.2784014642238617,
+      "learning_rate": 2.1407990853077298e-05,
+      "loss": 0.2719,
+      "step": 4075
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.2790541648864746,
+      "learning_rate": 2.1354358965773412e-05,
+      "loss": 0.2767,
+      "step": 4080
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.28797996044158936,
+      "learning_rate": 2.1300744215544967e-05,
+      "loss": 0.2858,
+      "step": 4085
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.28744253516197205,
+      "learning_rate": 2.1247146854418947e-05,
+      "loss": 0.2764,
+      "step": 4090
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.31796348094940186,
+      "learning_rate": 2.119356713434064e-05,
+      "loss": 0.2839,
+      "step": 4095
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.2939613461494446,
+      "learning_rate": 2.1140005307172393e-05,
+      "loss": 0.2848,
+      "step": 4100
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.31374332308769226,
+      "learning_rate": 2.1086461624692442e-05,
+      "loss": 0.2964,
+      "step": 4105
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.2767564058303833,
+      "learning_rate": 2.1032936338593718e-05,
+      "loss": 0.272,
+      "step": 4110
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.29127681255340576,
+      "learning_rate": 2.09794297004827e-05,
+      "loss": 0.2677,
+      "step": 4115
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.3078121840953827,
+      "learning_rate": 2.092594196187821e-05,
+      "loss": 0.2861,
+      "step": 4120
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.29198378324508667,
+      "learning_rate": 2.087247337421019e-05,
+      "loss": 0.2787,
+      "step": 4125
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.28829196095466614,
+      "learning_rate": 2.0819024188818613e-05,
+      "loss": 0.2858,
+      "step": 4130
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.29239264130592346,
+      "learning_rate": 2.0765594656952198e-05,
+      "loss": 0.2723,
+      "step": 4135
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.2830089032649994,
+      "learning_rate": 2.0712185029767316e-05,
+      "loss": 0.2727,
+      "step": 4140
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.2982872426509857,
+      "learning_rate": 2.0658795558326743e-05,
+      "loss": 0.2686,
+      "step": 4145
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.26791197061538696,
+      "learning_rate": 2.060542649359854e-05,
+      "loss": 0.2786,
+      "step": 4150
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.2683955132961273,
+      "learning_rate": 2.0552078086454805e-05,
+      "loss": 0.2946,
+      "step": 4155
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.2846348285675049,
+      "learning_rate": 2.0498750587670572e-05,
+      "loss": 0.2743,
+      "step": 4160
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.2987784743309021,
+      "learning_rate": 2.0445444247922552e-05,
+      "loss": 0.2888,
+      "step": 4165
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.31197643280029297,
+      "learning_rate": 2.0392159317788028e-05,
+      "loss": 0.2667,
+      "step": 4170
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.3034227788448334,
+      "learning_rate": 2.0338896047743607e-05,
+      "loss": 0.2748,
+      "step": 4175
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.2976836562156677,
+      "learning_rate": 2.0285654688164107e-05,
+      "loss": 0.2741,
+      "step": 4180
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.35139620304107666,
+      "learning_rate": 2.0232435489321354e-05,
+      "loss": 0.2709,
+      "step": 4185
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.30577874183654785,
+      "learning_rate": 2.0179238701382968e-05,
+      "loss": 0.2895,
+      "step": 4190
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.282113641500473,
+      "learning_rate": 2.012606457441127e-05,
+      "loss": 0.2868,
+      "step": 4195
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.30150479078292847,
+      "learning_rate": 2.0072913358362013e-05,
+      "loss": 0.2671,
+      "step": 4200
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.3025312125682831,
+      "learning_rate": 2.0019785303083276e-05,
+      "loss": 0.2752,
+      "step": 4205
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.28345394134521484,
+      "learning_rate": 1.9966680658314273e-05,
+      "loss": 0.2904,
+      "step": 4210
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.2951679229736328,
+      "learning_rate": 1.991359967368416e-05,
+      "loss": 0.274,
+      "step": 4215
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.30772852897644043,
+      "learning_rate": 1.9860542598710865e-05,
+      "loss": 0.2864,
+      "step": 4220
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.287640780210495,
+      "learning_rate": 1.980750968279995e-05,
+      "loss": 0.2752,
+      "step": 4225
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.30197873711586,
+      "learning_rate": 1.9754501175243387e-05,
+      "loss": 0.2646,
+      "step": 4230
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.2947797477245331,
+      "learning_rate": 1.970151732521842e-05,
+      "loss": 0.2796,
+      "step": 4235
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.28195419907569885,
+      "learning_rate": 1.96485583817864e-05,
+      "loss": 0.2839,
+      "step": 4240
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.35147541761398315,
+      "learning_rate": 1.9595624593891572e-05,
+      "loss": 0.2863,
+      "step": 4245
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.31853169202804565,
+      "learning_rate": 1.9542716210359964e-05,
+      "loss": 0.2981,
+      "step": 4250
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.29918932914733887,
+      "learning_rate": 1.948983347989815e-05,
+      "loss": 0.2807,
+      "step": 4255
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.28011664748191833,
+      "learning_rate": 1.9436976651092144e-05,
+      "loss": 0.2909,
+      "step": 4260
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.28573253750801086,
+      "learning_rate": 1.9384145972406193e-05,
+      "loss": 0.2858,
+      "step": 4265
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.28355416655540466,
+      "learning_rate": 1.9331341692181628e-05,
+      "loss": 0.2657,
+      "step": 4270
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.28709959983825684,
+      "learning_rate": 1.9278564058635666e-05,
+      "loss": 0.2822,
+      "step": 4275
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.30995383858680725,
+      "learning_rate": 1.9225813319860316e-05,
+      "loss": 0.277,
+      "step": 4280
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.3091290295124054,
+      "learning_rate": 1.917308972382109e-05,
+      "loss": 0.2725,
+      "step": 4285
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.2684681713581085,
+      "learning_rate": 1.9120393518355966e-05,
+      "loss": 0.2779,
+      "step": 4290
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.31061887741088867,
+      "learning_rate": 1.9067724951174167e-05,
+      "loss": 0.2751,
+      "step": 4295
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.345758318901062,
+      "learning_rate": 1.9015084269854956e-05,
+      "loss": 0.2742,
+      "step": 4300
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.28362998366355896,
+      "learning_rate": 1.8962471721846554e-05,
+      "loss": 0.2773,
+      "step": 4305
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.2853488028049469,
+      "learning_rate": 1.89098875544649e-05,
+      "loss": 0.2711,
+      "step": 4310
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.31866294145584106,
+      "learning_rate": 1.8857332014892558e-05,
+      "loss": 0.2752,
+      "step": 4315
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.28082075715065,
+      "learning_rate": 1.8804805350177505e-05,
+      "loss": 0.2848,
+      "step": 4320
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.28699925541877747,
+      "learning_rate": 1.8752307807231983e-05,
+      "loss": 0.2826,
+      "step": 4325
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.2884686589241028,
+      "learning_rate": 1.869983963283134e-05,
+      "loss": 0.2777,
+      "step": 4330
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.29716336727142334,
+      "learning_rate": 1.8647401073612892e-05,
+      "loss": 0.2669,
+      "step": 4335
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.32686588168144226,
+      "learning_rate": 1.8594992376074703e-05,
+      "loss": 0.2873,
+      "step": 4340
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.2946726083755493,
+      "learning_rate": 1.8542613786574503e-05,
+      "loss": 0.2743,
+      "step": 4345
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.30872857570648193,
+      "learning_rate": 1.8490265551328496e-05,
+      "loss": 0.279,
+      "step": 4350
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.2889072597026825,
+      "learning_rate": 1.8437947916410157e-05,
+      "loss": 0.2806,
+      "step": 4355
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.29989129304885864,
+      "learning_rate": 1.8385661127749166e-05,
+      "loss": 0.2684,
+      "step": 4360
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.29832085967063904,
+      "learning_rate": 1.833340543113016e-05,
+      "loss": 0.2671,
+      "step": 4365
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.30163899064064026,
+      "learning_rate": 1.8281181072191668e-05,
+      "loss": 0.2749,
+      "step": 4370
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.3076401352882385,
+      "learning_rate": 1.8228988296424877e-05,
+      "loss": 0.292,
+      "step": 4375
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.2921345829963684,
+      "learning_rate": 1.8176827349172516e-05,
+      "loss": 0.2828,
+      "step": 4380
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.2923434376716614,
+      "learning_rate": 1.8124698475627696e-05,
+      "loss": 0.2695,
+      "step": 4385
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.345413476228714,
+      "learning_rate": 1.8072601920832786e-05,
+      "loss": 0.2722,
+      "step": 4390
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.3174823820590973,
+      "learning_rate": 1.802053792967819e-05,
+      "loss": 0.279,
+      "step": 4395
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.6514738202095032,
+      "learning_rate": 1.796850674690127e-05,
+      "loss": 0.277,
+      "step": 4400
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.2825588285923004,
+      "learning_rate": 1.7916508617085175e-05,
+      "loss": 0.2736,
+      "step": 4405
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.2937125265598297,
+      "learning_rate": 1.7864543784657643e-05,
+      "loss": 0.2715,
+      "step": 4410
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.2738409638404846,
+      "learning_rate": 1.7812612493889928e-05,
+      "loss": 0.2863,
+      "step": 4415
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.31098294258117676,
+      "learning_rate": 1.7760714988895583e-05,
+      "loss": 0.287,
+      "step": 4420
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.30921077728271484,
+      "learning_rate": 1.7708851513629377e-05,
+      "loss": 0.2791,
+      "step": 4425
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.29440614581108093,
+      "learning_rate": 1.7657022311886097e-05,
+      "loss": 0.2708,
+      "step": 4430
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.3177744150161743,
+      "learning_rate": 1.7605227627299413e-05,
+      "loss": 0.2901,
+      "step": 4435
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.292756587266922,
+      "learning_rate": 1.7553467703340754e-05,
+      "loss": 0.2768,
+      "step": 4440
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.3144879937171936,
+      "learning_rate": 1.7501742783318154e-05,
+      "loss": 0.2865,
+      "step": 4445
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.3141010105609894,
+      "learning_rate": 1.745005311037508e-05,
+      "loss": 0.2727,
+      "step": 4450
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.2778961956501007,
+      "learning_rate": 1.7398398927489345e-05,
+      "loss": 0.2741,
+      "step": 4455
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.31049174070358276,
+      "learning_rate": 1.73467804774719e-05,
+      "loss": 0.2671,
+      "step": 4460
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.30341869592666626,
+      "learning_rate": 1.729519800296575e-05,
+      "loss": 0.2877,
+      "step": 4465
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.2848811149597168,
+      "learning_rate": 1.7243651746444795e-05,
+      "loss": 0.2923,
+      "step": 4470
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.3013759255409241,
+      "learning_rate": 1.7192141950212653e-05,
+      "loss": 0.2741,
+      "step": 4475
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.33438655734062195,
+      "learning_rate": 1.7140668856401588e-05,
+      "loss": 0.2814,
+      "step": 4480
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.1845710277557373,
+      "learning_rate": 1.7089232706971313e-05,
+      "loss": 0.2697,
+      "step": 4485
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.3131597936153412,
+      "learning_rate": 1.7037833743707892e-05,
+      "loss": 0.2694,
+      "step": 4490
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.2927083373069763,
+      "learning_rate": 1.6986472208222576e-05,
+      "loss": 0.2698,
+      "step": 4495
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.32785534858703613,
+      "learning_rate": 1.6935148341950692e-05,
+      "loss": 0.2734,
+      "step": 4500
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.2991541028022766,
+      "learning_rate": 1.6883862386150475e-05,
+      "loss": 0.2817,
+      "step": 4505
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.31159043312072754,
+      "learning_rate": 1.6832614581901986e-05,
+      "loss": 0.2835,
+      "step": 4510
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.31825435161590576,
+      "learning_rate": 1.6781405170105902e-05,
+      "loss": 0.2754,
+      "step": 4515
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.2784048318862915,
+      "learning_rate": 1.673023439148247e-05,
+      "loss": 0.2714,
+      "step": 4520
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.3185473382472992,
+      "learning_rate": 1.6679102486570328e-05,
+      "loss": 0.283,
+      "step": 4525
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.28926029801368713,
+      "learning_rate": 1.6628009695725346e-05,
+      "loss": 0.2691,
+      "step": 4530
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3041357100009918,
+      "learning_rate": 1.6576956259119576e-05,
+      "loss": 0.2662,
+      "step": 4535
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.290804386138916,
+      "learning_rate": 1.6525942416740043e-05,
+      "loss": 0.2599,
+      "step": 4540
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.2984979450702667,
+      "learning_rate": 1.647496840838767e-05,
+      "loss": 0.2801,
+      "step": 4545
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.2987411320209503,
+      "learning_rate": 1.642403447367612e-05,
+      "loss": 0.2744,
+      "step": 4550
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.30650317668914795,
+      "learning_rate": 1.6373140852030706e-05,
+      "loss": 0.2673,
+      "step": 4555
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.31252521276474,
+      "learning_rate": 1.6322287782687197e-05,
+      "loss": 0.2677,
+      "step": 4560
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.2668997645378113,
+      "learning_rate": 1.6271475504690792e-05,
+      "loss": 0.272,
+      "step": 4565
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.28516268730163574,
+      "learning_rate": 1.6220704256894888e-05,
+      "loss": 0.2851,
+      "step": 4570
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.2828245460987091,
+      "learning_rate": 1.616997427796005e-05,
+      "loss": 0.2816,
+      "step": 4575
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.33206111192703247,
+      "learning_rate": 1.611928580635285e-05,
+      "loss": 0.2813,
+      "step": 4580
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.32796260714530945,
+      "learning_rate": 1.6068639080344708e-05,
+      "loss": 0.2817,
+      "step": 4585
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.31739357113838196,
+      "learning_rate": 1.6018034338010857e-05,
+      "loss": 0.2736,
+      "step": 4590
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.2808535397052765,
+      "learning_rate": 1.5967471817229148e-05,
+      "loss": 0.2811,
+      "step": 4595
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.39438608288764954,
+      "learning_rate": 1.5916951755678973e-05,
+      "loss": 0.2791,
+      "step": 4600
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.3549308776855469,
+      "learning_rate": 1.5866474390840125e-05,
+      "loss": 0.274,
+      "step": 4605
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.33500468730926514,
+      "learning_rate": 1.5816039959991712e-05,
+      "loss": 0.2702,
+      "step": 4610
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.30193305015563965,
+      "learning_rate": 1.5765648700211e-05,
+      "loss": 0.2699,
+      "step": 4615
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.31212207674980164,
+      "learning_rate": 1.571530084837234e-05,
+      "loss": 0.2885,
+      "step": 4620
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.2984713315963745,
+      "learning_rate": 1.566499664114601e-05,
+      "loss": 0.275,
+      "step": 4625
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.4046083390712738,
+      "learning_rate": 1.5614736314997153e-05,
+      "loss": 0.2921,
+      "step": 4630
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.30109837651252747,
+      "learning_rate": 1.5564520106184644e-05,
+      "loss": 0.2731,
+      "step": 4635
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.298581600189209,
+      "learning_rate": 1.5514348250759936e-05,
+      "loss": 0.279,
+      "step": 4640
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.29031801223754883,
+      "learning_rate": 1.5464220984566037e-05,
+      "loss": 0.276,
+      "step": 4645
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.30218642950057983,
+      "learning_rate": 1.5414138543236316e-05,
+      "loss": 0.2803,
+      "step": 4650
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.34059423208236694,
+      "learning_rate": 1.5364101162193467e-05,
+      "loss": 0.2864,
+      "step": 4655
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.3055756986141205,
+      "learning_rate": 1.531410907664834e-05,
+      "loss": 0.2813,
+      "step": 4660
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.3241121768951416,
+      "learning_rate": 1.5264162521598892e-05,
+      "loss": 0.2827,
+      "step": 4665
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.29325103759765625,
+      "learning_rate": 1.5214261731829022e-05,
+      "loss": 0.2817,
+      "step": 4670
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.290559858083725,
+      "learning_rate": 1.5164406941907534e-05,
+      "loss": 0.2811,
+      "step": 4675
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3081018924713135,
+      "learning_rate": 1.5114598386186965e-05,
+      "loss": 0.2667,
+      "step": 4680
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.31803908944129944,
+      "learning_rate": 1.5064836298802551e-05,
+      "loss": 0.2608,
+      "step": 4685
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.2926519513130188,
+      "learning_rate": 1.5015120913671089e-05,
+      "loss": 0.2623,
+      "step": 4690
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.28197115659713745,
+      "learning_rate": 1.4965452464489815e-05,
+      "loss": 0.2753,
+      "step": 4695
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.2974069118499756,
+      "learning_rate": 1.4915831184735369e-05,
+      "loss": 0.2709,
+      "step": 4700
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.31076109409332275,
+      "learning_rate": 1.4866257307662634e-05,
+      "loss": 0.2746,
+      "step": 4705
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.3005169928073883,
+      "learning_rate": 1.4816731066303686e-05,
+      "loss": 0.2766,
+      "step": 4710
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.31110328435897827,
+      "learning_rate": 1.4767252693466665e-05,
+      "loss": 0.278,
+      "step": 4715
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.29102906584739685,
+      "learning_rate": 1.4717822421734718e-05,
+      "loss": 0.2852,
+      "step": 4720
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.29721471667289734,
+      "learning_rate": 1.466844048346484e-05,
+      "loss": 0.279,
+      "step": 4725
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.3029063642024994,
+      "learning_rate": 1.4619107110786884e-05,
+      "loss": 0.2748,
+      "step": 4730
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.29451948404312134,
+      "learning_rate": 1.456982253560235e-05,
+      "loss": 0.2715,
+      "step": 4735
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.3217031955718994,
+      "learning_rate": 1.4520586989583406e-05,
+      "loss": 0.2654,
+      "step": 4740
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.27577683329582214,
+      "learning_rate": 1.4471400704171706e-05,
+      "loss": 0.2725,
+      "step": 4745
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.28932300209999084,
+      "learning_rate": 1.4422263910577377e-05,
+      "loss": 0.2811,
+      "step": 4750
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.2884375751018524,
+      "learning_rate": 1.4373176839777896e-05,
+      "loss": 0.2662,
+      "step": 4755
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3114871680736542,
+      "learning_rate": 1.4324139722516982e-05,
+      "loss": 0.276,
+      "step": 4760
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.30992618203163147,
+      "learning_rate": 1.4275152789303575e-05,
+      "loss": 0.2836,
+      "step": 4765
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.29896673560142517,
+      "learning_rate": 1.4226216270410669e-05,
+      "loss": 0.2769,
+      "step": 4770
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.2877211570739746,
+      "learning_rate": 1.4177330395874327e-05,
+      "loss": 0.2716,
+      "step": 4775
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.30194801092147827,
+      "learning_rate": 1.4128495395492502e-05,
+      "loss": 0.272,
+      "step": 4780
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.36784809827804565,
+      "learning_rate": 1.4079711498824027e-05,
+      "loss": 0.2687,
+      "step": 4785
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.2926940619945526,
+      "learning_rate": 1.4030978935187511e-05,
+      "loss": 0.2521,
+      "step": 4790
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.2929438650608063,
+      "learning_rate": 1.3982297933660266e-05,
+      "loss": 0.2758,
+      "step": 4795
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.34613722562789917,
+      "learning_rate": 1.3933668723077198e-05,
+      "loss": 0.2796,
+      "step": 4800
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.2929108440876007,
+      "learning_rate": 1.3885091532029792e-05,
+      "loss": 0.2714,
+      "step": 4805
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.30324453115463257,
+      "learning_rate": 1.3836566588864999e-05,
+      "loss": 0.271,
+      "step": 4810
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.2864826023578644,
+      "learning_rate": 1.378809412168414e-05,
+      "loss": 0.2663,
+      "step": 4815
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.2974151372909546,
+      "learning_rate": 1.3739674358341903e-05,
+      "loss": 0.281,
+      "step": 4820
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.2970708906650543,
+      "learning_rate": 1.3691307526445191e-05,
+      "loss": 0.2856,
+      "step": 4825
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.29904627799987793,
+      "learning_rate": 1.3642993853352132e-05,
+      "loss": 0.2643,
+      "step": 4830
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.32226085662841797,
+      "learning_rate": 1.3594733566170926e-05,
+      "loss": 0.2807,
+      "step": 4835
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.2931012809276581,
+      "learning_rate": 1.3546526891758871e-05,
+      "loss": 0.2762,
+      "step": 4840
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.32732272148132324,
+      "learning_rate": 1.3498374056721197e-05,
+      "loss": 0.2655,
+      "step": 4845
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.2935151159763336,
+      "learning_rate": 1.3450275287410102e-05,
+      "loss": 0.2697,
+      "step": 4850
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.3281582295894623,
+      "learning_rate": 1.3402230809923586e-05,
+      "loss": 0.2717,
+      "step": 4855
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.3014606237411499,
+      "learning_rate": 1.3354240850104476e-05,
+      "loss": 0.276,
+      "step": 4860
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.31190937757492065,
+      "learning_rate": 1.3306305633539323e-05,
+      "loss": 0.2733,
+      "step": 4865
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.3211314380168915,
+      "learning_rate": 1.3258425385557322e-05,
+      "loss": 0.2753,
+      "step": 4870
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.3232608735561371,
+      "learning_rate": 1.3210600331229317e-05,
+      "loss": 0.2745,
+      "step": 4875
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.27900564670562744,
+      "learning_rate": 1.316283069536665e-05,
+      "loss": 0.2801,
+      "step": 4880
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.31238263845443726,
+      "learning_rate": 1.3115116702520219e-05,
+      "loss": 0.2741,
+      "step": 4885
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.2842240035533905,
+      "learning_rate": 1.3067458576979304e-05,
+      "loss": 0.2732,
+      "step": 4890
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.30593302845954895,
+      "learning_rate": 1.3019856542770607e-05,
+      "loss": 0.2755,
+      "step": 4895
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.32800737023353577,
+      "learning_rate": 1.297231082365714e-05,
+      "loss": 0.2695,
+      "step": 4900
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 7365,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 100,
+  "total_flos": 5.32027409351639e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}