diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5059 @@
+{
+  "best_metric": 0.928910891089109,
+  "best_model_checkpoint": "swin-base-patch4-window7-224-food101-16-7/checkpoint-8281",
+  "epoch": 6.995564941921859,
+  "eval_steps": 500,
+  "global_step": 8281,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.031363088057902e-07,
+      "loss": 4.6432,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2062726176115803e-06,
+      "loss": 4.649,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8094089264173706e-06,
+      "loss": 4.6442,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4125452352231606e-06,
+      "loss": 4.6446,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.015681544028951e-06,
+      "loss": 4.6096,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.618817852834741e-06,
+      "loss": 4.6262,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.221954161640531e-06,
+      "loss": 4.6307,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.825090470446321e-06,
+      "loss": 4.6098,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.428226779252111e-06,
+      "loss": 4.5854,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6.031363088057902e-06,
+      "loss": 4.5439,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.634499396863691e-06,
+      "loss": 4.5292,
+      "step": 110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.237635705669482e-06,
+      "loss": 4.5046,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.840772014475271e-06,
+      "loss": 4.4373,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.443908323281062e-06,
+      "loss": 4.3969,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.047044632086853e-06,
+      "loss": 4.3214,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.650180940892643e-06,
+      "loss": 4.2851,
+      "step": 160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.0253317249698432e-05,
+      "loss": 4.2182,
+      "step": 170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.0856453558504221e-05,
+      "loss": 4.1271,
+      "step": 180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.1459589867310012e-05,
+      "loss": 3.983,
+      "step": 190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.2062726176115804e-05,
+      "loss": 3.8543,
+      "step": 200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.2665862484921593e-05,
+      "loss": 3.6776,
+      "step": 210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3268998793727382e-05,
+      "loss": 3.5393,
+      "step": 220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.3872135102533174e-05,
+      "loss": 3.3374,
+      "step": 230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.4475271411338965e-05,
+      "loss": 3.1107,
+      "step": 240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.5078407720144752e-05,
+      "loss": 2.9508,
+      "step": 250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5681544028950542e-05,
+      "loss": 2.9108,
+      "step": 260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.6284680337756335e-05,
+      "loss": 2.7387,
+      "step": 270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.6887816646562124e-05,
+      "loss": 2.5738,
+      "step": 280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7490952955367913e-05,
+      "loss": 2.3587,
+      "step": 290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8094089264173706e-05,
+      "loss": 2.1593,
+      "step": 300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8697225572979492e-05,
+      "loss": 2.1114,
+      "step": 310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9300361881785285e-05,
+      "loss": 2.0665,
+      "step": 320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9903498190591075e-05,
+      "loss": 1.8669,
+      "step": 330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.0506634499396864e-05,
+      "loss": 1.7142,
+      "step": 340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1109770808202657e-05,
+      "loss": 1.8547,
+      "step": 350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1712907117008443e-05,
+      "loss": 1.6322,
+      "step": 360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.2316043425814236e-05,
+      "loss": 1.5511,
+      "step": 370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.2919179734620025e-05,
+      "loss": 1.5839,
+      "step": 380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.3522316043425814e-05,
+      "loss": 1.6223,
+      "step": 390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4125452352231607e-05,
+      "loss": 1.5052,
+      "step": 400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4728588661037397e-05,
+      "loss": 1.3546,
+      "step": 410
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.5331724969843186e-05,
+      "loss": 1.4439,
+      "step": 420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.5934861278648975e-05,
+      "loss": 1.407,
+      "step": 430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6537997587454765e-05,
+      "loss": 1.3609,
+      "step": 440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.7141133896260558e-05,
+      "loss": 1.2596,
+      "step": 450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.7744270205066347e-05,
+      "loss": 1.3737,
+      "step": 460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.8347406513872137e-05,
+      "loss": 1.3378,
+      "step": 470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.895054282267793e-05,
+      "loss": 1.3394,
+      "step": 480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.955367913148372e-05,
+      "loss": 1.2723,
+      "step": 490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0156815440289505e-05,
+      "loss": 1.1621,
+      "step": 500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0759951749095294e-05,
+      "loss": 1.2241,
+      "step": 510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1363088057901084e-05,
+      "loss": 1.1729,
+      "step": 520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.196622436670687e-05,
+      "loss": 1.0635,
+      "step": 530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.256936067551267e-05,
+      "loss": 1.1714,
+      "step": 540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.317249698431846e-05,
+      "loss": 1.1109,
+      "step": 550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.377563329312425e-05,
+      "loss": 1.2177,
+      "step": 560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.437876960193004e-05,
+      "loss": 1.0682,
+      "step": 570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.498190591073583e-05,
+      "loss": 1.1035,
+      "step": 580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.558504221954162e-05,
+      "loss": 1.2274,
+      "step": 590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 3.618817852834741e-05,
+      "loss": 1.0473,
+      "step": 600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.6791314837153195e-05,
+      "loss": 1.1054,
+      "step": 610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 3.7394451145958985e-05,
+      "loss": 1.0663,
+      "step": 620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 3.7997587454764774e-05,
+      "loss": 1.1639,
+      "step": 630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 3.860072376357057e-05,
+      "loss": 1.0283,
+      "step": 640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 3.920386007237636e-05,
+      "loss": 1.1588,
+      "step": 650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 3.980699638118215e-05,
+      "loss": 1.1216,
+      "step": 660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.041013268998794e-05,
+      "loss": 1.0618,
+      "step": 670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.101326899879373e-05,
+      "loss": 1.011,
+      "step": 680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1616405307599524e-05,
+      "loss": 1.0517,
+      "step": 690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2219541616405313e-05,
+      "loss": 1.1075,
+      "step": 700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.28226779252111e-05,
+      "loss": 1.0016,
+      "step": 710
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.3425814234016886e-05,
+      "loss": 1.0057,
+      "step": 720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.4028950542822675e-05,
+      "loss": 1.0903,
+      "step": 730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.463208685162847e-05,
+      "loss": 1.0357,
+      "step": 740
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.523522316043426e-05,
+      "loss": 1.1443,
+      "step": 750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.583835946924005e-05,
+      "loss": 1.0243,
+      "step": 760
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.644149577804584e-05,
+      "loss": 1.0206,
+      "step": 770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.704463208685163e-05,
+      "loss": 0.9825,
+      "step": 780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.7647768395657425e-05,
+      "loss": 1.1923,
+      "step": 790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8250904704463214e-05,
+      "loss": 0.9917,
+      "step": 800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8854041013269004e-05,
+      "loss": 1.0633,
+      "step": 810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.945717732207479e-05,
+      "loss": 1.015,
+      "step": 820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.999329039184112e-05,
+      "loss": 1.0894,
+      "step": 830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.9926194310252285e-05,
+      "loss": 0.9793,
+      "step": 840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.985909822866345e-05,
+      "loss": 0.949,
+      "step": 850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.979200214707461e-05,
+      "loss": 1.0262,
+      "step": 860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.9724906065485776e-05,
+      "loss": 0.9514,
+      "step": 870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.965780998389694e-05,
+      "loss": 1.0118,
+      "step": 880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.95907139023081e-05,
+      "loss": 1.0958,
+      "step": 890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9523617820719274e-05,
+      "loss": 1.0222,
+      "step": 900
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.945652173913044e-05,
+      "loss": 0.9155,
+      "step": 910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.93894256575416e-05,
+      "loss": 0.9821,
+      "step": 920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.932232957595277e-05,
+      "loss": 0.9444,
+      "step": 930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.9255233494363935e-05,
+      "loss": 1.0304,
+      "step": 940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.91881374127751e-05,
+      "loss": 0.9344,
+      "step": 950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.912104133118626e-05,
+      "loss": 0.8943,
+      "step": 960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.9053945249597426e-05,
+      "loss": 1.2308,
+      "step": 970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.898684916800859e-05,
+      "loss": 1.0398,
+      "step": 980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.891975308641975e-05,
+      "loss": 0.9748,
+      "step": 990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.885265700483092e-05,
+      "loss": 0.9154,
+      "step": 1000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.878556092324209e-05,
+      "loss": 0.8873,
+      "step": 1010
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.871846484165325e-05,
+      "loss": 0.8687,
+      "step": 1020
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8651368760064414e-05,
+      "loss": 0.9199,
+      "step": 1030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.858427267847558e-05,
+      "loss": 0.8948,
+      "step": 1040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.851717659688674e-05,
+      "loss": 0.9124,
+      "step": 1050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8450080515297905e-05,
+      "loss": 0.8505,
+      "step": 1060
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.838298443370907e-05,
+      "loss": 0.8998,
+      "step": 1070
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.831588835212024e-05,
+      "loss": 0.979,
+      "step": 1080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.82487922705314e-05,
+      "loss": 0.8806,
+      "step": 1090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.8181696188942566e-05,
+      "loss": 0.9287,
+      "step": 1100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.811460010735374e-05,
+      "loss": 0.9486,
+      "step": 1110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.80475040257649e-05,
+      "loss": 0.9874,
+      "step": 1120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.7980407944176064e-05,
+      "loss": 0.8716,
+      "step": 1130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.791331186258723e-05,
+      "loss": 0.8683,
+      "step": 1140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.784621578099839e-05,
+      "loss": 0.9064,
+      "step": 1150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7779119699409555e-05,
+      "loss": 0.8832,
+      "step": 1160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7712023617820725e-05,
+      "loss": 0.8954,
+      "step": 1170
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.764492753623189e-05,
+      "loss": 0.8681,
+      "step": 1180
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8731089108910891,
+      "eval_loss": 0.44370436668395996,
+      "eval_runtime": 407.6981,
+      "eval_samples_per_second": 61.933,
+      "eval_steps_per_second": 3.873,
+      "step": 1183
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.757783145464305e-05,
+      "loss": 0.8442,
+      "step": 1190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.7510735373054216e-05,
+      "loss": 0.802,
+      "step": 1200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.744363929146538e-05,
+      "loss": 0.818,
+      "step": 1210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.7376543209876543e-05,
+      "loss": 0.798,
+      "step": 1220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.730944712828771e-05,
+      "loss": 0.8002,
+      "step": 1230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.724235104669887e-05,
+      "loss": 0.8296,
+      "step": 1240
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.717525496511004e-05,
+      "loss": 0.7832,
+      "step": 1250
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.7108158883521205e-05,
+      "loss": 0.8956,
+      "step": 1260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.704106280193237e-05,
+      "loss": 0.8133,
+      "step": 1270
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.697396672034353e-05,
+      "loss": 0.8469,
+      "step": 1280
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.6906870638754696e-05,
+      "loss": 0.8012,
+      "step": 1290
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.6839774557165866e-05,
+      "loss": 0.737,
+      "step": 1300
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.677267847557703e-05,
+      "loss": 0.8695,
+      "step": 1310
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.670558239398819e-05,
+      "loss": 0.7718,
+      "step": 1320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.663848631239936e-05,
+      "loss": 0.7631,
+      "step": 1330
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.657139023081053e-05,
+      "loss": 0.8263,
+      "step": 1340
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.650429414922169e-05,
+      "loss": 0.8393,
+      "step": 1350
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.6437198067632854e-05,
+      "loss": 0.8479,
+      "step": 1360
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.637010198604402e-05,
+      "loss": 0.8147,
+      "step": 1370
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.630300590445518e-05,
+      "loss": 0.7623,
+      "step": 1380
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.6235909822866345e-05,
+      "loss": 0.7504,
+      "step": 1390
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.616881374127751e-05,
+      "loss": 0.7215,
+      "step": 1400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.610171765968867e-05,
+      "loss": 0.8449,
+      "step": 1410
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.603462157809984e-05,
+      "loss": 0.7671,
+      "step": 1420
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.5967525496511007e-05,
+      "loss": 0.753,
+      "step": 1430
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.590042941492217e-05,
+      "loss": 0.8065,
+      "step": 1440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.7258,
+      "step": 1450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.57662372517445e-05,
+      "loss": 0.7194,
+      "step": 1460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.569914117015566e-05,
+      "loss": 0.8929,
+      "step": 1470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.5632045088566825e-05,
+      "loss": 0.8532,
+      "step": 1480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.5564949006977995e-05,
+      "loss": 0.8744,
+      "step": 1490
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.549785292538916e-05,
+      "loss": 0.7241,
+      "step": 1500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.543075684380033e-05,
+      "loss": 0.6477,
+      "step": 1510
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.536366076221149e-05,
+      "loss": 0.7678,
+      "step": 1520
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.5296564680622656e-05,
+      "loss": 0.7956,
+      "step": 1530
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.522946859903382e-05,
+      "loss": 0.7876,
+      "step": 1540
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.5162372517444984e-05,
+      "loss": 0.7801,
+      "step": 1550
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.509527643585615e-05,
+      "loss": 0.8688,
+      "step": 1560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.502818035426731e-05,
+      "loss": 0.7611,
+      "step": 1570
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.4961084272678474e-05,
+      "loss": 0.752,
+      "step": 1580
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.4893988191089645e-05,
+      "loss": 0.8414,
+      "step": 1590
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.482689210950081e-05,
+      "loss": 0.7291,
+      "step": 1600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.475979602791197e-05,
+      "loss": 0.8054,
+      "step": 1610
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.4692699946323136e-05,
+      "loss": 0.8184,
+      "step": 1620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.46256038647343e-05,
+      "loss": 0.7482,
+      "step": 1630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.455850778314546e-05,
+      "loss": 0.8088,
+      "step": 1640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.449141170155663e-05,
+      "loss": 0.8412,
+      "step": 1650
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.442431561996779e-05,
+      "loss": 0.7775,
+      "step": 1660
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.435721953837896e-05,
+      "loss": 0.7194,
+      "step": 1670
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.429012345679013e-05,
+      "loss": 0.7445,
+      "step": 1680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.4223027375201295e-05,
+      "loss": 0.8271,
+      "step": 1690
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.415593129361246e-05,
+      "loss": 0.7927,
+      "step": 1700
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.408883521202362e-05,
+      "loss": 0.7354,
+      "step": 1710
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.4021739130434786e-05,
+      "loss": 0.863,
+      "step": 1720
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.395464304884595e-05,
+      "loss": 0.7567,
+      "step": 1730
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.388754696725711e-05,
+      "loss": 0.6566,
+      "step": 1740
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.3820450885668276e-05,
+      "loss": 0.8311,
+      "step": 1750
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.375335480407945e-05,
+      "loss": 0.7401,
+      "step": 1760
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.368625872249061e-05,
+      "loss": 0.7517,
+      "step": 1770
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.3619162640901774e-05,
+      "loss": 0.7797,
+      "step": 1780
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.355206655931294e-05,
+      "loss": 0.7237,
+      "step": 1790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.34849704777241e-05,
+      "loss": 0.7063,
+      "step": 1800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.3417874396135265e-05,
+      "loss": 0.7611,
+      "step": 1810
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.335077831454643e-05,
+      "loss": 0.6942,
+      "step": 1820
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.328368223295759e-05,
+      "loss": 0.7611,
+      "step": 1830
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.321658615136876e-05,
+      "loss": 0.7866,
+      "step": 1840
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.3149490069779926e-05,
+      "loss": 0.6945,
+      "step": 1850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.3082393988191097e-05,
+      "loss": 0.6864,
+      "step": 1860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.301529790660226e-05,
+      "loss": 0.8234,
+      "step": 1870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2948201825013424e-05,
+      "loss": 0.6629,
+      "step": 1880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.288110574342459e-05,
+      "loss": 0.7636,
+      "step": 1890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.281400966183575e-05,
+      "loss": 0.8126,
+      "step": 1900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.2746913580246915e-05,
+      "loss": 0.7373,
+      "step": 1910
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.267981749865808e-05,
+      "loss": 0.7383,
+      "step": 1920
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.261272141706925e-05,
+      "loss": 0.7185,
+      "step": 1930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.254562533548041e-05,
+      "loss": 0.7524,
+      "step": 1940
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.2478529253891576e-05,
+      "loss": 0.6379,
+      "step": 1950
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.241143317230274e-05,
+      "loss": 0.7145,
+      "step": 1960
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.23443370907139e-05,
+      "loss": 0.7139,
+      "step": 1970
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.227724100912507e-05,
+      "loss": 0.9402,
+      "step": 1980
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.221014492753623e-05,
+      "loss": 0.8974,
+      "step": 1990
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.2143048845947394e-05,
+      "loss": 0.7268,
+      "step": 2000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.2075952764358564e-05,
+      "loss": 0.6971,
+      "step": 2010
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.200885668276973e-05,
+      "loss": 0.6947,
+      "step": 2020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.194176060118089e-05,
+      "loss": 0.791,
+      "step": 2030
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.187466451959206e-05,
+      "loss": 0.6615,
+      "step": 2040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1807568438003226e-05,
+      "loss": 0.6294,
+      "step": 2050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.174047235641439e-05,
+      "loss": 0.7388,
+      "step": 2060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.167337627482555e-05,
+      "loss": 0.6579,
+      "step": 2070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1606280193236717e-05,
+      "loss": 0.7255,
+      "step": 2080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.153918411164788e-05,
+      "loss": 0.7341,
+      "step": 2090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.147208803005905e-05,
+      "loss": 0.7142,
+      "step": 2100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1404991948470214e-05,
+      "loss": 0.7042,
+      "step": 2110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.133789586688138e-05,
+      "loss": 0.7399,
+      "step": 2120
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.127079978529254e-05,
+      "loss": 0.8365,
+      "step": 2130
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.1203703703703705e-05,
+      "loss": 0.7981,
+      "step": 2140
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.113660762211487e-05,
+      "loss": 0.7135,
+      "step": 2150
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.106951154052603e-05,
+      "loss": 0.7685,
+      "step": 2160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.1002415458937196e-05,
+      "loss": 0.6902,
+      "step": 2170
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0935319377348366e-05,
+      "loss": 0.6474,
+      "step": 2180
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.086822329575953e-05,
+      "loss": 0.7644,
+      "step": 2190
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.0801127214170694e-05,
+      "loss": 0.7102,
+      "step": 2200
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.073403113258186e-05,
+      "loss": 0.7549,
+      "step": 2210
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.066693505099302e-05,
+      "loss": 0.7211,
+      "step": 2220
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.059983896940419e-05,
+      "loss": 0.794,
+      "step": 2230
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.0532742887815355e-05,
+      "loss": 0.5939,
+      "step": 2240
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.046564680622652e-05,
+      "loss": 0.7167,
+      "step": 2250
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.039855072463768e-05,
+      "loss": 0.7734,
+      "step": 2260
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.033145464304885e-05,
+      "loss": 0.7067,
+      "step": 2270
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.0264358561460016e-05,
+      "loss": 0.7067,
+      "step": 2280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.019726247987118e-05,
+      "loss": 0.7468,
+      "step": 2290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.0130166398282343e-05,
+      "loss": 0.7585,
+      "step": 2300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 4.006307031669351e-05,
+      "loss": 0.6175,
+      "step": 2310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.999597423510467e-05,
+      "loss": 0.7804,
+      "step": 2320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.9928878153515834e-05,
+      "loss": 0.7373,
+      "step": 2330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.9861782071927e-05,
+      "loss": 0.7641,
+      "step": 2340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.979468599033817e-05,
+      "loss": 0.691,
+      "step": 2350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.972758990874933e-05,
+      "loss": 0.6919,
+      "step": 2360
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9038415841584159,
+      "eval_loss": 0.33233964443206787,
+      "eval_runtime": 409.4477,
+      "eval_samples_per_second": 61.668,
+      "eval_steps_per_second": 3.856,
+      "step": 2367
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.9660493827160496e-05,
+      "loss": 0.6591,
+      "step": 2370
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.959339774557166e-05,
+      "loss": 0.5642,
+      "step": 2380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.952630166398282e-05,
+      "loss": 0.6206,
+      "step": 2390
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9459205582393986e-05,
+      "loss": 0.6085,
+      "step": 2400
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.939210950080515e-05,
+      "loss": 0.5174,
+      "step": 2410
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.932501341921632e-05,
+      "loss": 0.6653,
+      "step": 2420
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.9257917337627484e-05,
+      "loss": 0.6464,
+      "step": 2430
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.9190821256038654e-05,
+      "loss": 0.641,
+      "step": 2440
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.912372517444982e-05,
+      "loss": 0.56,
+      "step": 2450
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.905662909286098e-05,
+      "loss": 0.6378,
+      "step": 2460
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.8989533011272145e-05,
+      "loss": 0.6674,
+      "step": 2470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.892243692968331e-05,
+      "loss": 0.6061,
+      "step": 2480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.885534084809447e-05,
+      "loss": 0.7563,
+      "step": 2490
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.8788244766505636e-05,
+      "loss": 0.6304,
+      "step": 2500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.87211486849168e-05,
+      "loss": 0.5949,
+      "step": 2510
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.865405260332797e-05,
+      "loss": 0.6201,
+      "step": 2520
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.8586956521739134e-05,
+      "loss": 0.5424,
+      "step": 2530
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.85198604401503e-05,
+      "loss": 0.5528,
+      "step": 2540
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.845276435856146e-05,
+      "loss": 0.582,
+      "step": 2550
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8385668276972625e-05,
+      "loss": 0.5896,
+      "step": 2560
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.831857219538379e-05,
+      "loss": 0.7052,
+      "step": 2570
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.825147611379495e-05,
+      "loss": 0.5849,
+      "step": 2580
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.8184380032206116e-05,
+      "loss": 0.6328,
+      "step": 2590
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.8117283950617286e-05,
+      "loss": 0.606,
+      "step": 2600
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.8050187869028456e-05,
+      "loss": 0.5958,
+      "step": 2610
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.798309178743962e-05,
+      "loss": 0.6429,
+      "step": 2620
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.7915995705850784e-05,
+      "loss": 0.6674,
+      "step": 2630
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.784889962426195e-05,
+      "loss": 0.6623,
+      "step": 2640
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.778180354267311e-05,
+      "loss": 0.543,
+      "step": 2650
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.7714707461084274e-05,
+      "loss": 0.584,
+      "step": 2660
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.764761137949544e-05,
+      "loss": 0.6371,
+      "step": 2670
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.75805152979066e-05,
+      "loss": 0.6221,
+      "step": 2680
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.751341921631777e-05,
+      "loss": 0.5602,
+      "step": 2690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.7446323134728936e-05,
+      "loss": 0.6468,
+      "step": 2700
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.73792270531401e-05,
+      "loss": 0.6885,
+      "step": 2710
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.731213097155126e-05,
+      "loss": 0.5594,
+      "step": 2720
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.724503488996243e-05,
+      "loss": 0.5833,
+      "step": 2730
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.717793880837359e-05,
+      "loss": 0.5923,
+      "step": 2740
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.7110842726784754e-05,
+      "loss": 0.6079,
+      "step": 2750
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.704374664519592e-05,
+      "loss": 0.5498,
+      "step": 2760
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.697665056360709e-05,
+      "loss": 0.5448,
+      "step": 2770
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.690955448201825e-05,
+      "loss": 0.7015,
+      "step": 2780
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.684245840042942e-05,
+      "loss": 0.654,
+      "step": 2790
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.6775362318840586e-05,
+      "loss": 0.724,
+      "step": 2800
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.670826623725175e-05,
+      "loss": 0.6532,
+      "step": 2810
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.664117015566291e-05,
+      "loss": 0.6913,
+      "step": 2820
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.6574074074074076e-05,
+      "loss": 0.5872,
+      "step": 2830
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 3.650697799248524e-05,
+      "loss": 0.6574,
+      "step": 2840
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.6439881910896404e-05,
+      "loss": 0.5782,
+      "step": 2850
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.6372785829307574e-05,
+      "loss": 0.5749,
+      "step": 2860
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.630568974771874e-05,
+      "loss": 0.598,
+      "step": 2870
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.62385936661299e-05,
+      "loss": 0.6566,
+      "step": 2880
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.6171497584541065e-05,
+      "loss": 0.5799,
+      "step": 2890
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.610440150295223e-05,
+      "loss": 0.5143,
+      "step": 2900
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.603730542136339e-05,
+      "loss": 0.6099,
+      "step": 2910
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.5970209339774556e-05,
+      "loss": 0.5819,
+      "step": 2920
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.590311325818572e-05,
+      "loss": 0.689,
+      "step": 2930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.583601717659689e-05,
+      "loss": 0.5902,
+      "step": 2940
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.5768921095008053e-05,
+      "loss": 0.6991,
+      "step": 2950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.570182501341922e-05,
+      "loss": 0.6297,
+      "step": 2960
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.563472893183038e-05,
+      "loss": 0.5992,
+      "step": 2970
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.556763285024155e-05,
+      "loss": 0.5836,
+      "step": 2980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.5500536768652715e-05,
+      "loss": 0.6248,
+      "step": 2990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.543344068706388e-05,
+      "loss": 0.6381,
+      "step": 3000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.536634460547504e-05,
+      "loss": 0.6116,
+      "step": 3010
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.5299248523886206e-05,
+      "loss": 0.5965,
+      "step": 3020
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.523215244229737e-05,
+      "loss": 0.6187,
+      "step": 3030
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.516505636070854e-05,
+      "loss": 0.599,
+      "step": 3040
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.50979602791197e-05,
+      "loss": 0.5887,
+      "step": 3050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.503086419753087e-05,
+      "loss": 0.6477,
+      "step": 3060
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.496376811594203e-05,
+      "loss": 0.569,
+      "step": 3070
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.4896672034353194e-05,
+      "loss": 0.5754,
+      "step": 3080
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.482957595276436e-05,
+      "loss": 0.6396,
+      "step": 3090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.476247987117552e-05,
+      "loss": 0.6094,
+      "step": 3100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.469538378958669e-05,
+      "loss": 0.6197,
+      "step": 3110
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.4628287707997855e-05,
+      "loss": 0.6004,
+      "step": 3120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.456119162640902e-05,
+      "loss": 0.5206,
+      "step": 3130
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.449409554482018e-05,
+      "loss": 0.5334,
+      "step": 3140
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4426999463231346e-05,
+      "loss": 0.6192,
+      "step": 3150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.4359903381642517e-05,
+      "loss": 0.609,
+      "step": 3160
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.429280730005368e-05,
+      "loss": 0.6226,
+      "step": 3170
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.4225711218464844e-05,
+      "loss": 0.6571,
+      "step": 3180
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.415861513687601e-05,
+      "loss": 0.5952,
+      "step": 3190
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.409151905528717e-05,
+      "loss": 0.5891,
+      "step": 3200
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.402442297369834e-05,
+      "loss": 0.4672,
+      "step": 3210
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.3957326892109505e-05,
+      "loss": 0.5808,
+      "step": 3220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.389023081052067e-05,
+      "loss": 0.6208,
+      "step": 3230
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.382313472893183e-05,
+      "loss": 0.6188,
+      "step": 3240
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.3756038647342996e-05,
+      "loss": 0.6743,
+      "step": 3250
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.368894256575416e-05,
+      "loss": 0.573,
+      "step": 3260
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.362184648416532e-05,
+      "loss": 0.5899,
+      "step": 3270
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.3554750402576494e-05,
+      "loss": 0.5871,
+      "step": 3280
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.348765432098766e-05,
+      "loss": 0.5499,
+      "step": 3290
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.342055823939882e-05,
+      "loss": 0.605,
+      "step": 3300
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3353462157809984e-05,
+      "loss": 0.6665,
+      "step": 3310
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.328636607622115e-05,
+      "loss": 0.573,
+      "step": 3320
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.321926999463231e-05,
+      "loss": 0.6375,
+      "step": 3330
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.3152173913043475e-05,
+      "loss": 0.599,
+      "step": 3340
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.3085077831454646e-05,
+      "loss": 0.6503,
+      "step": 3350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.301798174986581e-05,
+      "loss": 0.5831,
+      "step": 3360
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.295088566827697e-05,
+      "loss": 0.6334,
+      "step": 3370
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.2883789586688143e-05,
+      "loss": 0.557,
+      "step": 3380
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.281669350509931e-05,
+      "loss": 0.5436,
+      "step": 3390
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.274959742351047e-05,
+      "loss": 0.4956,
+      "step": 3400
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.2682501341921634e-05,
+      "loss": 0.547,
+      "step": 3410
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.26154052603328e-05,
+      "loss": 0.6534,
+      "step": 3420
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.254830917874396e-05,
+      "loss": 0.6153,
+      "step": 3430
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.2481213097155125e-05,
+      "loss": 0.5774,
+      "step": 3440
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.241411701556629e-05,
+      "loss": 0.5621,
+      "step": 3450
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.234702093397746e-05,
+      "loss": 0.6906,
+      "step": 3460
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.227992485238862e-05,
+      "loss": 0.5393,
+      "step": 3470
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.2212828770799786e-05,
+      "loss": 0.4736,
+      "step": 3480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.214573268921095e-05,
+      "loss": 0.5821,
+      "step": 3490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.2078636607622114e-05,
+      "loss": 0.6094,
+      "step": 3500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.201154052603328e-05,
+      "loss": 0.6534,
+      "step": 3510
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.5601,
+      "step": 3520
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.187734836285561e-05,
+      "loss": 0.6296,
+      "step": 3530
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1810252281266775e-05,
+      "loss": 0.623,
+      "step": 3540
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1743156199677945e-05,
+      "loss": 0.4668,
+      "step": 3550
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9157623762376238,
+      "eval_loss": 0.2927907705307007,
+      "eval_runtime": 406.2996,
+      "eval_samples_per_second": 62.146,
+      "eval_steps_per_second": 3.886,
+      "step": 3551
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.167606011808911e-05,
+      "loss": 0.4494,
+      "step": 3560
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.160896403650027e-05,
+      "loss": 0.5114,
+      "step": 3570
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.1541867954911436e-05,
+      "loss": 0.5162,
+      "step": 3580
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.14747718733226e-05,
+      "loss": 0.4549,
+      "step": 3590
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.1407675791733763e-05,
+      "loss": 0.5093,
+      "step": 3600
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.134057971014493e-05,
+      "loss": 0.5832,
+      "step": 3610
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.127348362855609e-05,
+      "loss": 0.5485,
+      "step": 3620
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.120638754696726e-05,
+      "loss": 0.4862,
+      "step": 3630
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.1139291465378425e-05,
+      "loss": 0.579,
+      "step": 3640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.107219538378959e-05,
+      "loss": 0.5122,
+      "step": 3650
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.100509930220075e-05,
+      "loss": 0.5398,
+      "step": 3660
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.0938003220611916e-05,
+      "loss": 0.5274,
+      "step": 3670
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.087090713902308e-05,
+      "loss": 0.4969,
+      "step": 3680
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.080381105743424e-05,
+      "loss": 0.4823,
+      "step": 3690
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.073671497584541e-05,
+      "loss": 0.496,
+      "step": 3700
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.066961889425658e-05,
+      "loss": 0.3574,
+      "step": 3710
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.060252281266775e-05,
+      "loss": 0.5095,
+      "step": 3720
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.053542673107891e-05,
+      "loss": 0.5287,
+      "step": 3730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.0468330649490074e-05,
+      "loss": 0.5009,
+      "step": 3740
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.0401234567901238e-05,
+      "loss": 0.5129,
+      "step": 3750
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.0334138486312402e-05,
+      "loss": 0.4764,
+      "step": 3760
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.0267042404723565e-05,
+      "loss": 0.5205,
+      "step": 3770
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.0199946323134732e-05,
+      "loss": 0.497,
+      "step": 3780
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0132850241545896e-05,
+      "loss": 0.5178,
+      "step": 3790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.006575415995706e-05,
+      "loss": 0.4867,
+      "step": 3800
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.9998658078368223e-05,
+      "loss": 0.4379,
+      "step": 3810
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.993156199677939e-05,
+      "loss": 0.5452,
+      "step": 3820
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.9864465915190554e-05,
+      "loss": 0.4547,
+      "step": 3830
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.9797369833601717e-05,
+      "loss": 0.5176,
+      "step": 3840
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.973027375201288e-05,
+      "loss": 0.5673,
+      "step": 3850
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.9663177670424048e-05,
+      "loss": 0.4695,
+      "step": 3860
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.9596081588835212e-05,
+      "loss": 0.5306,
+      "step": 3870
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.9528985507246375e-05,
+      "loss": 0.5388,
+      "step": 3880
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.946188942565754e-05,
+      "loss": 0.5403,
+      "step": 3890
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.9394793344068706e-05,
+      "loss": 0.4351,
+      "step": 3900
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.9327697262479876e-05,
+      "loss": 0.5779,
+      "step": 3910
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.926060118089104e-05,
+      "loss": 0.396,
+      "step": 3920
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.9193505099302204e-05,
+      "loss": 0.5509,
+      "step": 3930
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.9126409017713367e-05,
+      "loss": 0.4572,
+      "step": 3940
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.9059312936124534e-05,
+      "loss": 0.4868,
+      "step": 3950
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.8992216854535698e-05,
+      "loss": 0.4601,
+      "step": 3960
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.892512077294686e-05,
+      "loss": 0.5508,
+      "step": 3970
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.8858024691358025e-05,
+      "loss": 0.534,
+      "step": 3980
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.8790928609769192e-05,
+      "loss": 0.537,
+      "step": 3990
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.8723832528180356e-05,
+      "loss": 0.4439,
+      "step": 4000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.865673644659152e-05,
+      "loss": 0.6213,
+      "step": 4010
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.8589640365002683e-05,
+      "loss": 0.5037,
+      "step": 4020
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.852254428341385e-05,
+      "loss": 0.4869,
+      "step": 4030
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.8455448201825014e-05,
+      "loss": 0.5354,
+      "step": 4040
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.8388352120236177e-05,
+      "loss": 0.5654,
+      "step": 4050
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.832125603864734e-05,
+      "loss": 0.4897,
+      "step": 4060
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.8254159957058508e-05,
+      "loss": 0.5188,
+      "step": 4070
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.818706387546967e-05,
+      "loss": 0.5579,
+      "step": 4080
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.8119967793880842e-05,
+      "loss": 0.4569,
+      "step": 4090
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.8052871712292006e-05,
+      "loss": 0.5914,
+      "step": 4100
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.798577563070317e-05,
+      "loss": 0.4699,
+      "step": 4110
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.7918679549114336e-05,
+      "loss": 0.5459,
+      "step": 4120
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.78515834675255e-05,
+      "loss": 0.4761,
+      "step": 4130
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.7784487385936663e-05,
+      "loss": 0.5127,
+      "step": 4140
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.7717391304347827e-05,
+      "loss": 0.5452,
+      "step": 4150
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.7650295222758994e-05,
+      "loss": 0.5532,
+      "step": 4160
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.7583199141170158e-05,
+      "loss": 0.5435,
+      "step": 4170
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.751610305958132e-05,
+      "loss": 0.5191,
+      "step": 4180
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.7449006977992485e-05,
+      "loss": 0.5271,
+      "step": 4190
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.7381910896403652e-05,
+      "loss": 0.4802,
+      "step": 4200
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.7314814814814816e-05,
+      "loss": 0.5373,
+      "step": 4210
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.724771873322598e-05,
+      "loss": 0.5416,
+      "step": 4220
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.7180622651637143e-05,
+      "loss": 0.4691,
+      "step": 4230
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 2.711352657004831e-05,
+      "loss": 0.517,
+      "step": 4240
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 2.7046430488459473e-05,
+      "loss": 0.548,
+      "step": 4250
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 2.6979334406870637e-05,
+      "loss": 0.3983,
+      "step": 4260
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 2.69122383252818e-05,
+      "loss": 0.5338,
+      "step": 4270
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.684514224369297e-05,
+      "loss": 0.5325,
+      "step": 4280
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 2.6778046162104138e-05,
+      "loss": 0.4098,
+      "step": 4290
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 2.6710950080515302e-05,
+      "loss": 0.4194,
+      "step": 4300
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 2.6643853998926465e-05,
+      "loss": 0.5277,
+      "step": 4310
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 2.657675791733763e-05,
+      "loss": 0.4749,
+      "step": 4320
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 2.6509661835748796e-05,
+      "loss": 0.4769,
+      "step": 4330
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.644256575415996e-05,
+      "loss": 0.5218,
+      "step": 4340
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 2.6375469672571123e-05,
+      "loss": 0.5074,
+      "step": 4350
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 2.6308373590982287e-05,
+      "loss": 0.5536,
+      "step": 4360
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 2.6241277509393454e-05,
+      "loss": 0.5381,
+      "step": 4370
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.6174181427804617e-05,
+      "loss": 0.4784,
+      "step": 4380
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.610708534621578e-05,
+      "loss": 0.5196,
+      "step": 4390
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.6039989264626945e-05,
+      "loss": 0.5772,
+      "step": 4400
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.5972893183038112e-05,
+      "loss": 0.5499,
+      "step": 4410
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.5905797101449275e-05,
+      "loss": 0.4025,
+      "step": 4420
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.583870101986044e-05,
+      "loss": 0.4915,
+      "step": 4430
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.5771604938271603e-05,
+      "loss": 0.4872,
+      "step": 4440
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 2.570450885668277e-05,
+      "loss": 0.4751,
+      "step": 4450
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.5637412775093933e-05,
+      "loss": 0.5216,
+      "step": 4460
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.5570316693505104e-05,
+      "loss": 0.5404,
+      "step": 4470
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.5503220611916267e-05,
+      "loss": 0.5335,
+      "step": 4480
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.543612453032743e-05,
+      "loss": 0.525,
+      "step": 4490
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.5369028448738598e-05,
+      "loss": 0.4855,
+      "step": 4500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.530193236714976e-05,
+      "loss": 0.455,
+      "step": 4510
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.5234836285560925e-05,
+      "loss": 0.5023,
+      "step": 4520
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.516774020397209e-05,
+      "loss": 0.4287,
+      "step": 4530
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.5100644122383256e-05,
+      "loss": 0.483,
+      "step": 4540
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.503354804079442e-05,
+      "loss": 0.4484,
+      "step": 4550
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.4966451959205583e-05,
+      "loss": 0.6273,
+      "step": 4560
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.4899355877616747e-05,
+      "loss": 0.4636,
+      "step": 4570
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.4832259796027914e-05,
+      "loss": 0.4916,
+      "step": 4580
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.4765163714439077e-05,
+      "loss": 0.4491,
+      "step": 4590
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.469806763285024e-05,
+      "loss": 0.4828,
+      "step": 4600
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4630971551261408e-05,
+      "loss": 0.4927,
+      "step": 4610
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.456387546967257e-05,
+      "loss": 0.5052,
+      "step": 4620
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.449677938808374e-05,
+      "loss": 0.4043,
+      "step": 4630
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.4429683306494902e-05,
+      "loss": 0.4726,
+      "step": 4640
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.4362587224906066e-05,
+      "loss": 0.5479,
+      "step": 4650
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.429549114331723e-05,
+      "loss": 0.4143,
+      "step": 4660
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.4228395061728396e-05,
+      "loss": 0.493,
+      "step": 4670
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.416129898013956e-05,
+      "loss": 0.475,
+      "step": 4680
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.4094202898550724e-05,
+      "loss": 0.5184,
+      "step": 4690
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.402710681696189e-05,
+      "loss": 0.4779,
+      "step": 4700
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.3960010735373058e-05,
+      "loss": 0.4878,
+      "step": 4710
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.389291465378422e-05,
+      "loss": 0.4925,
+      "step": 4720
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.3825818572195385e-05,
+      "loss": 0.5488,
+      "step": 4730
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.920910891089109,
+      "eval_loss": 0.27520403265953064,
+      "eval_runtime": 406.5032,
+      "eval_samples_per_second": 62.115,
+      "eval_steps_per_second": 3.884,
+      "step": 4735
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.375872249060655e-05,
+      "loss": 0.4702,
+      "step": 4740
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.3691626409017716e-05,
+      "loss": 0.438,
+      "step": 4750
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.362453032742888e-05,
+      "loss": 0.437,
+      "step": 4760
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.3557434245840043e-05,
+      "loss": 0.4423,
+      "step": 4770
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.3490338164251206e-05,
+      "loss": 0.4678,
+      "step": 4780
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.3423242082662373e-05,
+      "loss": 0.3891,
+      "step": 4790
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.335614600107354e-05,
+      "loss": 0.4133,
+      "step": 4800
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.3289049919484704e-05,
+      "loss": 0.4515,
+      "step": 4810
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.3221953837895868e-05,
+      "loss": 0.3847,
+      "step": 4820
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.315485775630703e-05,
+      "loss": 0.4034,
+      "step": 4830
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.30877616747182e-05,
+      "loss": 0.4844,
+      "step": 4840
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.3020665593129362e-05,
+      "loss": 0.4264,
+      "step": 4850
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.2953569511540526e-05,
+      "loss": 0.4721,
+      "step": 4860
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.288647342995169e-05,
+      "loss": 0.4572,
+      "step": 4870
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.2819377348362856e-05,
+      "loss": 0.3871,
+      "step": 4880
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.2752281266774023e-05,
+      "loss": 0.4147,
+      "step": 4890
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.2685185185185187e-05,
+      "loss": 0.4352,
+      "step": 4900
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.261808910359635e-05,
+      "loss": 0.4346,
+      "step": 4910
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.2550993022007517e-05,
+      "loss": 0.43,
+      "step": 4920
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.248389694041868e-05,
+      "loss": 0.4396,
+      "step": 4930
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.2416800858829845e-05,
+      "loss": 0.4214,
+      "step": 4940
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.234970477724101e-05,
+      "loss": 0.3948,
+      "step": 4950
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.2282608695652175e-05,
+      "loss": 0.4116,
+      "step": 4960
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.221551261406334e-05,
+      "loss": 0.4448,
+      "step": 4970
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.2148416532474506e-05,
+      "loss": 0.4443,
+      "step": 4980
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.208132045088567e-05,
+      "loss": 0.4902,
+      "step": 4990
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.2014224369296833e-05,
+      "loss": 0.4488,
+      "step": 5000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 2.1947128287708e-05,
+      "loss": 0.429,
+      "step": 5010
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.1880032206119164e-05,
+      "loss": 0.3614,
+      "step": 5020
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.1812936124530327e-05,
+      "loss": 0.3985,
+      "step": 5030
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.174584004294149e-05,
+      "loss": 0.4337,
+      "step": 5040
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.1678743961352658e-05,
+      "loss": 0.4836,
+      "step": 5050
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.1611647879763822e-05,
+      "loss": 0.5104,
+      "step": 5060
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.154455179817499e-05,
+      "loss": 0.4572,
+      "step": 5070
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.1477455716586152e-05,
+      "loss": 0.3736,
+      "step": 5080
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.141035963499732e-05,
+      "loss": 0.4681,
+      "step": 5090
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.1343263553408483e-05,
+      "loss": 0.4808,
+      "step": 5100
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.1276167471819647e-05,
+      "loss": 0.4718,
+      "step": 5110
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.120907139023081e-05,
+      "loss": 0.4537,
+      "step": 5120
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.1141975308641977e-05,
+      "loss": 0.4686,
+      "step": 5130
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.107487922705314e-05,
+      "loss": 0.3562,
+      "step": 5140
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.1007783145464305e-05,
+      "loss": 0.453,
+      "step": 5150
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.0940687063875468e-05,
+      "loss": 0.3651,
+      "step": 5160
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.0873590982286635e-05,
+      "loss": 0.4608,
+      "step": 5170
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0806494900697802e-05,
+      "loss": 0.4224,
+      "step": 5180
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.0739398819108966e-05,
+      "loss": 0.4482,
+      "step": 5190
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.067230273752013e-05,
+      "loss": 0.3867,
+      "step": 5200
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.0605206655931293e-05,
+      "loss": 0.4543,
+      "step": 5210
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.053811057434246e-05,
+      "loss": 0.4875,
+      "step": 5220
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.0471014492753624e-05,
+      "loss": 0.4759,
+      "step": 5230
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.0403918411164787e-05,
+      "loss": 0.3958,
+      "step": 5240
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.033682232957595e-05,
+      "loss": 0.496,
+      "step": 5250
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.026972624798712e-05,
+      "loss": 0.4213,
+      "step": 5260
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.0202630166398285e-05,
+      "loss": 0.3921,
+      "step": 5270
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.013553408480945e-05,
+      "loss": 0.3664,
+      "step": 5280
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.0068438003220612e-05,
+      "loss": 0.4583,
+      "step": 5290
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.000134192163178e-05,
+      "loss": 0.4207,
+      "step": 5300
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.9934245840042943e-05,
+      "loss": 0.4283,
+      "step": 5310
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.9867149758454106e-05,
+      "loss": 0.4012,
+      "step": 5320
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.980005367686527e-05,
+      "loss": 0.444,
+      "step": 5330
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9732957595276437e-05,
+      "loss": 0.4419,
+      "step": 5340
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9665861513687604e-05,
+      "loss": 0.4532,
+      "step": 5350
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.9598765432098768e-05,
+      "loss": 0.4268,
+      "step": 5360
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.953166935050993e-05,
+      "loss": 0.4851,
+      "step": 5370
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.9464573268921095e-05,
+      "loss": 0.393,
+      "step": 5380
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.9397477187332262e-05,
+      "loss": 0.4428,
+      "step": 5390
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.9330381105743426e-05,
+      "loss": 0.4332,
+      "step": 5400
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.926328502415459e-05,
+      "loss": 0.4311,
+      "step": 5410
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.9196188942565753e-05,
+      "loss": 0.458,
+      "step": 5420
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.912909286097692e-05,
+      "loss": 0.4833,
+      "step": 5430
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.9061996779388087e-05,
+      "loss": 0.3519,
+      "step": 5440
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.899490069779925e-05,
+      "loss": 0.4852,
+      "step": 5450
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.8927804616210414e-05,
+      "loss": 0.3888,
+      "step": 5460
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.886070853462158e-05,
+      "loss": 0.4399,
+      "step": 5470
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.8793612453032745e-05,
+      "loss": 0.4736,
+      "step": 5480
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.872651637144391e-05,
+      "loss": 0.3755,
+      "step": 5490
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.8659420289855072e-05,
+      "loss": 0.4528,
+      "step": 5500
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.859232420826624e-05,
+      "loss": 0.3739,
+      "step": 5510
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.8525228126677403e-05,
+      "loss": 0.4313,
+      "step": 5520
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.8458132045088566e-05,
+      "loss": 0.3996,
+      "step": 5530
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.8391035963499733e-05,
+      "loss": 0.3957,
+      "step": 5540
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.8323939881910897e-05,
+      "loss": 0.3689,
+      "step": 5550
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8256843800322064e-05,
+      "loss": 0.4107,
+      "step": 5560
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.8189747718733227e-05,
+      "loss": 0.4122,
+      "step": 5570
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.812265163714439e-05,
+      "loss": 0.4545,
+      "step": 5580
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.8055555555555555e-05,
+      "loss": 0.4937,
+      "step": 5590
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.7988459473966722e-05,
+      "loss": 0.414,
+      "step": 5600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.7921363392377885e-05,
+      "loss": 0.3985,
+      "step": 5610
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.785426731078905e-05,
+      "loss": 0.4149,
+      "step": 5620
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7787171229200216e-05,
+      "loss": 0.4272,
+      "step": 5630
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7720075147611383e-05,
+      "loss": 0.4113,
+      "step": 5640
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.7652979066022547e-05,
+      "loss": 0.4205,
+      "step": 5650
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.758588298443371e-05,
+      "loss": 0.4442,
+      "step": 5660
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.7518786902844874e-05,
+      "loss": 0.4958,
+      "step": 5670
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.745169082125604e-05,
+      "loss": 0.4831,
+      "step": 5680
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.7384594739667205e-05,
+      "loss": 0.469,
+      "step": 5690
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7317498658078368e-05,
+      "loss": 0.4265,
+      "step": 5700
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7250402576489532e-05,
+      "loss": 0.4247,
+      "step": 5710
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.71833064949007e-05,
+      "loss": 0.4296,
+      "step": 5720
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.7116210413311866e-05,
+      "loss": 0.4157,
+      "step": 5730
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.704911433172303e-05,
+      "loss": 0.438,
+      "step": 5740
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.6982018250134193e-05,
+      "loss": 0.4071,
+      "step": 5750
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6914922168545357e-05,
+      "loss": 0.446,
+      "step": 5760
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6847826086956524e-05,
+      "loss": 0.4436,
+      "step": 5770
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.6780730005367687e-05,
+      "loss": 0.4553,
+      "step": 5780
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.671363392377885e-05,
+      "loss": 0.3775,
+      "step": 5790
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.6646537842190015e-05,
+      "loss": 0.3559,
+      "step": 5800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.657944176060118e-05,
+      "loss": 0.4753,
+      "step": 5810
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.651234567901235e-05,
+      "loss": 0.428,
+      "step": 5820
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6445249597423512e-05,
+      "loss": 0.4214,
+      "step": 5830
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.6378153515834676e-05,
+      "loss": 0.4276,
+      "step": 5840
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.631105743424584e-05,
+      "loss": 0.4947,
+      "step": 5850
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.6243961352657006e-05,
+      "loss": 0.4876,
+      "step": 5860
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.617686527106817e-05,
+      "loss": 0.4058,
+      "step": 5870
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.6109769189479334e-05,
+      "loss": 0.3293,
+      "step": 5880
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.60426731078905e-05,
+      "loss": 0.408,
+      "step": 5890
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5975577026301664e-05,
+      "loss": 0.4163,
+      "step": 5900
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.590848094471283e-05,
+      "loss": 0.4527,
+      "step": 5910
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9254653465346535,
+      "eval_loss": 0.2600358724594116,
+      "eval_runtime": 404.2244,
+      "eval_samples_per_second": 62.465,
+      "eval_steps_per_second": 3.906,
+      "step": 5918
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5841384863123995e-05,
+      "loss": 0.4546,
+      "step": 5920
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.577428878153516e-05,
+      "loss": 0.4224,
+      "step": 5930
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5707192699946326e-05,
+      "loss": 0.3777,
+      "step": 5940
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.564009661835749e-05,
+      "loss": 0.4111,
+      "step": 5950
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5573000536768653e-05,
+      "loss": 0.3774,
+      "step": 5960
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5505904455179816e-05,
+      "loss": 0.3476,
+      "step": 5970
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5438808373590983e-05,
+      "loss": 0.4537,
+      "step": 5980
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.5371712292002147e-05,
+      "loss": 0.3923,
+      "step": 5990
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5304616210413314e-05,
+      "loss": 0.3981,
+      "step": 6000
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.5237520128824478e-05,
+      "loss": 0.4233,
+      "step": 6010
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5170424047235643e-05,
+      "loss": 0.4101,
+      "step": 6020
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5103327965646807e-05,
+      "loss": 0.3733,
+      "step": 6030
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.5036231884057972e-05,
+      "loss": 0.3557,
+      "step": 6040
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.4969135802469136e-05,
+      "loss": 0.3727,
+      "step": 6050
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.4902039720880301e-05,
+      "loss": 0.2747,
+      "step": 6060
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.4834943639291465e-05,
+      "loss": 0.36,
+      "step": 6070
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.476784755770263e-05,
+      "loss": 0.3925,
+      "step": 6080
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.4700751476113795e-05,
+      "loss": 0.3208,
+      "step": 6090
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.4633655394524962e-05,
+      "loss": 0.411,
+      "step": 6100
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.4566559312936126e-05,
+      "loss": 0.3541,
+      "step": 6110
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.4499463231347291e-05,
+      "loss": 0.3226,
+      "step": 6120
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.4432367149758455e-05,
+      "loss": 0.3397,
+      "step": 6130
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.436527106816962e-05,
+      "loss": 0.393,
+      "step": 6140
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4298174986580784e-05,
+      "loss": 0.3464,
+      "step": 6150
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.4231078904991949e-05,
+      "loss": 0.3783,
+      "step": 6160
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.4163982823403113e-05,
+      "loss": 0.3059,
+      "step": 6170
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.4096886741814278e-05,
+      "loss": 0.3839,
+      "step": 6180
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.4029790660225445e-05,
+      "loss": 0.3846,
+      "step": 6190
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.3962694578636609e-05,
+      "loss": 0.3753,
+      "step": 6200
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.3895598497047774e-05,
+      "loss": 0.3844,
+      "step": 6210
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.3828502415458937e-05,
+      "loss": 0.4096,
+      "step": 6220
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.3761406333870103e-05,
+      "loss": 0.3862,
+      "step": 6230
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.3694310252281266e-05,
+      "loss": 0.3715,
+      "step": 6240
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.3627214170692432e-05,
+      "loss": 0.442,
+      "step": 6250
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.3560118089103595e-05,
+      "loss": 0.3484,
+      "step": 6260
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.349302200751476e-05,
+      "loss": 0.3228,
+      "step": 6270
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.3425925925925928e-05,
+      "loss": 0.3554,
+      "step": 6280
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.3358829844337093e-05,
+      "loss": 0.3982,
+      "step": 6290
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.3291733762748257e-05,
+      "loss": 0.3629,
+      "step": 6300
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.3224637681159422e-05,
+      "loss": 0.3859,
+      "step": 6310
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.3157541599570586e-05,
+      "loss": 0.4495,
+      "step": 6320
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.3090445517981751e-05,
+      "loss": 0.4382,
+      "step": 6330
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.3023349436392915e-05,
+      "loss": 0.4376,
+      "step": 6340
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.295625335480408e-05,
+      "loss": 0.3863,
+      "step": 6350
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.2889157273215243e-05,
+      "loss": 0.4027,
+      "step": 6360
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.2822061191626409e-05,
+      "loss": 0.3861,
+      "step": 6370
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.2754965110037576e-05,
+      "loss": 0.317,
+      "step": 6380
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.268786902844874e-05,
+      "loss": 0.4036,
+      "step": 6390
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.2620772946859905e-05,
+      "loss": 0.377,
+      "step": 6400
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.2553676865271068e-05,
+      "loss": 0.4009,
+      "step": 6410
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.2486580783682234e-05,
+      "loss": 0.4451,
+      "step": 6420
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.2419484702093397e-05,
+      "loss": 0.4013,
+      "step": 6430
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.2352388620504563e-05,
+      "loss": 0.4352,
+      "step": 6440
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.2285292538915728e-05,
+      "loss": 0.3146,
+      "step": 6450
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.2218196457326893e-05,
+      "loss": 0.4974,
+      "step": 6460
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.2151100375738057e-05,
+      "loss": 0.3879,
+      "step": 6470
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.2084004294149222e-05,
+      "loss": 0.3095,
+      "step": 6480
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.2016908212560387e-05,
+      "loss": 0.3223,
+      "step": 6490
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.1949812130971553e-05,
+      "loss": 0.3637,
+      "step": 6500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.1882716049382716e-05,
+      "loss": 0.3744,
+      "step": 6510
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.1815619967793882e-05,
+      "loss": 0.3693,
+      "step": 6520
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.1748523886205045e-05,
+      "loss": 0.3547,
+      "step": 6530
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.168142780461621e-05,
+      "loss": 0.393,
+      "step": 6540
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.1614331723027376e-05,
+      "loss": 0.3454,
+      "step": 6550
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.1547235641438541e-05,
+      "loss": 0.4265,
+      "step": 6560
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.1480139559849705e-05,
+      "loss": 0.2771,
+      "step": 6570
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.141304347826087e-05,
+      "loss": 0.3607,
+      "step": 6580
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.1345947396672036e-05,
+      "loss": 0.4114,
+      "step": 6590
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.12788513150832e-05,
+      "loss": 0.3702,
+      "step": 6600
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.1211755233494365e-05,
+      "loss": 0.3729,
+      "step": 6610
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.1144659151905528e-05,
+      "loss": 0.2976,
+      "step": 6620
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.1077563070316695e-05,
+      "loss": 0.3772,
+      "step": 6630
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.1010466988727859e-05,
+      "loss": 0.3785,
+      "step": 6640
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.0943370907139024e-05,
+      "loss": 0.3918,
+      "step": 6650
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.0876274825550188e-05,
+      "loss": 0.4002,
+      "step": 6660
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.0809178743961353e-05,
+      "loss": 0.2988,
+      "step": 6670
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.0742082662372518e-05,
+      "loss": 0.2904,
+      "step": 6680
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.0674986580783684e-05,
+      "loss": 0.4303,
+      "step": 6690
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.0607890499194847e-05,
+      "loss": 0.4125,
+      "step": 6700
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.0540794417606013e-05,
+      "loss": 0.354,
+      "step": 6710
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.0473698336017176e-05,
+      "loss": 0.3508,
+      "step": 6720
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.0406602254428342e-05,
+      "loss": 0.4478,
+      "step": 6730
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.0339506172839507e-05,
+      "loss": 0.2989,
+      "step": 6740
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.027241009125067e-05,
+      "loss": 0.3582,
+      "step": 6750
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.0205314009661836e-05,
+      "loss": 0.4106,
+      "step": 6760
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.0138217928073001e-05,
+      "loss": 0.3462,
+      "step": 6770
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.0071121846484166e-05,
+      "loss": 0.4752,
+      "step": 6780
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.000402576489533e-05,
+      "loss": 0.3162,
+      "step": 6790
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.936929683306495e-06,
+      "loss": 0.3622,
+      "step": 6800
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.869833601717659e-06,
+      "loss": 0.3441,
+      "step": 6810
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 9.802737520128826e-06,
+      "loss": 0.3836,
+      "step": 6820
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 9.73564143853999e-06,
+      "loss": 0.4264,
+      "step": 6830
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 9.668545356951155e-06,
+      "loss": 0.3951,
+      "step": 6840
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 9.601449275362319e-06,
+      "loss": 0.3384,
+      "step": 6850
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.534353193773484e-06,
+      "loss": 0.3336,
+      "step": 6860
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.46725711218465e-06,
+      "loss": 0.3547,
+      "step": 6870
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.400161030595815e-06,
+      "loss": 0.3235,
+      "step": 6880
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.333064949006978e-06,
+      "loss": 0.3729,
+      "step": 6890
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 9.265968867418143e-06,
+      "loss": 0.4095,
+      "step": 6900
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 9.198872785829309e-06,
+      "loss": 0.3553,
+      "step": 6910
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 9.131776704240472e-06,
+      "loss": 0.3447,
+      "step": 6920
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 9.064680622651638e-06,
+      "loss": 0.3497,
+      "step": 6930
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 8.997584541062801e-06,
+      "loss": 0.3942,
+      "step": 6940
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 8.930488459473967e-06,
+      "loss": 0.3625,
+      "step": 6950
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 8.863392377885132e-06,
+      "loss": 0.3395,
+      "step": 6960
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 8.796296296296297e-06,
+      "loss": 0.4519,
+      "step": 6970
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 8.729200214707461e-06,
+      "loss": 0.295,
+      "step": 6980
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 8.662104133118626e-06,
+      "loss": 0.3756,
+      "step": 6990
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 8.59500805152979e-06,
+      "loss": 0.4562,
+      "step": 7000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 8.527911969940957e-06,
+      "loss": 0.3559,
+      "step": 7010
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 8.46081588835212e-06,
+      "loss": 0.3502,
+      "step": 7020
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 8.393719806763286e-06,
+      "loss": 0.3596,
+      "step": 7030
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 8.32662372517445e-06,
+      "loss": 0.3637,
+      "step": 7040
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 8.259527643585616e-06,
+      "loss": 0.3959,
+      "step": 7050
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 8.19243156199678e-06,
+      "loss": 0.3546,
+      "step": 7060
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 8.125335480407945e-06,
+      "loss": 0.4052,
+      "step": 7070
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 8.058239398819109e-06,
+      "loss": 0.3444,
+      "step": 7080
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.991143317230274e-06,
+      "loss": 0.3889,
+      "step": 7090
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.92404723564144e-06,
+      "loss": 0.3692,
+      "step": 7100
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9272475247524753,
+      "eval_loss": 0.2519378662109375,
+      "eval_runtime": 407.0035,
+      "eval_samples_per_second": 62.039,
+      "eval_steps_per_second": 3.88,
+      "step": 7102
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.856951154052603e-06,
+      "loss": 0.4123,
+      "step": 7110
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.789855072463769e-06,
+      "loss": 0.3047,
+      "step": 7120
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.722758990874932e-06,
+      "loss": 0.2708,
+      "step": 7130
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.6556629092861e-06,
+      "loss": 0.3236,
+      "step": 7140
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.588566827697263e-06,
+      "loss": 0.3952,
+      "step": 7150
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.521470746108428e-06,
+      "loss": 0.3507,
+      "step": 7160
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.454374664519593e-06,
+      "loss": 0.3079,
+      "step": 7170
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.387278582930757e-06,
+      "loss": 0.4077,
+      "step": 7180
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.320182501341922e-06,
+      "loss": 0.2877,
+      "step": 7190
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.253086419753087e-06,
+      "loss": 0.3456,
+      "step": 7200
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.185990338164251e-06,
+      "loss": 0.3498,
+      "step": 7210
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 7.118894256575416e-06,
+      "loss": 0.3442,
+      "step": 7220
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 7.05179817498658e-06,
+      "loss": 0.3833,
+      "step": 7230
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.9847020933977464e-06,
+      "loss": 0.3155,
+      "step": 7240
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.917606011808911e-06,
+      "loss": 0.3417,
+      "step": 7250
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.850509930220075e-06,
+      "loss": 0.2992,
+      "step": 7260
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.78341384863124e-06,
+      "loss": 0.4262,
+      "step": 7270
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.716317767042406e-06,
+      "loss": 0.3631,
+      "step": 7280
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.6492216854535705e-06,
+      "loss": 0.4147,
+      "step": 7290
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.582125603864735e-06,
+      "loss": 0.356,
+      "step": 7300
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.515029522275899e-06,
+      "loss": 0.3378,
+      "step": 7310
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.447933440687064e-06,
+      "loss": 0.28,
+      "step": 7320
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.380837359098229e-06,
+      "loss": 0.3808,
+      "step": 7330
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.313741277509394e-06,
+      "loss": 0.3424,
+      "step": 7340
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.246645195920558e-06,
+      "loss": 0.3462,
+      "step": 7350
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.179549114331723e-06,
+      "loss": 0.3508,
+      "step": 7360
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.112453032742888e-06,
+      "loss": 0.3141,
+      "step": 7370
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.045356951154052e-06,
+      "loss": 0.3214,
+      "step": 7380
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 5.978260869565218e-06,
+      "loss": 0.3332,
+      "step": 7390
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 5.911164787976382e-06,
+      "loss": 0.3817,
+      "step": 7400
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 5.8440687063875475e-06,
+      "loss": 0.3602,
+      "step": 7410
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 5.776972624798712e-06,
+      "loss": 0.2888,
+      "step": 7420
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 5.7098765432098764e-06,
+      "loss": 0.3167,
+      "step": 7430
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.642780461621042e-06,
+      "loss": 0.366,
+      "step": 7440
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 5.575684380032206e-06,
+      "loss": 0.3613,
+      "step": 7450
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 5.5085882984433715e-06,
+      "loss": 0.3911,
+      "step": 7460
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 5.441492216854536e-06,
+      "loss": 0.3058,
+      "step": 7470
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 5.374396135265701e-06,
+      "loss": 0.2861,
+      "step": 7480
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 5.307300053676866e-06,
+      "loss": 0.3622,
+      "step": 7490
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.24020397208803e-06,
+      "loss": 0.3764,
+      "step": 7500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 5.173107890499196e-06,
+      "loss": 0.3326,
+      "step": 7510
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 5.10601180891036e-06,
+      "loss": 0.336,
+      "step": 7520
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 5.0389157273215245e-06,
+      "loss": 0.3021,
+      "step": 7530
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 4.971819645732689e-06,
+      "loss": 0.3657,
+      "step": 7540
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 4.904723564143854e-06,
+      "loss": 0.3544,
+      "step": 7550
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.837627482555019e-06,
+      "loss": 0.3104,
+      "step": 7560
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 4.770531400966183e-06,
+      "loss": 0.382,
+      "step": 7570
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 4.7034353193773486e-06,
+      "loss": 0.377,
+      "step": 7580
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 4.636339237788513e-06,
+      "loss": 0.3392,
+      "step": 7590
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 4.569243156199678e-06,
+      "loss": 0.2541,
+      "step": 7600
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 4.502147074610843e-06,
+      "loss": 0.3582,
+      "step": 7610
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 4.435050993022008e-06,
+      "loss": 0.3047,
+      "step": 7620
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.367954911433173e-06,
+      "loss": 0.3324,
+      "step": 7630
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 4.300858829844337e-06,
+      "loss": 0.3744,
+      "step": 7640
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 4.233762748255502e-06,
+      "loss": 0.3741,
+      "step": 7650
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.3683,
+      "step": 7660
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 4.099570585077832e-06,
+      "loss": 0.3265,
+      "step": 7670
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 4.032474503488997e-06,
+      "loss": 0.3449,
+      "step": 7680
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.965378421900161e-06,
+      "loss": 0.3083,
+      "step": 7690
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.898282340311326e-06,
+      "loss": 0.4392,
+      "step": 7700
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.83118625872249e-06,
+      "loss": 0.333,
+      "step": 7710
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 3.764090177133656e-06,
+      "loss": 0.2682,
+      "step": 7720
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.6969940955448203e-06,
+      "loss": 0.3805,
+      "step": 7730
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 3.629898013955985e-06,
+      "loss": 0.3493,
+      "step": 7740
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.5628019323671496e-06,
+      "loss": 0.3319,
+      "step": 7750
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.495705850778315e-06,
+      "loss": 0.3455,
+      "step": 7760
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 3.4286097691894794e-06,
+      "loss": 0.3389,
+      "step": 7770
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.361513687600644e-06,
+      "loss": 0.3685,
+      "step": 7780
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 3.294417606011809e-06,
+      "loss": 0.3595,
+      "step": 7790
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.2273215244229737e-06,
+      "loss": 0.3194,
+      "step": 7800
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 3.160225442834139e-06,
+      "loss": 0.3005,
+      "step": 7810
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.0931293612453035e-06,
+      "loss": 0.3638,
+      "step": 7820
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.026033279656468e-06,
+      "loss": 0.3484,
+      "step": 7830
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 2.958937198067633e-06,
+      "loss": 0.2803,
+      "step": 7840
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 2.8918411164787977e-06,
+      "loss": 0.3512,
+      "step": 7850
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 2.8247450348899626e-06,
+      "loss": 0.3223,
+      "step": 7860
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 2.7576489533011275e-06,
+      "loss": 0.3594,
+      "step": 7870
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 2.6905528717122924e-06,
+      "loss": 0.3005,
+      "step": 7880
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.623456790123457e-06,
+      "loss": 0.3444,
+      "step": 7890
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 2.5563607085346218e-06,
+      "loss": 0.3464,
+      "step": 7900
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 2.4892646269457862e-06,
+      "loss": 0.327,
+      "step": 7910
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 2.422168545356951e-06,
+      "loss": 0.3225,
+      "step": 7920
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 2.355072463768116e-06,
+      "loss": 0.3719,
+      "step": 7930
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 2.287976382179281e-06,
+      "loss": 0.3101,
+      "step": 7940
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.220880300590446e-06,
+      "loss": 0.303,
+      "step": 7950
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 2.1537842190016103e-06,
+      "loss": 0.459,
+      "step": 7960
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 2.086688137412775e-06,
+      "loss": 0.2967,
+      "step": 7970
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 2.01959205582394e-06,
+      "loss": 0.3771,
+      "step": 7980
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.952495974235105e-06,
+      "loss": 0.326,
+      "step": 7990
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.8853998926462696e-06,
+      "loss": 0.351,
+      "step": 8000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.8183038110574345e-06,
+      "loss": 0.324,
+      "step": 8010
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.7512077294685992e-06,
+      "loss": 0.3017,
+      "step": 8020
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6841116478797637e-06,
+      "loss": 0.4255,
+      "step": 8030
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6170155662909286e-06,
+      "loss": 0.2741,
+      "step": 8040
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5499194847020935e-06,
+      "loss": 0.3028,
+      "step": 8050
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4828234031132581e-06,
+      "loss": 0.281,
+      "step": 8060
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.415727321524423e-06,
+      "loss": 0.3243,
+      "step": 8070
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.3486312399355877e-06,
+      "loss": 0.3361,
+      "step": 8080
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.2815351583467526e-06,
+      "loss": 0.3363,
+      "step": 8090
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.2144390767579175e-06,
+      "loss": 0.3142,
+      "step": 8100
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.1473429951690822e-06,
+      "loss": 0.3222,
+      "step": 8110
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.0802469135802469e-06,
+      "loss": 0.3898,
+      "step": 8120
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.0131508319914118e-06,
+      "loss": 0.3503,
+      "step": 8130
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.460547504025766e-07,
+      "loss": 0.2973,
+      "step": 8140
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 8.789586688137412e-07,
+      "loss": 0.3506,
+      "step": 8150
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 8.11862587224906e-07,
+      "loss": 0.349,
+      "step": 8160
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 7.447665056360709e-07,
+      "loss": 0.3323,
+      "step": 8170
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 6.776704240472356e-07,
+      "loss": 0.3286,
+      "step": 8180
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 6.105743424584005e-07,
+      "loss": 0.334,
+      "step": 8190
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 5.434782608695653e-07,
+      "loss": 0.3741,
+      "step": 8200
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.7638217928073006e-07,
+      "loss": 0.3811,
+      "step": 8210
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.092860976918948e-07,
+      "loss": 0.2902,
+      "step": 8220
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.421900161030596e-07,
+      "loss": 0.3058,
+      "step": 8230
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 2.7509393451422437e-07,
+      "loss": 0.3061,
+      "step": 8240
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 2.0799785292538919e-07,
+      "loss": 0.3699,
+      "step": 8250
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.4090177133655395e-07,
+      "loss": 0.3062,
+      "step": 8260
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 7.380568974771873e-08,
+      "loss": 0.3713,
+      "step": 8270
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 6.709608158883521e-09,
+      "loss": 0.3731,
+      "step": 8280
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.928910891089109,
+      "eval_loss": 0.24779632687568665,
+      "eval_runtime": 406.392,
+      "eval_samples_per_second": 62.132,
+      "eval_steps_per_second": 3.885,
+      "step": 8281
+    },
+    {
+      "epoch": 7.0,
+      "step": 8281,
+      "total_flos": 4.15657572108913e+19,
+      "train_loss": 0.6947555202905746,
+      "train_runtime": 23204.4343,
+      "train_samples_per_second": 22.851,
+      "train_steps_per_second": 0.357
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 8281,
+  "num_train_epochs": 7,
+  "save_steps": 500,
+  "total_flos": 4.15657572108913e+19,
+  "trial_name": null,
+  "trial_params": null
+}