diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6817 @@
+{
+  "best_metric": 0.991156462585034,
+  "best_model_checkpoint": "vit-base-patch16-224-finetuned-main-gpu-20e-final/checkpoint-10469",
+  "epoch": 19.99909338168631,
+  "global_step": 11020,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.537205081669692e-07,
+      "loss": 1.4008,
+      "step": 10
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.074410163339384e-07,
+      "loss": 1.3823,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3611615245009075e-06,
+      "loss": 1.3568,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8148820326678768e-06,
+      "loss": 1.2922,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.268602540834846e-06,
+      "loss": 1.2212,
+      "step": 50
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.722323049001815e-06,
+      "loss": 1.1398,
+      "step": 60
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.176043557168784e-06,
+      "loss": 1.0566,
+      "step": 70
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6297640653357536e-06,
+      "loss": 0.9813,
+      "step": 80
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.083484573502722e-06,
+      "loss": 0.923,
+      "step": 90
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.537205081669692e-06,
+      "loss": 0.8694,
+      "step": 100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9909255898366605e-06,
+      "loss": 0.8238,
+      "step": 110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.44464609800363e-06,
+      "loss": 0.8267,
+      "step": 120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.8983666061705996e-06,
+      "loss": 0.7945,
+      "step": 130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.352087114337568e-06,
+      "loss": 0.7639,
+      "step": 140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 6.805807622504537e-06,
+      "loss": 0.7407,
+      "step": 150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.259528130671507e-06,
+      "loss": 0.7238,
+      "step": 160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.713248638838475e-06,
+      "loss": 0.6893,
+      "step": 170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.166969147005445e-06,
+      "loss": 0.6843,
+      "step": 180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.620689655172414e-06,
+      "loss": 0.6971,
+      "step": 190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.074410163339384e-06,
+      "loss": 0.6873,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.528130671506351e-06,
+      "loss": 0.6959,
+      "step": 210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.981851179673321e-06,
+      "loss": 0.6907,
+      "step": 220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.043557168784029e-05,
+      "loss": 0.6568,
+      "step": 230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.088929219600726e-05,
+      "loss": 0.6438,
+      "step": 240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1343012704174228e-05,
+      "loss": 0.635,
+      "step": 250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1796733212341199e-05,
+      "loss": 0.6178,
+      "step": 260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2250453720508167e-05,
+      "loss": 0.6333,
+      "step": 270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2704174228675136e-05,
+      "loss": 0.6082,
+      "step": 280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 0.6191,
+      "step": 290
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3611615245009074e-05,
+      "loss": 0.6187,
+      "step": 300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4065335753176045e-05,
+      "loss": 0.6194,
+      "step": 310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.4519056261343015e-05,
+      "loss": 0.563,
+      "step": 320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.4972776769509982e-05,
+      "loss": 0.5597,
+      "step": 330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.542649727767695e-05,
+      "loss": 0.5863,
+      "step": 340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.588021778584392e-05,
+      "loss": 0.5854,
+      "step": 350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.633393829401089e-05,
+      "loss": 0.562,
+      "step": 360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.6787658802177857e-05,
+      "loss": 0.5369,
+      "step": 370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7241379310344828e-05,
+      "loss": 0.5441,
+      "step": 380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.76950998185118e-05,
+      "loss": 0.5438,
+      "step": 390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8148820326678767e-05,
+      "loss": 0.5295,
+      "step": 400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8602540834845735e-05,
+      "loss": 0.5359,
+      "step": 410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9056261343012703e-05,
+      "loss": 0.5553,
+      "step": 420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9509981851179674e-05,
+      "loss": 0.5457,
+      "step": 430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9963702359346642e-05,
+      "loss": 0.4994,
+      "step": 440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.0417422867513613e-05,
+      "loss": 0.5178,
+      "step": 450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.087114337568058e-05,
+      "loss": 0.5147,
+      "step": 460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1324863883847552e-05,
+      "loss": 0.49,
+      "step": 470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.177858439201452e-05,
+      "loss": 0.5035,
+      "step": 480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2232304900181488e-05,
+      "loss": 0.4859,
+      "step": 490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2686025408348456e-05,
+      "loss": 0.5102,
+      "step": 500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3139745916515427e-05,
+      "loss": 0.5489,
+      "step": 510
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3593466424682398e-05,
+      "loss": 0.4851,
+      "step": 520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4047186932849366e-05,
+      "loss": 0.481,
+      "step": 530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4500907441016334e-05,
+      "loss": 0.4762,
+      "step": 540
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4954627949183305e-05,
+      "loss": 0.4852,
+      "step": 550
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.8041666666666667,
+      "eval_loss": 0.4533345103263855,
+      "eval_runtime": 88.4947,
+      "eval_samples_per_second": 132.889,
+      "eval_steps_per_second": 4.158,
+      "step": 551
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5408348457350273e-05,
+      "loss": 0.4793,
+      "step": 560
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.5862068965517244e-05,
+      "loss": 0.4701,
+      "step": 570
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.4398,
+      "step": 580
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.6769509981851183e-05,
+      "loss": 0.4399,
+      "step": 590
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7223230490018148e-05,
+      "loss": 0.4783,
+      "step": 600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.767695099818512e-05,
+      "loss": 0.4607,
+      "step": 610
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.813067150635209e-05,
+      "loss": 0.4361,
+      "step": 620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.8584392014519058e-05,
+      "loss": 0.4282,
+      "step": 630
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.903811252268603e-05,
+      "loss": 0.4311,
+      "step": 640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9491833030852994e-05,
+      "loss": 0.424,
+      "step": 650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9945553539019965e-05,
+      "loss": 0.4138,
+      "step": 660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 3.0399274047186933e-05,
+      "loss": 0.4233,
+      "step": 670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 3.08529945553539e-05,
+      "loss": 0.4143,
+      "step": 680
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 3.1306715063520875e-05,
+      "loss": 0.4098,
+      "step": 690
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 3.176043557168784e-05,
+      "loss": 0.4038,
+      "step": 700
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 3.221415607985481e-05,
+      "loss": 0.4072,
+      "step": 710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 3.266787658802178e-05,
+      "loss": 0.4479,
+      "step": 720
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 3.312159709618875e-05,
+      "loss": 0.4373,
+      "step": 730
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 3.3575317604355714e-05,
+      "loss": 0.4132,
+      "step": 740
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 3.402903811252269e-05,
+      "loss": 0.4029,
+      "step": 750
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 3.4482758620689657e-05,
+      "loss": 0.3568,
+      "step": 760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 3.4936479128856624e-05,
+      "loss": 0.3954,
+      "step": 770
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 3.53901996370236e-05,
+      "loss": 0.3904,
+      "step": 780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 3.584392014519056e-05,
+      "loss": 0.396,
+      "step": 790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 3.6297640653357535e-05,
+      "loss": 0.3848,
+      "step": 800
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 3.67513611615245e-05,
+      "loss": 0.3961,
+      "step": 810
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 3.720508166969147e-05,
+      "loss": 0.3862,
+      "step": 820
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 3.7658802177858445e-05,
+      "loss": 0.384,
+      "step": 830
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 3.8112522686025406e-05,
+      "loss": 0.3694,
+      "step": 840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 3.856624319419238e-05,
+      "loss": 0.3728,
+      "step": 850
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 3.901996370235935e-05,
+      "loss": 0.4158,
+      "step": 860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.3759,
+      "step": 870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 3.9927404718693284e-05,
+      "loss": 0.4083,
+      "step": 880
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.038112522686026e-05,
+      "loss": 0.3776,
+      "step": 890
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.0834845735027227e-05,
+      "loss": 0.3503,
+      "step": 900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1288566243194194e-05,
+      "loss": 0.3768,
+      "step": 910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.174228675136116e-05,
+      "loss": 0.3598,
+      "step": 920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.219600725952813e-05,
+      "loss": 0.3541,
+      "step": 930
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.2649727767695105e-05,
+      "loss": 0.3566,
+      "step": 940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.3103448275862066e-05,
+      "loss": 0.3992,
+      "step": 950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.355716878402904e-05,
+      "loss": 0.3649,
+      "step": 960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.401088929219601e-05,
+      "loss": 0.3492,
+      "step": 970
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.4464609800362976e-05,
+      "loss": 0.3827,
+      "step": 980
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.491833030852995e-05,
+      "loss": 0.3357,
+      "step": 990
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.537205081669691e-05,
+      "loss": 0.3399,
+      "step": 1000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.5825771324863886e-05,
+      "loss": 0.3417,
+      "step": 1010
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.6279491833030854e-05,
+      "loss": 0.337,
+      "step": 1020
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.673321234119782e-05,
+      "loss": 0.3275,
+      "step": 1030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.7186932849364796e-05,
+      "loss": 0.3662,
+      "step": 1040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.7640653357531764e-05,
+      "loss": 0.3411,
+      "step": 1050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.809437386569873e-05,
+      "loss": 0.356,
+      "step": 1060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.85480943738657e-05,
+      "loss": 0.303,
+      "step": 1070
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.900181488203267e-05,
+      "loss": 0.3145,
+      "step": 1080
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.945553539019964e-05,
+      "loss": 0.3489,
+      "step": 1090
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.990925589836661e-05,
+      "loss": 0.3033,
+      "step": 1100
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9157312925170068,
+      "eval_loss": 0.21573838591575623,
+      "eval_runtime": 87.9072,
+      "eval_samples_per_second": 133.777,
+      "eval_steps_per_second": 4.186,
+      "step": 1102
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.995966928816294e-05,
+      "loss": 0.3447,
+      "step": 1110
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.990925589836661e-05,
+      "loss": 0.3406,
+      "step": 1120
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.9858842508570283e-05,
+      "loss": 0.3339,
+      "step": 1130
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.980842911877395e-05,
+      "loss": 0.3295,
+      "step": 1140
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.9758015728977616e-05,
+      "loss": 0.3034,
+      "step": 1150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.970760233918128e-05,
+      "loss": 0.3239,
+      "step": 1160
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.9657188949384956e-05,
+      "loss": 0.3401,
+      "step": 1170
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.960677555958863e-05,
+      "loss": 0.3184,
+      "step": 1180
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 4.9556362169792296e-05,
+      "loss": 0.2689,
+      "step": 1190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 4.950594877999597e-05,
+      "loss": 0.3159,
+      "step": 1200
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 4.945553539019964e-05,
+      "loss": 0.3356,
+      "step": 1210
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 4.940512200040331e-05,
+      "loss": 0.3247,
+      "step": 1220
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 4.935470861060698e-05,
+      "loss": 0.3114,
+      "step": 1230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 4.930429522081065e-05,
+      "loss": 0.2737,
+      "step": 1240
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 4.925388183101432e-05,
+      "loss": 0.3002,
+      "step": 1250
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.9203468441217995e-05,
+      "loss": 0.3299,
+      "step": 1260
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 4.915305505142166e-05,
+      "loss": 0.2771,
+      "step": 1270
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 4.9102641661625335e-05,
+      "loss": 0.3119,
+      "step": 1280
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 4.9052228271829e-05,
+      "loss": 0.2983,
+      "step": 1290
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.900181488203267e-05,
+      "loss": 0.267,
+      "step": 1300
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.895140149223634e-05,
+      "loss": 0.2923,
+      "step": 1310
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.890098810244001e-05,
+      "loss": 0.2502,
+      "step": 1320
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 4.885057471264368e-05,
+      "loss": 0.2816,
+      "step": 1330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.880016132284735e-05,
+      "loss": 0.2835,
+      "step": 1340
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.874974793305102e-05,
+      "loss": 0.2981,
+      "step": 1350
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.8699334543254694e-05,
+      "loss": 0.2902,
+      "step": 1360
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 4.864892115345836e-05,
+      "loss": 0.2723,
+      "step": 1370
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.8598507763662033e-05,
+      "loss": 0.2722,
+      "step": 1380
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.85480943738657e-05,
+      "loss": 0.2963,
+      "step": 1390
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.849768098406937e-05,
+      "loss": 0.2728,
+      "step": 1400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.8447267594273046e-05,
+      "loss": 0.2498,
+      "step": 1410
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.839685420447671e-05,
+      "loss": 0.2541,
+      "step": 1420
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.834644081468038e-05,
+      "loss": 0.2769,
+      "step": 1430
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.829602742488405e-05,
+      "loss": 0.2544,
+      "step": 1440
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.824561403508772e-05,
+      "loss": 0.2452,
+      "step": 1450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.819520064529139e-05,
+      "loss": 0.2801,
+      "step": 1460
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.814478725549506e-05,
+      "loss": 0.2868,
+      "step": 1470
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.809437386569873e-05,
+      "loss": 0.2837,
+      "step": 1480
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.80439604759024e-05,
+      "loss": 0.2455,
+      "step": 1490
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.799354708610607e-05,
+      "loss": 0.2456,
+      "step": 1500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7943133696309745e-05,
+      "loss": 0.2753,
+      "step": 1510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.789272030651341e-05,
+      "loss": 0.2619,
+      "step": 1520
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.7842306916717085e-05,
+      "loss": 0.2493,
+      "step": 1530
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.779189352692075e-05,
+      "loss": 0.2547,
+      "step": 1540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.7741480137124425e-05,
+      "loss": 0.2275,
+      "step": 1550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.76910667473281e-05,
+      "loss": 0.2321,
+      "step": 1560
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.7640653357531764e-05,
+      "loss": 0.2563,
+      "step": 1570
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.759023996773543e-05,
+      "loss": 0.2569,
+      "step": 1580
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.75398265779391e-05,
+      "loss": 0.2592,
+      "step": 1590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.748941318814277e-05,
+      "loss": 0.2604,
+      "step": 1600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.7438999798346444e-05,
+      "loss": 0.2228,
+      "step": 1610
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.738858640855011e-05,
+      "loss": 0.2804,
+      "step": 1620
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.7338173018753783e-05,
+      "loss": 0.2436,
+      "step": 1630
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.728775962895745e-05,
+      "loss": 0.2005,
+      "step": 1640
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.723734623916112e-05,
+      "loss": 0.2339,
+      "step": 1650
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9534013605442176,
+      "eval_loss": 0.1212291494011879,
+      "eval_runtime": 89.978,
+      "eval_samples_per_second": 130.699,
+      "eval_steps_per_second": 4.09,
+      "step": 1653
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.7186932849364796e-05,
+      "loss": 0.2426,
+      "step": 1660
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.713651945956846e-05,
+      "loss": 0.2559,
+      "step": 1670
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.7086106069772136e-05,
+      "loss": 0.2467,
+      "step": 1680
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.70356926799758e-05,
+      "loss": 0.2338,
+      "step": 1690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.6985279290179476e-05,
+      "loss": 0.2344,
+      "step": 1700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.693486590038315e-05,
+      "loss": 0.2327,
+      "step": 1710
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.6884452510586816e-05,
+      "loss": 0.2154,
+      "step": 1720
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.683403912079048e-05,
+      "loss": 0.2058,
+      "step": 1730
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.678362573099415e-05,
+      "loss": 0.2511,
+      "step": 1740
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.673321234119782e-05,
+      "loss": 0.2592,
+      "step": 1750
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.6682798951401495e-05,
+      "loss": 0.2454,
+      "step": 1760
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.663238556160516e-05,
+      "loss": 0.2243,
+      "step": 1770
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.6581972171808835e-05,
+      "loss": 0.24,
+      "step": 1780
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.65315587820125e-05,
+      "loss": 0.2145,
+      "step": 1790
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.6481145392216175e-05,
+      "loss": 0.2399,
+      "step": 1800
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.643073200241985e-05,
+      "loss": 0.2172,
+      "step": 1810
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.6380318612623514e-05,
+      "loss": 0.1855,
+      "step": 1820
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.632990522282719e-05,
+      "loss": 0.241,
+      "step": 1830
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.6279491833030854e-05,
+      "loss": 0.2381,
+      "step": 1840
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.622907844323453e-05,
+      "loss": 0.2036,
+      "step": 1850
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.6178665053438194e-05,
+      "loss": 0.2099,
+      "step": 1860
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.612825166364187e-05,
+      "loss": 0.22,
+      "step": 1870
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.6077838273845533e-05,
+      "loss": 0.2231,
+      "step": 1880
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.602742488404921e-05,
+      "loss": 0.223,
+      "step": 1890
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.597701149425287e-05,
+      "loss": 0.2243,
+      "step": 1900
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.5926598104456546e-05,
+      "loss": 0.2182,
+      "step": 1910
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.587618471466021e-05,
+      "loss": 0.2212,
+      "step": 1920
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.5825771324863886e-05,
+      "loss": 0.1938,
+      "step": 1930
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.577535793506756e-05,
+      "loss": 0.1853,
+      "step": 1940
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.5724944545271226e-05,
+      "loss": 0.2206,
+      "step": 1950
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.56745311554749e-05,
+      "loss": 0.2096,
+      "step": 1960
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.5624117765678566e-05,
+      "loss": 0.2272,
+      "step": 1970
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 4.557370437588224e-05,
+      "loss": 0.2211,
+      "step": 1980
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.552329098608591e-05,
+      "loss": 0.2403,
+      "step": 1990
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.547287759628958e-05,
+      "loss": 0.2077,
+      "step": 2000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.5422464206493245e-05,
+      "loss": 0.1759,
+      "step": 2010
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 4.537205081669691e-05,
+      "loss": 0.1996,
+      "step": 2020
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 4.5321637426900585e-05,
+      "loss": 0.2303,
+      "step": 2030
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 4.527122403710426e-05,
+      "loss": 0.2274,
+      "step": 2040
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 4.5220810647307925e-05,
+      "loss": 0.2278,
+      "step": 2050
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 4.51703972575116e-05,
+      "loss": 0.1844,
+      "step": 2060
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.5119983867715264e-05,
+      "loss": 0.1692,
+      "step": 2070
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.506957047791894e-05,
+      "loss": 0.2069,
+      "step": 2080
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.501915708812261e-05,
+      "loss": 0.1937,
+      "step": 2090
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.496874369832628e-05,
+      "loss": 0.2025,
+      "step": 2100
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.491833030852995e-05,
+      "loss": 0.2119,
+      "step": 2110
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.486791691873362e-05,
+      "loss": 0.2081,
+      "step": 2120
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.481750352893729e-05,
+      "loss": 0.1984,
+      "step": 2130
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4767090139140964e-05,
+      "loss": 0.2239,
+      "step": 2140
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.471667674934463e-05,
+      "loss": 0.1887,
+      "step": 2150
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.4666263359548296e-05,
+      "loss": 0.2091,
+      "step": 2160
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.461584996975196e-05,
+      "loss": 0.1692,
+      "step": 2170
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.4565436579955636e-05,
+      "loss": 0.187,
+      "step": 2180
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.451502319015931e-05,
+      "loss": 0.1936,
+      "step": 2190
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.4464609800362976e-05,
+      "loss": 0.1694,
+      "step": 2200
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9602891156462585,
+      "eval_loss": 0.10755564272403717,
+      "eval_runtime": 87.6819,
+      "eval_samples_per_second": 134.121,
+      "eval_steps_per_second": 4.197,
+      "step": 2204
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.441419641056665e-05,
+      "loss": 0.2099,
+      "step": 2210
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.4363783020770316e-05,
+      "loss": 0.191,
+      "step": 2220
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.431336963097399e-05,
+      "loss": 0.2144,
+      "step": 2230
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.426295624117766e-05,
+      "loss": 0.1877,
+      "step": 2240
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.421254285138133e-05,
+      "loss": 0.1611,
+      "step": 2250
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.4162129461585e-05,
+      "loss": 0.2162,
+      "step": 2260
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.411171607178867e-05,
+      "loss": 0.1706,
+      "step": 2270
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.406130268199234e-05,
+      "loss": 0.1692,
+      "step": 2280
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.401088929219601e-05,
+      "loss": 0.1567,
+      "step": 2290
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.396047590239968e-05,
+      "loss": 0.1983,
+      "step": 2300
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.391006251260335e-05,
+      "loss": 0.1839,
+      "step": 2310
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.3859649122807014e-05,
+      "loss": 0.167,
+      "step": 2320
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.380923573301069e-05,
+      "loss": 0.1889,
+      "step": 2330
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.375882234321436e-05,
+      "loss": 0.158,
+      "step": 2340
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.370840895341803e-05,
+      "loss": 0.1842,
+      "step": 2350
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.36579955636217e-05,
+      "loss": 0.1819,
+      "step": 2360
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.360758217382537e-05,
+      "loss": 0.1646,
+      "step": 2370
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.355716878402904e-05,
+      "loss": 0.1926,
+      "step": 2380
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.3506755394232714e-05,
+      "loss": 0.1908,
+      "step": 2390
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.345634200443638e-05,
+      "loss": 0.1871,
+      "step": 2400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.340592861464005e-05,
+      "loss": 0.1672,
+      "step": 2410
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.335551522484372e-05,
+      "loss": 0.1662,
+      "step": 2420
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.330510183504739e-05,
+      "loss": 0.1943,
+      "step": 2430
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.325468844525106e-05,
+      "loss": 0.1692,
+      "step": 2440
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.3204275055454726e-05,
+      "loss": 0.1848,
+      "step": 2450
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.31538616656584e-05,
+      "loss": 0.1749,
+      "step": 2460
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.3103448275862066e-05,
+      "loss": 0.174,
+      "step": 2470
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.305303488606574e-05,
+      "loss": 0.1806,
+      "step": 2480
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.300262149626941e-05,
+      "loss": 0.1768,
+      "step": 2490
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.295220810647308e-05,
+      "loss": 0.2011,
+      "step": 2500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.290179471667675e-05,
+      "loss": 0.1888,
+      "step": 2510
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.285138132688042e-05,
+      "loss": 0.1701,
+      "step": 2520
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.280096793708409e-05,
+      "loss": 0.1777,
+      "step": 2530
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.2750554547287765e-05,
+      "loss": 0.1535,
+      "step": 2540
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.270014115749143e-05,
+      "loss": 0.1755,
+      "step": 2550
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.2649727767695105e-05,
+      "loss": 0.1591,
+      "step": 2560
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.259931437789878e-05,
+      "loss": 0.1649,
+      "step": 2570
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.2548900988102444e-05,
+      "loss": 0.1567,
+      "step": 2580
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.249848759830611e-05,
+      "loss": 0.1858,
+      "step": 2590
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.244807420850978e-05,
+      "loss": 0.1301,
+      "step": 2600
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.239766081871345e-05,
+      "loss": 0.202,
+      "step": 2610
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.2347247428917124e-05,
+      "loss": 0.1552,
+      "step": 2620
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.229683403912079e-05,
+      "loss": 0.1754,
+      "step": 2630
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.2246420649324464e-05,
+      "loss": 0.1661,
+      "step": 2640
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 4.219600725952813e-05,
+      "loss": 0.1688,
+      "step": 2650
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 4.21455938697318e-05,
+      "loss": 0.1806,
+      "step": 2660
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 4.2095180479935477e-05,
+      "loss": 0.1675,
+      "step": 2670
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.204476709013914e-05,
+      "loss": 0.19,
+      "step": 2680
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.1994353700342816e-05,
+      "loss": 0.1632,
+      "step": 2690
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.194394031054648e-05,
+      "loss": 0.196,
+      "step": 2700
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 4.1893526920750156e-05,
+      "loss": 0.1557,
+      "step": 2710
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.184311353095382e-05,
+      "loss": 0.1644,
+      "step": 2720
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.1792700141157496e-05,
+      "loss": 0.171,
+      "step": 2730
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 4.174228675136116e-05,
+      "loss": 0.1626,
+      "step": 2740
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.169187336156483e-05,
+      "loss": 0.1715,
+      "step": 2750
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9692176870748299,
+      "eval_loss": 0.08295176178216934,
+      "eval_runtime": 88.1408,
+      "eval_samples_per_second": 133.423,
+      "eval_steps_per_second": 4.175,
+      "step": 2755
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 4.16414599717685e-05,
+      "loss": 0.1657,
+      "step": 2760
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 4.1591046581972175e-05,
+      "loss": 0.1604,
+      "step": 2770
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 4.154063319217584e-05,
+      "loss": 0.1604,
+      "step": 2780
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 4.1490219802379515e-05,
+      "loss": 0.1509,
+      "step": 2790
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 4.143980641258318e-05,
+      "loss": 0.1605,
+      "step": 2800
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 4.1389393022786855e-05,
+      "loss": 0.1911,
+      "step": 2810
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 4.133897963299053e-05,
+      "loss": 0.169,
+      "step": 2820
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 4.1288566243194194e-05,
+      "loss": 0.1814,
+      "step": 2830
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 4.123815285339787e-05,
+      "loss": 0.1571,
+      "step": 2840
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 4.1187739463601534e-05,
+      "loss": 0.1743,
+      "step": 2850
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 4.113732607380521e-05,
+      "loss": 0.1688,
+      "step": 2860
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 4.1086912684008874e-05,
+      "loss": 0.1645,
+      "step": 2870
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 4.103649929421255e-05,
+      "loss": 0.1556,
+      "step": 2880
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 4.0986085904416214e-05,
+      "loss": 0.1721,
+      "step": 2890
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 4.093567251461988e-05,
+      "loss": 0.1665,
+      "step": 2900
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 4.088525912482355e-05,
+      "loss": 0.1398,
+      "step": 2910
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 4.0834845735027227e-05,
+      "loss": 0.1631,
+      "step": 2920
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 4.078443234523089e-05,
+      "loss": 0.1364,
+      "step": 2930
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 4.0734018955434566e-05,
+      "loss": 0.1551,
+      "step": 2940
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 4.068360556563823e-05,
+      "loss": 0.1696,
+      "step": 2950
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 4.0633192175841906e-05,
+      "loss": 0.1638,
+      "step": 2960
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 4.058277878604558e-05,
+      "loss": 0.1479,
+      "step": 2970
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 4.0532365396249246e-05,
+      "loss": 0.1291,
+      "step": 2980
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 4.048195200645292e-05,
+      "loss": 0.1694,
+      "step": 2990
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 4.0431538616656585e-05,
+      "loss": 0.1659,
+      "step": 3000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 4.038112522686026e-05,
+      "loss": 0.159,
+      "step": 3010
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 4.0330711837063925e-05,
+      "loss": 0.1535,
+      "step": 3020
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 4.028029844726759e-05,
+      "loss": 0.1589,
+      "step": 3030
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 4.0229885057471265e-05,
+      "loss": 0.1556,
+      "step": 3040
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 4.017947166767493e-05,
+      "loss": 0.1625,
+      "step": 3050
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 4.0129058277878605e-05,
+      "loss": 0.1335,
+      "step": 3060
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 4.007864488808228e-05,
+      "loss": 0.1302,
+      "step": 3070
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 4.0028231498285944e-05,
+      "loss": 0.1851,
+      "step": 3080
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 3.997781810848962e-05,
+      "loss": 0.159,
+      "step": 3090
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 3.9927404718693284e-05,
+      "loss": 0.1508,
+      "step": 3100
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 3.987699132889696e-05,
+      "loss": 0.1418,
+      "step": 3110
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 3.982657793910063e-05,
+      "loss": 0.1691,
+      "step": 3120
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 3.97761645493043e-05,
+      "loss": 0.1386,
+      "step": 3130
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 3.972575115950797e-05,
+      "loss": 0.1399,
+      "step": 3140
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 3.967533776971164e-05,
+      "loss": 0.1337,
+      "step": 3150
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 3.962492437991531e-05,
+      "loss": 0.165,
+      "step": 3160
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 3.9574510990118977e-05,
+      "loss": 0.1495,
+      "step": 3170
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 3.952409760032264e-05,
+      "loss": 0.1552,
+      "step": 3180
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.1467,
+      "step": 3190
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 3.942327082072998e-05,
+      "loss": 0.1539,
+      "step": 3200
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 3.9372857430933656e-05,
+      "loss": 0.1476,
+      "step": 3210
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 3.932244404113733e-05,
+      "loss": 0.1684,
+      "step": 3220
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 3.9272030651340996e-05,
+      "loss": 0.146,
+      "step": 3230
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 3.922161726154467e-05,
+      "loss": 0.1522,
+      "step": 3240
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 3.917120387174834e-05,
+      "loss": 0.1433,
+      "step": 3250
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 3.912079048195201e-05,
+      "loss": 0.1394,
+      "step": 3260
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 3.907037709215568e-05,
+      "loss": 0.1515,
+      "step": 3270
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 3.901996370235935e-05,
+      "loss": 0.1527,
+      "step": 3280
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 3.896955031256302e-05,
+      "loss": 0.1354,
+      "step": 3290
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 3.891913692276669e-05,
+      "loss": 0.1339,
+      "step": 3300
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9761904761904762,
+      "eval_loss": 0.06736525893211365,
+      "eval_runtime": 87.7746,
+      "eval_samples_per_second": 133.98,
+      "eval_steps_per_second": 4.193,
+      "step": 3306
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.886872353297036e-05,
+      "loss": 0.1747,
+      "step": 3310
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 3.881831014317403e-05,
+      "loss": 0.1727,
+      "step": 3320
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 3.8767896753377694e-05,
+      "loss": 0.1411,
+      "step": 3330
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 3.871748336358137e-05,
+      "loss": 0.1277,
+      "step": 3340
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 3.866706997378504e-05,
+      "loss": 0.1319,
+      "step": 3350
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 3.861665658398871e-05,
+      "loss": 0.1431,
+      "step": 3360
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 3.856624319419238e-05,
+      "loss": 0.1552,
+      "step": 3370
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 3.851582980439605e-05,
+      "loss": 0.152,
+      "step": 3380
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 3.846541641459972e-05,
+      "loss": 0.1272,
+      "step": 3390
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 3.8415003024803394e-05,
+      "loss": 0.1386,
+      "step": 3400
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 3.836458963500706e-05,
+      "loss": 0.131,
+      "step": 3410
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 3.831417624521073e-05,
+      "loss": 0.1536,
+      "step": 3420
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 3.82637628554144e-05,
+      "loss": 0.1405,
+      "step": 3430
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 3.821334946561807e-05,
+      "loss": 0.1458,
+      "step": 3440
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 3.816293607582174e-05,
+      "loss": 0.1206,
+      "step": 3450
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 3.8112522686025406e-05,
+      "loss": 0.1698,
+      "step": 3460
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 3.806210929622908e-05,
+      "loss": 0.1422,
+      "step": 3470
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 3.8011695906432746e-05,
+      "loss": 0.1396,
+      "step": 3480
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 3.796128251663642e-05,
+      "loss": 0.1377,
+      "step": 3490
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 3.791086912684009e-05,
+      "loss": 0.1179,
+      "step": 3500
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 3.786045573704376e-05,
+      "loss": 0.1545,
+      "step": 3510
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 3.781004234724743e-05,
+      "loss": 0.1277,
+      "step": 3520
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 3.77596289574511e-05,
+      "loss": 0.1456,
+      "step": 3530
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 3.770921556765477e-05,
+      "loss": 0.1368,
+      "step": 3540
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 3.7658802177858445e-05,
+      "loss": 0.1565,
+      "step": 3550
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 3.760838878806211e-05,
+      "loss": 0.1659,
+      "step": 3560
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 3.7557975398265785e-05,
+      "loss": 0.1201,
+      "step": 3570
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 3.750756200846945e-05,
+      "loss": 0.1354,
+      "step": 3580
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 3.7457148618673124e-05,
+      "loss": 0.1398,
+      "step": 3590
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 3.740673522887679e-05,
+      "loss": 0.1523,
+      "step": 3600
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 3.735632183908046e-05,
+      "loss": 0.1263,
+      "step": 3610
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 3.730590844928413e-05,
+      "loss": 0.1408,
+      "step": 3620
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 3.72554950594878e-05,
+      "loss": 0.1452,
+      "step": 3630
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 3.720508166969147e-05,
+      "loss": 0.1646,
+      "step": 3640
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 3.7154668279895144e-05,
+      "loss": 0.1152,
+      "step": 3650
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 3.710425489009881e-05,
+      "loss": 0.1287,
+      "step": 3660
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 3.705384150030248e-05,
+      "loss": 0.1456,
+      "step": 3670
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 3.700342811050615e-05,
+      "loss": 0.1374,
+      "step": 3680
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 3.695301472070982e-05,
+      "loss": 0.1625,
+      "step": 3690
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 3.6902601330913496e-05,
+      "loss": 0.1331,
+      "step": 3700
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 3.685218794111716e-05,
+      "loss": 0.1161,
+      "step": 3710
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 3.6801774551320836e-05,
+      "loss": 0.1443,
+      "step": 3720
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 3.67513611615245e-05,
+      "loss": 0.1478,
+      "step": 3730
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 3.6700947771728176e-05,
+      "loss": 0.1448,
+      "step": 3740
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 3.665053438193184e-05,
+      "loss": 0.1346,
+      "step": 3750
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 3.660012099213551e-05,
+      "loss": 0.1034,
+      "step": 3760
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 3.654970760233918e-05,
+      "loss": 0.138,
+      "step": 3770
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 3.649929421254285e-05,
+      "loss": 0.1162,
+      "step": 3780
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 3.644888082274652e-05,
+      "loss": 0.1288,
+      "step": 3790
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 3.6398467432950195e-05,
+      "loss": 0.1164,
+      "step": 3800
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 3.634805404315386e-05,
+      "loss": 0.1238,
+      "step": 3810
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 3.6297640653357535e-05,
+      "loss": 0.1308,
+      "step": 3820
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 3.62472272635612e-05,
+      "loss": 0.136,
+      "step": 3830
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 3.6196813873764874e-05,
+      "loss": 0.1397,
+      "step": 3840
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 3.614640048396855e-05,
+      "loss": 0.1527,
+      "step": 3850
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9790816326530613,
+      "eval_loss": 0.055576082319021225,
+      "eval_runtime": 85.539,
+      "eval_samples_per_second": 137.481,
+      "eval_steps_per_second": 4.302,
+      "step": 3857
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.6095987094172214e-05,
+      "loss": 0.1517,
+      "step": 3860
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 3.604557370437589e-05,
+      "loss": 0.1541,
+      "step": 3870
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 3.5995160314579554e-05,
+      "loss": 0.1417,
+      "step": 3880
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 3.594474692478322e-05,
+      "loss": 0.142,
+      "step": 3890
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 3.5894333534986894e-05,
+      "loss": 0.131,
+      "step": 3900
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 3.584392014519056e-05,
+      "loss": 0.1187,
+      "step": 3910
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 3.579350675539423e-05,
+      "loss": 0.1444,
+      "step": 3920
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 3.5743093365597907e-05,
+      "loss": 0.1289,
+      "step": 3930
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.569267997580157e-05,
+      "loss": 0.1441,
+      "step": 3940
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.5642266586005246e-05,
+      "loss": 0.1278,
+      "step": 3950
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.559185319620891e-05,
+      "loss": 0.1272,
+      "step": 3960
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.5541439806412586e-05,
+      "loss": 0.1091,
+      "step": 3970
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.549102641661626e-05,
+      "loss": 0.1025,
+      "step": 3980
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.5440613026819926e-05,
+      "loss": 0.1204,
+      "step": 3990
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.53901996370236e-05,
+      "loss": 0.1171,
+      "step": 4000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.5339786247227266e-05,
+      "loss": 0.1314,
+      "step": 4010
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.528937285743094e-05,
+      "loss": 0.1315,
+      "step": 4020
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.5238959467634605e-05,
+      "loss": 0.1258,
+      "step": 4030
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.518854607783827e-05,
+      "loss": 0.1206,
+      "step": 4040
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.5138132688041945e-05,
+      "loss": 0.1189,
+      "step": 4050
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.508771929824561e-05,
+      "loss": 0.1413,
+      "step": 4060
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.5037305908449285e-05,
+      "loss": 0.118,
+      "step": 4070
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.498689251865296e-05,
+      "loss": 0.0862,
+      "step": 4080
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.4936479128856624e-05,
+      "loss": 0.1454,
+      "step": 4090
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.48860657390603e-05,
+      "loss": 0.1447,
+      "step": 4100
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.4835652349263964e-05,
+      "loss": 0.1035,
+      "step": 4110
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.478523895946764e-05,
+      "loss": 0.1218,
+      "step": 4120
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.473482556967131e-05,
+      "loss": 0.126,
+      "step": 4130
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.468441217987498e-05,
+      "loss": 0.1162,
+      "step": 4140
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.463399879007865e-05,
+      "loss": 0.115,
+      "step": 4150
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 3.458358540028232e-05,
+      "loss": 0.1203,
+      "step": 4160
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 3.453317201048599e-05,
+      "loss": 0.1421,
+      "step": 4170
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 3.4482758620689657e-05,
+      "loss": 0.1266,
+      "step": 4180
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 3.443234523089332e-05,
+      "loss": 0.1264,
+      "step": 4190
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 3.4381931841096996e-05,
+      "loss": 0.1243,
+      "step": 4200
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 3.433151845130066e-05,
+      "loss": 0.1331,
+      "step": 4210
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 3.4281105061504336e-05,
+      "loss": 0.1357,
+      "step": 4220
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 3.423069167170801e-05,
+      "loss": 0.1223,
+      "step": 4230
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.4180278281911676e-05,
+      "loss": 0.115,
+      "step": 4240
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.412986489211535e-05,
+      "loss": 0.1169,
+      "step": 4250
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.4079451502319016e-05,
+      "loss": 0.1202,
+      "step": 4260
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.402903811252269e-05,
+      "loss": 0.1136,
+      "step": 4270
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.397862472272636e-05,
+      "loss": 0.105,
+      "step": 4280
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 3.392821133293003e-05,
+      "loss": 0.1189,
+      "step": 4290
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 3.38777979431337e-05,
+      "loss": 0.1271,
+      "step": 4300
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 3.382738455333737e-05,
+      "loss": 0.1212,
+      "step": 4310
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 3.3776971163541035e-05,
+      "loss": 0.1272,
+      "step": 4320
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 3.372655777374471e-05,
+      "loss": 0.123,
+      "step": 4330
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 3.3676144383948374e-05,
+      "loss": 0.1026,
+      "step": 4340
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 3.362573099415205e-05,
+      "loss": 0.1049,
+      "step": 4350
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 3.3575317604355714e-05,
+      "loss": 0.1249,
+      "step": 4360
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 3.352490421455939e-05,
+      "loss": 0.12,
+      "step": 4370
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 3.347449082476306e-05,
+      "loss": 0.112,
+      "step": 4380
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 3.342407743496673e-05,
+      "loss": 0.1355,
+      "step": 4390
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 3.33736640451704e-05,
+      "loss": 0.1214,
+      "step": 4400
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.9832482993197279,
+      "eval_loss": 0.04550003260374069,
+      "eval_runtime": 87.0544,
+      "eval_samples_per_second": 135.088,
+      "eval_steps_per_second": 4.227,
+      "step": 4408
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 3.332325065537407e-05,
+      "loss": 0.1301,
+      "step": 4410
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 3.327283726557774e-05,
+      "loss": 0.1193,
+      "step": 4420
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 3.322242387578141e-05,
+      "loss": 0.1349,
+      "step": 4430
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 3.317201048598508e-05,
+      "loss": 0.1043,
+      "step": 4440
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 3.312159709618875e-05,
+      "loss": 0.1367,
+      "step": 4450
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 3.307118370639242e-05,
+      "loss": 0.1249,
+      "step": 4460
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 3.3020770316596086e-05,
+      "loss": 0.1303,
+      "step": 4470
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 3.297035692679976e-05,
+      "loss": 0.1064,
+      "step": 4480
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 3.2919943537003426e-05,
+      "loss": 0.1058,
+      "step": 4490
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 3.28695301472071e-05,
+      "loss": 0.1168,
+      "step": 4500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 3.2819116757410766e-05,
+      "loss": 0.1123,
+      "step": 4510
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 3.276870336761444e-05,
+      "loss": 0.1178,
+      "step": 4520
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 3.271828997781811e-05,
+      "loss": 0.119,
+      "step": 4530
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 3.266787658802178e-05,
+      "loss": 0.1193,
+      "step": 4540
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 3.261746319822545e-05,
+      "loss": 0.1086,
+      "step": 4550
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 3.256704980842912e-05,
+      "loss": 0.1282,
+      "step": 4560
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 3.251663641863279e-05,
+      "loss": 0.1266,
+      "step": 4570
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 3.2466223028836465e-05,
+      "loss": 0.1362,
+      "step": 4580
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 3.241580963904013e-05,
+      "loss": 0.1099,
+      "step": 4590
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 3.2365396249243804e-05,
+      "loss": 0.1363,
+      "step": 4600
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 3.231498285944747e-05,
+      "loss": 0.1077,
+      "step": 4610
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 3.226456946965114e-05,
+      "loss": 0.1167,
+      "step": 4620
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 3.221415607985481e-05,
+      "loss": 0.1495,
+      "step": 4630
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 3.216374269005848e-05,
+      "loss": 0.1075,
+      "step": 4640
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 3.211332930026215e-05,
+      "loss": 0.1139,
+      "step": 4650
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 3.2062915910465824e-05,
+      "loss": 0.1296,
+      "step": 4660
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 3.201250252066949e-05,
+      "loss": 0.1011,
+      "step": 4670
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 3.196208913087316e-05,
+      "loss": 0.1158,
+      "step": 4680
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 3.191167574107683e-05,
+      "loss": 0.1071,
+      "step": 4690
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 3.18612623512805e-05,
+      "loss": 0.1237,
+      "step": 4700
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 3.1810848961484176e-05,
+      "loss": 0.1136,
+      "step": 4710
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 3.176043557168784e-05,
+      "loss": 0.1135,
+      "step": 4720
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 3.1710022181891516e-05,
+      "loss": 0.09,
+      "step": 4730
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 3.165960879209518e-05,
+      "loss": 0.1141,
+      "step": 4740
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 3.160919540229885e-05,
+      "loss": 0.0941,
+      "step": 4750
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 3.155878201250252e-05,
+      "loss": 0.1283,
+      "step": 4760
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 3.150836862270619e-05,
+      "loss": 0.1499,
+      "step": 4770
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 3.145795523290986e-05,
+      "loss": 0.111,
+      "step": 4780
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 3.140754184311353e-05,
+      "loss": 0.0984,
+      "step": 4790
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 3.13571284533172e-05,
+      "loss": 0.1128,
+      "step": 4800
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 3.1306715063520875e-05,
+      "loss": 0.1191,
+      "step": 4810
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 3.125630167372454e-05,
+      "loss": 0.1353,
+      "step": 4820
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 3.1205888283928215e-05,
+      "loss": 0.1481,
+      "step": 4830
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 3.115547489413188e-05,
+      "loss": 0.1245,
+      "step": 4840
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 3.1105061504335554e-05,
+      "loss": 0.1093,
+      "step": 4850
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 3.105464811453923e-05,
+      "loss": 0.1188,
+      "step": 4860
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 3.1004234724742894e-05,
+      "loss": 0.1088,
+      "step": 4870
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 3.095382133494657e-05,
+      "loss": 0.127,
+      "step": 4880
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 3.0903407945150234e-05,
+      "loss": 0.1311,
+      "step": 4890
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 3.08529945553539e-05,
+      "loss": 0.1191,
+      "step": 4900
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 3.0802581165557574e-05,
+      "loss": 0.1241,
+      "step": 4910
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 3.075216777576124e-05,
+      "loss": 0.1085,
+      "step": 4920
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 3.0701754385964913e-05,
+      "loss": 0.0979,
+      "step": 4930
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 3.065134099616858e-05,
+      "loss": 0.1216,
+      "step": 4940
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 3.060092760637225e-05,
+      "loss": 0.1062,
+      "step": 4950
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9829081632653062,
+      "eval_loss": 0.046558648347854614,
+      "eval_runtime": 88.0948,
+      "eval_samples_per_second": 133.493,
+      "eval_steps_per_second": 4.177,
+      "step": 4959
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 3.0550514216575926e-05,
+      "loss": 0.1116,
+      "step": 4960
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 3.0500100826779593e-05,
+      "loss": 0.1042,
+      "step": 4970
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 3.0449687436983266e-05,
+      "loss": 0.0936,
+      "step": 4980
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 3.0399274047186933e-05,
+      "loss": 0.0999,
+      "step": 4990
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 3.0348860657390606e-05,
+      "loss": 0.1071,
+      "step": 5000
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 3.0298447267594276e-05,
+      "loss": 0.1024,
+      "step": 5010
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 3.0248033877797942e-05,
+      "loss": 0.0965,
+      "step": 5020
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.0197620488001615e-05,
+      "loss": 0.0989,
+      "step": 5030
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.0147207098205282e-05,
+      "loss": 0.0989,
+      "step": 5040
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.0096793708408955e-05,
+      "loss": 0.1078,
+      "step": 5050
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.004638031861263e-05,
+      "loss": 0.1469,
+      "step": 5060
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 2.9995966928816295e-05,
+      "loss": 0.1222,
+      "step": 5070
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 2.9945553539019965e-05,
+      "loss": 0.1176,
+      "step": 5080
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 2.989514014922363e-05,
+      "loss": 0.0941,
+      "step": 5090
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.9844726759427304e-05,
+      "loss": 0.1035,
+      "step": 5100
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.9794313369630978e-05,
+      "loss": 0.0865,
+      "step": 5110
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.9743899979834644e-05,
+      "loss": 0.1118,
+      "step": 5120
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.9693486590038317e-05,
+      "loss": 0.1015,
+      "step": 5130
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.9643073200241984e-05,
+      "loss": 0.1055,
+      "step": 5140
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.9592659810445654e-05,
+      "loss": 0.1184,
+      "step": 5150
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.9542246420649327e-05,
+      "loss": 0.116,
+      "step": 5160
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.9491833030852994e-05,
+      "loss": 0.1119,
+      "step": 5170
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.9441419641056667e-05,
+      "loss": 0.0995,
+      "step": 5180
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.9391006251260333e-05,
+      "loss": 0.115,
+      "step": 5190
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.9340592861464007e-05,
+      "loss": 0.102,
+      "step": 5200
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.9290179471667676e-05,
+      "loss": 0.1157,
+      "step": 5210
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.9239766081871346e-05,
+      "loss": 0.1056,
+      "step": 5220
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.9189352692075016e-05,
+      "loss": 0.1278,
+      "step": 5230
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.9138939302278683e-05,
+      "loss": 0.0994,
+      "step": 5240
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.9088525912482356e-05,
+      "loss": 0.0926,
+      "step": 5250
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.903811252268603e-05,
+      "loss": 0.0937,
+      "step": 5260
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.8987699132889696e-05,
+      "loss": 0.1062,
+      "step": 5270
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.893728574309337e-05,
+      "loss": 0.1211,
+      "step": 5280
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.888687235329704e-05,
+      "loss": 0.1128,
+      "step": 5290
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 2.8836458963500705e-05,
+      "loss": 0.0928,
+      "step": 5300
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 2.878604557370438e-05,
+      "loss": 0.1199,
+      "step": 5310
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 2.8735632183908045e-05,
+      "loss": 0.1051,
+      "step": 5320
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 2.8685218794111718e-05,
+      "loss": 0.1167,
+      "step": 5330
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 2.863480540431539e-05,
+      "loss": 0.1156,
+      "step": 5340
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 2.8584392014519058e-05,
+      "loss": 0.102,
+      "step": 5350
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 2.8533978624722728e-05,
+      "loss": 0.0978,
+      "step": 5360
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 2.8483565234926398e-05,
+      "loss": 0.094,
+      "step": 5370
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 2.8433151845130067e-05,
+      "loss": 0.101,
+      "step": 5380
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.838273845533374e-05,
+      "loss": 0.0907,
+      "step": 5390
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.8332325065537407e-05,
+      "loss": 0.0975,
+      "step": 5400
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 2.828191167574108e-05,
+      "loss": 0.122,
+      "step": 5410
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 2.8231498285944747e-05,
+      "loss": 0.1171,
+      "step": 5420
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 2.818108489614842e-05,
+      "loss": 0.1094,
+      "step": 5430
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 2.813067150635209e-05,
+      "loss": 0.0983,
+      "step": 5440
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 2.8080258116555757e-05,
+      "loss": 0.0873,
+      "step": 5450
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 2.802984472675943e-05,
+      "loss": 0.1062,
+      "step": 5460
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.7979431336963096e-05,
+      "loss": 0.1126,
+      "step": 5470
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.792901794716677e-05,
+      "loss": 0.1219,
+      "step": 5480
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.7878604557370443e-05,
+      "loss": 0.0994,
+      "step": 5490
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 2.782819116757411e-05,
+      "loss": 0.0926,
+      "step": 5500
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.0974,
+      "step": 5510
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.9848639455782313,
+      "eval_loss": 0.04026130214333534,
+      "eval_runtime": 87.861,
+      "eval_samples_per_second": 133.848,
+      "eval_steps_per_second": 4.188,
+      "step": 5510
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 2.7727364387981446e-05,
+      "loss": 0.1126,
+      "step": 5520
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 2.767695099818512e-05,
+      "loss": 0.0963,
+      "step": 5530
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 2.7626537608388792e-05,
+      "loss": 0.0916,
+      "step": 5540
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 2.757612421859246e-05,
+      "loss": 0.1058,
+      "step": 5550
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 2.7525710828796132e-05,
+      "loss": 0.0979,
+      "step": 5560
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 2.7475297438999798e-05,
+      "loss": 0.0975,
+      "step": 5570
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 2.7424884049203468e-05,
+      "loss": 0.1101,
+      "step": 5580
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 2.737447065940714e-05,
+      "loss": 0.0944,
+      "step": 5590
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 2.7324057269610808e-05,
+      "loss": 0.1149,
+      "step": 5600
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 2.727364387981448e-05,
+      "loss": 0.1024,
+      "step": 5610
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 2.7223230490018148e-05,
+      "loss": 0.0832,
+      "step": 5620
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 2.717281710022182e-05,
+      "loss": 0.1002,
+      "step": 5630
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 2.7122403710425494e-05,
+      "loss": 0.1248,
+      "step": 5640
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 2.707199032062916e-05,
+      "loss": 0.0897,
+      "step": 5650
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 2.702157693083283e-05,
+      "loss": 0.1105,
+      "step": 5660
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 2.6971163541036497e-05,
+      "loss": 0.0862,
+      "step": 5670
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 2.692075015124017e-05,
+      "loss": 0.1075,
+      "step": 5680
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 2.6870336761443843e-05,
+      "loss": 0.1038,
+      "step": 5690
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 2.681992337164751e-05,
+      "loss": 0.1037,
+      "step": 5700
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 2.6769509981851183e-05,
+      "loss": 0.1005,
+      "step": 5710
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 2.671909659205485e-05,
+      "loss": 0.1124,
+      "step": 5720
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 2.666868320225852e-05,
+      "loss": 0.1075,
+      "step": 5730
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 2.6618269812462193e-05,
+      "loss": 0.1079,
+      "step": 5740
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 2.656785642266586e-05,
+      "loss": 0.0876,
+      "step": 5750
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 2.6517443032869533e-05,
+      "loss": 0.0917,
+      "step": 5760
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 2.64670296430732e-05,
+      "loss": 0.0873,
+      "step": 5770
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 2.6416616253276872e-05,
+      "loss": 0.0958,
+      "step": 5780
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 2.6366202863480542e-05,
+      "loss": 0.1083,
+      "step": 5790
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.1011,
+      "step": 5800
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 2.6265376083887882e-05,
+      "loss": 0.0949,
+      "step": 5810
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 2.621496269409155e-05,
+      "loss": 0.0828,
+      "step": 5820
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 2.616454930429522e-05,
+      "loss": 0.0864,
+      "step": 5830
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 2.6114135914498895e-05,
+      "loss": 0.085,
+      "step": 5840
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 2.606372252470256e-05,
+      "loss": 0.0956,
+      "step": 5850
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 2.6013309134906235e-05,
+      "loss": 0.0919,
+      "step": 5860
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 2.59628957451099e-05,
+      "loss": 0.1228,
+      "step": 5870
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 2.591248235531357e-05,
+      "loss": 0.0908,
+      "step": 5880
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 2.5862068965517244e-05,
+      "loss": 0.0997,
+      "step": 5890
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 2.581165557572091e-05,
+      "loss": 0.0946,
+      "step": 5900
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 2.5761242185924584e-05,
+      "loss": 0.1115,
+      "step": 5910
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 2.571082879612825e-05,
+      "loss": 0.0874,
+      "step": 5920
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 2.5660415406331924e-05,
+      "loss": 0.0704,
+      "step": 5930
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 2.5610002016535593e-05,
+      "loss": 0.1027,
+      "step": 5940
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 2.555958862673926e-05,
+      "loss": 0.0952,
+      "step": 5950
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 2.5509175236942933e-05,
+      "loss": 0.1035,
+      "step": 5960
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 2.5458761847146606e-05,
+      "loss": 0.0798,
+      "step": 5970
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 2.5408348457350273e-05,
+      "loss": 0.0982,
+      "step": 5980
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 2.5357935067553946e-05,
+      "loss": 0.0973,
+      "step": 5990
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 2.5307521677757613e-05,
+      "loss": 0.1048,
+      "step": 6000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 2.5257108287961283e-05,
+      "loss": 0.0766,
+      "step": 6010
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 2.5206694898164956e-05,
+      "loss": 0.0851,
+      "step": 6020
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 2.5156281508368622e-05,
+      "loss": 0.1029,
+      "step": 6030
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 2.5105868118572295e-05,
+      "loss": 0.0887,
+      "step": 6040
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 2.5055454728775962e-05,
+      "loss": 0.0953,
+      "step": 6050
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 2.5005041338979635e-05,
+      "loss": 0.0875,
+      "step": 6060
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.985969387755102,
+      "eval_loss": 0.038536228239536285,
+      "eval_runtime": 86.7471,
+      "eval_samples_per_second": 135.566,
+      "eval_steps_per_second": 4.242,
+      "step": 6061
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 2.4954627949183305e-05,
+      "loss": 0.1304,
+      "step": 6070
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 2.4904214559386975e-05,
+      "loss": 0.101,
+      "step": 6080
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 2.485380116959064e-05,
+      "loss": 0.1011,
+      "step": 6090
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 2.4803387779794315e-05,
+      "loss": 0.0824,
+      "step": 6100
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 2.4752974389997985e-05,
+      "loss": 0.0836,
+      "step": 6110
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 2.4702561000201654e-05,
+      "loss": 0.0978,
+      "step": 6120
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 2.4652147610405324e-05,
+      "loss": 0.1097,
+      "step": 6130
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 2.4601734220608998e-05,
+      "loss": 0.0909,
+      "step": 6140
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 2.4551320830812667e-05,
+      "loss": 0.0958,
+      "step": 6150
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 2.4500907441016334e-05,
+      "loss": 0.094,
+      "step": 6160
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 2.4450494051220004e-05,
+      "loss": 0.0853,
+      "step": 6170
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 2.4400080661423674e-05,
+      "loss": 0.0936,
+      "step": 6180
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 2.4349667271627347e-05,
+      "loss": 0.1244,
+      "step": 6190
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 2.4299253881831017e-05,
+      "loss": 0.1087,
+      "step": 6200
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 2.4248840492034687e-05,
+      "loss": 0.1012,
+      "step": 6210
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 2.4198427102238356e-05,
+      "loss": 0.0902,
+      "step": 6220
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 2.4148013712442026e-05,
+      "loss": 0.0852,
+      "step": 6230
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 2.4097600322645696e-05,
+      "loss": 0.095,
+      "step": 6240
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 2.4047186932849366e-05,
+      "loss": 0.1148,
+      "step": 6250
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 2.3996773543053036e-05,
+      "loss": 0.0829,
+      "step": 6260
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 2.3946360153256706e-05,
+      "loss": 0.0898,
+      "step": 6270
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 2.3895946763460376e-05,
+      "loss": 0.0887,
+      "step": 6280
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 2.384553337366405e-05,
+      "loss": 0.0979,
+      "step": 6290
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 2.3795119983867715e-05,
+      "loss": 0.0848,
+      "step": 6300
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 2.3744706594071385e-05,
+      "loss": 0.11,
+      "step": 6310
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 2.3694293204275055e-05,
+      "loss": 0.1022,
+      "step": 6320
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 2.3643879814478725e-05,
+      "loss": 0.1009,
+      "step": 6330
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 2.3593466424682398e-05,
+      "loss": 0.0976,
+      "step": 6340
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 2.3543053034886068e-05,
+      "loss": 0.0904,
+      "step": 6350
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 2.3492639645089738e-05,
+      "loss": 0.0976,
+      "step": 6360
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 2.3442226255293408e-05,
+      "loss": 0.1009,
+      "step": 6370
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 2.3391812865497074e-05,
+      "loss": 0.0895,
+      "step": 6380
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 2.3341399475700748e-05,
+      "loss": 0.0749,
+      "step": 6390
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 2.3290986085904417e-05,
+      "loss": 0.0854,
+      "step": 6400
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 2.3240572696108087e-05,
+      "loss": 0.0847,
+      "step": 6410
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 2.3190159306311757e-05,
+      "loss": 0.0876,
+      "step": 6420
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 2.3139745916515427e-05,
+      "loss": 0.1056,
+      "step": 6430
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 2.3089332526719097e-05,
+      "loss": 0.0808,
+      "step": 6440
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 2.3038919136922767e-05,
+      "loss": 0.0776,
+      "step": 6450
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 2.2988505747126437e-05,
+      "loss": 0.0701,
+      "step": 6460
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 2.2938092357330106e-05,
+      "loss": 0.0747,
+      "step": 6470
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 2.288767896753378e-05,
+      "loss": 0.0886,
+      "step": 6480
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 2.283726557773745e-05,
+      "loss": 0.0982,
+      "step": 6490
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 2.278685218794112e-05,
+      "loss": 0.0818,
+      "step": 6500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 2.273643879814479e-05,
+      "loss": 0.1133,
+      "step": 6510
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 2.2686025408348456e-05,
+      "loss": 0.1023,
+      "step": 6520
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 2.263561201855213e-05,
+      "loss": 0.0821,
+      "step": 6530
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 2.25851986287558e-05,
+      "loss": 0.1041,
+      "step": 6540
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 2.253478523895947e-05,
+      "loss": 0.0906,
+      "step": 6550
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 2.248437184916314e-05,
+      "loss": 0.0985,
+      "step": 6560
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 2.243395845936681e-05,
+      "loss": 0.0835,
+      "step": 6570
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 2.2383545069570482e-05,
+      "loss": 0.0922,
+      "step": 6580
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 2.2333131679774148e-05,
+      "loss": 0.0765,
+      "step": 6590
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 2.2282718289977818e-05,
+      "loss": 0.067,
+      "step": 6600
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 2.2232304900181488e-05,
+      "loss": 0.0992,
+      "step": 6610
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.9869897959183673,
+      "eval_loss": 0.03759635239839554,
+      "eval_runtime": 85.2108,
+      "eval_samples_per_second": 138.011,
+      "eval_steps_per_second": 4.319,
+      "step": 6612
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.2181891510385158e-05,
+      "loss": 0.1131,
+      "step": 6620
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.213147812058883e-05,
+      "loss": 0.0865,
+      "step": 6630
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.20810647307925e-05,
+      "loss": 0.0688,
+      "step": 6640
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.203065134099617e-05,
+      "loss": 0.1034,
+      "step": 6650
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.198023795119984e-05,
+      "loss": 0.0796,
+      "step": 6660
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.1929824561403507e-05,
+      "loss": 0.0899,
+      "step": 6670
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.187941117160718e-05,
+      "loss": 0.0838,
+      "step": 6680
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.182899778181085e-05,
+      "loss": 0.074,
+      "step": 6690
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.177858439201452e-05,
+      "loss": 0.0825,
+      "step": 6700
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.172817100221819e-05,
+      "loss": 0.1185,
+      "step": 6710
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.167775761242186e-05,
+      "loss": 0.0779,
+      "step": 6720
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.162734422262553e-05,
+      "loss": 0.0763,
+      "step": 6730
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.15769308328292e-05,
+      "loss": 0.0988,
+      "step": 6740
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.152651744303287e-05,
+      "loss": 0.091,
+      "step": 6750
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.147610405323654e-05,
+      "loss": 0.0736,
+      "step": 6760
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.142569066344021e-05,
+      "loss": 0.0974,
+      "step": 6770
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.1375277273643882e-05,
+      "loss": 0.1104,
+      "step": 6780
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.1324863883847552e-05,
+      "loss": 0.1004,
+      "step": 6790
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.1274450494051222e-05,
+      "loss": 0.077,
+      "step": 6800
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.122403710425489e-05,
+      "loss": 0.0915,
+      "step": 6810
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.1173623714458562e-05,
+      "loss": 0.0825,
+      "step": 6820
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.1123210324662232e-05,
+      "loss": 0.0819,
+      "step": 6830
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.10727969348659e-05,
+      "loss": 0.0661,
+      "step": 6840
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.102238354506957e-05,
+      "loss": 0.0901,
+      "step": 6850
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.097197015527324e-05,
+      "loss": 0.085,
+      "step": 6860
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.092155676547691e-05,
+      "loss": 0.0857,
+      "step": 6870
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.087114337568058e-05,
+      "loss": 0.0854,
+      "step": 6880
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.082072998588425e-05,
+      "loss": 0.0809,
+      "step": 6890
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.077031659608792e-05,
+      "loss": 0.1004,
+      "step": 6900
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.071990320629159e-05,
+      "loss": 0.0665,
+      "step": 6910
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.0669489816495264e-05,
+      "loss": 0.0812,
+      "step": 6920
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.0619076426698934e-05,
+      "loss": 0.0821,
+      "step": 6930
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.0568663036902604e-05,
+      "loss": 0.0809,
+      "step": 6940
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.0518249647106274e-05,
+      "loss": 0.095,
+      "step": 6950
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.046783625730994e-05,
+      "loss": 0.0845,
+      "step": 6960
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.0417422867513613e-05,
+      "loss": 0.0919,
+      "step": 6970
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.0367009477717283e-05,
+      "loss": 0.1,
+      "step": 6980
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.0316596087920953e-05,
+      "loss": 0.0617,
+      "step": 6990
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.0266182698124623e-05,
+      "loss": 0.079,
+      "step": 7000
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.0215769308328293e-05,
+      "loss": 0.0923,
+      "step": 7010
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.0165355918531963e-05,
+      "loss": 0.0717,
+      "step": 7020
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.0114942528735632e-05,
+      "loss": 0.0806,
+      "step": 7030
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.0064529138939302e-05,
+      "loss": 0.0735,
+      "step": 7040
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.0014115749142972e-05,
+      "loss": 0.0858,
+      "step": 7050
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.9963702359346642e-05,
+      "loss": 0.0739,
+      "step": 7060
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 1.9913288969550315e-05,
+      "loss": 0.0726,
+      "step": 7070
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 1.9862875579753985e-05,
+      "loss": 0.0999,
+      "step": 7080
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 1.9812462189957655e-05,
+      "loss": 0.1066,
+      "step": 7090
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.976204880016132e-05,
+      "loss": 0.085,
+      "step": 7100
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.971163541036499e-05,
+      "loss": 0.0974,
+      "step": 7110
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 1.9661222020568665e-05,
+      "loss": 0.0884,
+      "step": 7120
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 1.9610808630772334e-05,
+      "loss": 0.078,
+      "step": 7130
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 1.9560395240976004e-05,
+      "loss": 0.0936,
+      "step": 7140
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 1.9509981851179674e-05,
+      "loss": 0.0863,
+      "step": 7150
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.9459568461383344e-05,
+      "loss": 0.065,
+      "step": 7160
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.9863945578231292,
+      "eval_loss": 0.039155878126621246,
+      "eval_runtime": 88.5016,
+      "eval_samples_per_second": 132.879,
+      "eval_steps_per_second": 4.158,
+      "step": 7163
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 1.9409155071587014e-05,
+      "loss": 0.0942,
+      "step": 7170
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 1.9358741681790684e-05,
+      "loss": 0.0707,
+      "step": 7180
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 1.9308328291994354e-05,
+      "loss": 0.0883,
+      "step": 7190
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 1.9257914902198024e-05,
+      "loss": 0.0721,
+      "step": 7200
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 1.9207501512401697e-05,
+      "loss": 0.0755,
+      "step": 7210
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 1.9157088122605367e-05,
+      "loss": 0.0709,
+      "step": 7220
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 1.9106674732809037e-05,
+      "loss": 0.0659,
+      "step": 7230
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 1.9056261343012703e-05,
+      "loss": 0.086,
+      "step": 7240
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 1.9005847953216373e-05,
+      "loss": 0.0885,
+      "step": 7250
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 1.8955434563420046e-05,
+      "loss": 0.078,
+      "step": 7260
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 1.8905021173623716e-05,
+      "loss": 0.0848,
+      "step": 7270
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 1.8854607783827386e-05,
+      "loss": 0.0831,
+      "step": 7280
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 1.8804194394031056e-05,
+      "loss": 0.0802,
+      "step": 7290
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 1.8753781004234726e-05,
+      "loss": 0.0618,
+      "step": 7300
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 1.8703367614438395e-05,
+      "loss": 0.0666,
+      "step": 7310
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 1.8652954224642065e-05,
+      "loss": 0.0737,
+      "step": 7320
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 1.8602540834845735e-05,
+      "loss": 0.0697,
+      "step": 7330
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 1.8552127445049405e-05,
+      "loss": 0.0812,
+      "step": 7340
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 1.8501714055253075e-05,
+      "loss": 0.0884,
+      "step": 7350
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 1.8451300665456748e-05,
+      "loss": 0.0648,
+      "step": 7360
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.8400887275660418e-05,
+      "loss": 0.0641,
+      "step": 7370
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 1.8350473885864088e-05,
+      "loss": 0.0977,
+      "step": 7380
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 1.8300060496067754e-05,
+      "loss": 0.0778,
+      "step": 7390
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 1.8249647106271424e-05,
+      "loss": 0.0746,
+      "step": 7400
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 1.8199233716475097e-05,
+      "loss": 0.0943,
+      "step": 7410
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.8148820326678767e-05,
+      "loss": 0.0971,
+      "step": 7420
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 1.8098406936882437e-05,
+      "loss": 0.0848,
+      "step": 7430
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 1.8047993547086107e-05,
+      "loss": 0.0712,
+      "step": 7440
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 1.7997580157289777e-05,
+      "loss": 0.0672,
+      "step": 7450
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.7947166767493447e-05,
+      "loss": 0.0748,
+      "step": 7460
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.7896753377697117e-05,
+      "loss": 0.0965,
+      "step": 7470
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 1.7846339987900787e-05,
+      "loss": 0.0606,
+      "step": 7480
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.7795926598104456e-05,
+      "loss": 0.0924,
+      "step": 7490
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 1.774551320830813e-05,
+      "loss": 0.082,
+      "step": 7500
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.76950998185118e-05,
+      "loss": 0.0886,
+      "step": 7510
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.764468642871547e-05,
+      "loss": 0.0798,
+      "step": 7520
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.7594273038919136e-05,
+      "loss": 0.0855,
+      "step": 7530
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.7543859649122806e-05,
+      "loss": 0.069,
+      "step": 7540
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 1.749344625932648e-05,
+      "loss": 0.0726,
+      "step": 7550
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.744303286953015e-05,
+      "loss": 0.0892,
+      "step": 7560
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.739261947973382e-05,
+      "loss": 0.0788,
+      "step": 7570
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.734220608993749e-05,
+      "loss": 0.0941,
+      "step": 7580
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.729179270014116e-05,
+      "loss": 0.066,
+      "step": 7590
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.7241379310344828e-05,
+      "loss": 0.0976,
+      "step": 7600
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.7190965920548498e-05,
+      "loss": 0.089,
+      "step": 7610
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.7140552530752168e-05,
+      "loss": 0.0751,
+      "step": 7620
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.7090139140955838e-05,
+      "loss": 0.0887,
+      "step": 7630
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.7039725751159508e-05,
+      "loss": 0.0667,
+      "step": 7640
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.698931236136318e-05,
+      "loss": 0.0626,
+      "step": 7650
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.693889897156685e-05,
+      "loss": 0.067,
+      "step": 7660
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.6888485581770517e-05,
+      "loss": 0.0727,
+      "step": 7670
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.6838072191974187e-05,
+      "loss": 0.0698,
+      "step": 7680
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.6787658802177857e-05,
+      "loss": 0.0746,
+      "step": 7690
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.673724541238153e-05,
+      "loss": 0.078,
+      "step": 7700
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.66868320225852e-05,
+      "loss": 0.0775,
+      "step": 7710
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.989030612244898,
+      "eval_loss": 0.03436902537941933,
+      "eval_runtime": 86.6273,
+      "eval_samples_per_second": 135.754,
+      "eval_steps_per_second": 4.248,
+      "step": 7714
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.663641863278887e-05,
+      "loss": 0.0703,
+      "step": 7720
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.658600524299254e-05,
+      "loss": 0.0832,
+      "step": 7730
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.653559185319621e-05,
+      "loss": 0.0781,
+      "step": 7740
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 1.648517846339988e-05,
+      "loss": 0.0659,
+      "step": 7750
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.643476507360355e-05,
+      "loss": 0.078,
+      "step": 7760
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.638435168380722e-05,
+      "loss": 0.0825,
+      "step": 7770
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.633393829401089e-05,
+      "loss": 0.0685,
+      "step": 7780
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.628352490421456e-05,
+      "loss": 0.0666,
+      "step": 7790
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 1.6233111514418232e-05,
+      "loss": 0.0704,
+      "step": 7800
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.6182698124621902e-05,
+      "loss": 0.0571,
+      "step": 7810
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.613228473482557e-05,
+      "loss": 0.0601,
+      "step": 7820
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.608187134502924e-05,
+      "loss": 0.0725,
+      "step": 7830
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.6031457955232912e-05,
+      "loss": 0.0708,
+      "step": 7840
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 1.598104456543658e-05,
+      "loss": 0.0542,
+      "step": 7850
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.593063117564025e-05,
+      "loss": 0.0956,
+      "step": 7860
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.588021778584392e-05,
+      "loss": 0.064,
+      "step": 7870
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.582980439604759e-05,
+      "loss": 0.0761,
+      "step": 7880
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.577939100625126e-05,
+      "loss": 0.0725,
+      "step": 7890
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 1.572897761645493e-05,
+      "loss": 0.0752,
+      "step": 7900
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 1.56785642266586e-05,
+      "loss": 0.0584,
+      "step": 7910
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.562815083686227e-05,
+      "loss": 0.0824,
+      "step": 7920
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.557773744706594e-05,
+      "loss": 0.0826,
+      "step": 7930
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.5527324057269614e-05,
+      "loss": 0.0608,
+      "step": 7940
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 1.5476910667473284e-05,
+      "loss": 0.0756,
+      "step": 7950
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 1.542649727767695e-05,
+      "loss": 0.0818,
+      "step": 7960
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.537608388788062e-05,
+      "loss": 0.1,
+      "step": 7970
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.532567049808429e-05,
+      "loss": 0.0737,
+      "step": 7980
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.5275257108287963e-05,
+      "loss": 0.0766,
+      "step": 7990
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 1.5224843718491633e-05,
+      "loss": 0.0822,
+      "step": 8000
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 1.5174430328695303e-05,
+      "loss": 0.0924,
+      "step": 8010
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.5124016938898971e-05,
+      "loss": 0.0675,
+      "step": 8020
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.5073603549102641e-05,
+      "loss": 0.065,
+      "step": 8030
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.5023190159306314e-05,
+      "loss": 0.0894,
+      "step": 8040
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 1.4972776769509982e-05,
+      "loss": 0.0931,
+      "step": 8050
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 1.4922363379713652e-05,
+      "loss": 0.0672,
+      "step": 8060
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 1.4871949989917322e-05,
+      "loss": 0.065,
+      "step": 8070
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.4821536600120992e-05,
+      "loss": 0.0756,
+      "step": 8080
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.4771123210324664e-05,
+      "loss": 0.085,
+      "step": 8090
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 1.4720709820528333e-05,
+      "loss": 0.0682,
+      "step": 8100
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 1.4670296430732003e-05,
+      "loss": 0.0597,
+      "step": 8110
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 1.4619883040935673e-05,
+      "loss": 0.0624,
+      "step": 8120
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.4569469651139341e-05,
+      "loss": 0.0851,
+      "step": 8130
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.4519056261343015e-05,
+      "loss": 0.0716,
+      "step": 8140
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 1.4468642871546684e-05,
+      "loss": 0.0796,
+      "step": 8150
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 1.4418229481750353e-05,
+      "loss": 0.063,
+      "step": 8160
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 1.4367816091954022e-05,
+      "loss": 0.0815,
+      "step": 8170
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.4317402702157696e-05,
+      "loss": 0.0669,
+      "step": 8180
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.4266989312361364e-05,
+      "loss": 0.066,
+      "step": 8190
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 1.4216575922565034e-05,
+      "loss": 0.0735,
+      "step": 8200
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 1.4166162532768704e-05,
+      "loss": 0.0652,
+      "step": 8210
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 1.4115749142972373e-05,
+      "loss": 0.069,
+      "step": 8220
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 1.4065335753176045e-05,
+      "loss": 0.0755,
+      "step": 8230
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.4014922363379715e-05,
+      "loss": 0.0636,
+      "step": 8240
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 1.3964508973583385e-05,
+      "loss": 0.0756,
+      "step": 8250
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 1.3914095583787055e-05,
+      "loss": 0.0544,
+      "step": 8260
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.9887755102040816,
+      "eval_loss": 0.03623654320836067,
+      "eval_runtime": 88.3574,
+      "eval_samples_per_second": 133.096,
+      "eval_steps_per_second": 4.165,
+      "step": 8265
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 1.3863682193990723e-05,
+      "loss": 0.0831,
+      "step": 8270
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 1.3813268804194396e-05,
+      "loss": 0.07,
+      "step": 8280
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 1.3762855414398066e-05,
+      "loss": 0.0852,
+      "step": 8290
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 1.3712442024601734e-05,
+      "loss": 0.0575,
+      "step": 8300
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 1.3662028634805404e-05,
+      "loss": 0.0737,
+      "step": 8310
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 1.3611615245009074e-05,
+      "loss": 0.0693,
+      "step": 8320
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 1.3561201855212747e-05,
+      "loss": 0.0708,
+      "step": 8330
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 1.3510788465416415e-05,
+      "loss": 0.0759,
+      "step": 8340
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 1.3460375075620085e-05,
+      "loss": 0.0678,
+      "step": 8350
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 1.3409961685823755e-05,
+      "loss": 0.0606,
+      "step": 8360
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 1.3359548296027425e-05,
+      "loss": 0.065,
+      "step": 8370
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 1.3309134906231096e-05,
+      "loss": 0.0713,
+      "step": 8380
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 1.3258721516434766e-05,
+      "loss": 0.0872,
+      "step": 8390
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 1.3208308126638436e-05,
+      "loss": 0.0777,
+      "step": 8400
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 0.0878,
+      "step": 8410
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 1.3107481347045774e-05,
+      "loss": 0.0805,
+      "step": 8420
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 1.3057067957249447e-05,
+      "loss": 0.0696,
+      "step": 8430
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 1.3006654567453117e-05,
+      "loss": 0.0699,
+      "step": 8440
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.2956241177656785e-05,
+      "loss": 0.0793,
+      "step": 8450
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 1.2905827787860455e-05,
+      "loss": 0.0752,
+      "step": 8460
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 1.2855414398064125e-05,
+      "loss": 0.091,
+      "step": 8470
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 1.2805001008267797e-05,
+      "loss": 0.0651,
+      "step": 8480
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 1.2754587618471467e-05,
+      "loss": 0.0617,
+      "step": 8490
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.2704174228675136e-05,
+      "loss": 0.0574,
+      "step": 8500
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 1.2653760838878806e-05,
+      "loss": 0.0677,
+      "step": 8510
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 1.2603347449082478e-05,
+      "loss": 0.0746,
+      "step": 8520
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 1.2552934059286148e-05,
+      "loss": 0.0875,
+      "step": 8530
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 1.2502520669489818e-05,
+      "loss": 0.0603,
+      "step": 8540
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.2452107279693487e-05,
+      "loss": 0.0726,
+      "step": 8550
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 1.2401693889897157e-05,
+      "loss": 0.0632,
+      "step": 8560
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 1.2351280500100827e-05,
+      "loss": 0.0648,
+      "step": 8570
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 1.2300867110304499e-05,
+      "loss": 0.0891,
+      "step": 8580
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 1.2250453720508167e-05,
+      "loss": 0.0689,
+      "step": 8590
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.2200040330711837e-05,
+      "loss": 0.0763,
+      "step": 8600
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.2149626940915508e-05,
+      "loss": 0.0655,
+      "step": 8610
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 1.2099213551119178e-05,
+      "loss": 0.0752,
+      "step": 8620
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 1.2048800161322848e-05,
+      "loss": 0.07,
+      "step": 8630
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 1.1998386771526518e-05,
+      "loss": 0.0694,
+      "step": 8640
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.1947973381730188e-05,
+      "loss": 0.0606,
+      "step": 8650
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.1897559991933858e-05,
+      "loss": 0.0614,
+      "step": 8660
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 1.1847146602137528e-05,
+      "loss": 0.0669,
+      "step": 8670
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 1.1796733212341199e-05,
+      "loss": 0.066,
+      "step": 8680
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 1.1746319822544869e-05,
+      "loss": 0.0651,
+      "step": 8690
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.1695906432748537e-05,
+      "loss": 0.0757,
+      "step": 8700
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.1645493042952209e-05,
+      "loss": 0.0607,
+      "step": 8710
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.1595079653155879e-05,
+      "loss": 0.0573,
+      "step": 8720
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 1.1544666263359548e-05,
+      "loss": 0.0734,
+      "step": 8730
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 1.1494252873563218e-05,
+      "loss": 0.0565,
+      "step": 8740
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.144383948376689e-05,
+      "loss": 0.0644,
+      "step": 8750
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.139342609397056e-05,
+      "loss": 0.0699,
+      "step": 8760
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.1343012704174228e-05,
+      "loss": 0.0658,
+      "step": 8770
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 1.12925993143779e-05,
+      "loss": 0.0631,
+      "step": 8780
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 1.124218592458157e-05,
+      "loss": 0.0764,
+      "step": 8790
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.1191772534785241e-05,
+      "loss": 0.0786,
+      "step": 8800
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.1141359144988909e-05,
+      "loss": 0.0584,
+      "step": 8810
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.9871598639455782,
+      "eval_loss": 0.042227257043123245,
+      "eval_runtime": 87.9973,
+      "eval_samples_per_second": 133.64,
+      "eval_steps_per_second": 4.182,
+      "step": 8816
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.1090945755192579e-05,
+      "loss": 0.0871,
+      "step": 8820
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 1.104053236539625e-05,
+      "loss": 0.0745,
+      "step": 8830
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 1.099011897559992e-05,
+      "loss": 0.0585,
+      "step": 8840
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 1.093970558580359e-05,
+      "loss": 0.0705,
+      "step": 8850
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 1.088929219600726e-05,
+      "loss": 0.074,
+      "step": 8860
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 1.083887880621093e-05,
+      "loss": 0.0669,
+      "step": 8870
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 1.07884654164146e-05,
+      "loss": 0.0717,
+      "step": 8880
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 1.073805202661827e-05,
+      "loss": 0.0899,
+      "step": 8890
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 1.0687638636821941e-05,
+      "loss": 0.0631,
+      "step": 8900
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 1.0637225247025611e-05,
+      "loss": 0.0753,
+      "step": 8910
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 1.0586811857229281e-05,
+      "loss": 0.0657,
+      "step": 8920
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 1.053639846743295e-05,
+      "loss": 0.0646,
+      "step": 8930
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 1.048598507763662e-05,
+      "loss": 0.065,
+      "step": 8940
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.043557168784029e-05,
+      "loss": 0.0574,
+      "step": 8950
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.038515829804396e-05,
+      "loss": 0.0601,
+      "step": 8960
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.0334744908247632e-05,
+      "loss": 0.0754,
+      "step": 8970
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.0284331518451302e-05,
+      "loss": 0.0879,
+      "step": 8980
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 1.023391812865497e-05,
+      "loss": 0.0614,
+      "step": 8990
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.0183504738858642e-05,
+      "loss": 0.0762,
+      "step": 9000
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.0133091349062311e-05,
+      "loss": 0.0584,
+      "step": 9010
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.0082677959265981e-05,
+      "loss": 0.072,
+      "step": 9020
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.0032264569469651e-05,
+      "loss": 0.0591,
+      "step": 9030
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 9.981851179673321e-06,
+      "loss": 0.0673,
+      "step": 9040
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 9.931437789876993e-06,
+      "loss": 0.0699,
+      "step": 9050
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 9.88102440008066e-06,
+      "loss": 0.0507,
+      "step": 9060
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 9.830611010284332e-06,
+      "loss": 0.0776,
+      "step": 9070
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 9.780197620488002e-06,
+      "loss": 0.061,
+      "step": 9080
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 9.729784230691672e-06,
+      "loss": 0.0696,
+      "step": 9090
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 9.679370840895342e-06,
+      "loss": 0.0689,
+      "step": 9100
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 9.628957451099012e-06,
+      "loss": 0.0511,
+      "step": 9110
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 9.578544061302683e-06,
+      "loss": 0.0776,
+      "step": 9120
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 9.528130671506351e-06,
+      "loss": 0.0617,
+      "step": 9130
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 9.477717281710023e-06,
+      "loss": 0.0779,
+      "step": 9140
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 9.427303891913693e-06,
+      "loss": 0.0665,
+      "step": 9150
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 9.376890502117363e-06,
+      "loss": 0.0697,
+      "step": 9160
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 9.326477112321033e-06,
+      "loss": 0.0654,
+      "step": 9170
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 9.276063722524703e-06,
+      "loss": 0.0703,
+      "step": 9180
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 9.225650332728374e-06,
+      "loss": 0.0712,
+      "step": 9190
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 9.175236942932044e-06,
+      "loss": 0.0612,
+      "step": 9200
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 9.124823553135712e-06,
+      "loss": 0.0695,
+      "step": 9210
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 9.074410163339384e-06,
+      "loss": 0.0693,
+      "step": 9220
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 9.023996773543054e-06,
+      "loss": 0.0746,
+      "step": 9230
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 8.973583383746723e-06,
+      "loss": 0.0535,
+      "step": 9240
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 8.923169993950393e-06,
+      "loss": 0.0784,
+      "step": 9250
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 8.872756604154065e-06,
+      "loss": 0.0849,
+      "step": 9260
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 8.822343214357735e-06,
+      "loss": 0.0601,
+      "step": 9270
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 8.771929824561403e-06,
+      "loss": 0.0531,
+      "step": 9280
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 8.721516434765074e-06,
+      "loss": 0.0619,
+      "step": 9290
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 8.671103044968744e-06,
+      "loss": 0.0697,
+      "step": 9300
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 8.620689655172414e-06,
+      "loss": 0.0756,
+      "step": 9310
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 8.570276265376084e-06,
+      "loss": 0.0671,
+      "step": 9320
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 8.519862875579754e-06,
+      "loss": 0.0671,
+      "step": 9330
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 8.469449485783425e-06,
+      "loss": 0.0734,
+      "step": 9340
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 8.419036095987094e-06,
+      "loss": 0.0757,
+      "step": 9350
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 8.368622706190765e-06,
+      "loss": 0.0722,
+      "step": 9360
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.9899659863945578,
+      "eval_loss": 0.031354524195194244,
+      "eval_runtime": 87.5888,
+      "eval_samples_per_second": 134.264,
+      "eval_steps_per_second": 4.201,
+      "step": 9367
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 8.318209316394435e-06,
+      "loss": 0.0767,
+      "step": 9370
+    },
+    {
+      "epoch": 17.02,
+      "learning_rate": 8.267795926598105e-06,
+      "loss": 0.0749,
+      "step": 9380
+    },
+    {
+      "epoch": 17.04,
+      "learning_rate": 8.217382536801775e-06,
+      "loss": 0.0726,
+      "step": 9390
+    },
+    {
+      "epoch": 17.06,
+      "learning_rate": 8.166969147005445e-06,
+      "loss": 0.0734,
+      "step": 9400
+    },
+    {
+      "epoch": 17.08,
+      "learning_rate": 8.116555757209116e-06,
+      "loss": 0.0644,
+      "step": 9410
+    },
+    {
+      "epoch": 17.1,
+      "learning_rate": 8.066142367412784e-06,
+      "loss": 0.0716,
+      "step": 9420
+    },
+    {
+      "epoch": 17.11,
+      "learning_rate": 8.015728977616456e-06,
+      "loss": 0.0562,
+      "step": 9430
+    },
+    {
+      "epoch": 17.13,
+      "learning_rate": 7.965315587820126e-06,
+      "loss": 0.0667,
+      "step": 9440
+    },
+    {
+      "epoch": 17.15,
+      "learning_rate": 7.914902198023796e-06,
+      "loss": 0.0726,
+      "step": 9450
+    },
+    {
+      "epoch": 17.17,
+      "learning_rate": 7.864488808227466e-06,
+      "loss": 0.0711,
+      "step": 9460
+    },
+    {
+      "epoch": 17.19,
+      "learning_rate": 7.814075418431135e-06,
+      "loss": 0.0874,
+      "step": 9470
+    },
+    {
+      "epoch": 17.2,
+      "learning_rate": 7.763662028634807e-06,
+      "loss": 0.0639,
+      "step": 9480
+    },
+    {
+      "epoch": 17.22,
+      "learning_rate": 7.713248638838475e-06,
+      "loss": 0.0673,
+      "step": 9490
+    },
+    {
+      "epoch": 17.24,
+      "learning_rate": 7.662835249042145e-06,
+      "loss": 0.0554,
+      "step": 9500
+    },
+    {
+      "epoch": 17.26,
+      "learning_rate": 7.6124218592458165e-06,
+      "loss": 0.0734,
+      "step": 9510
+    },
+    {
+      "epoch": 17.28,
+      "learning_rate": 7.5620084694494855e-06,
+      "loss": 0.071,
+      "step": 9520
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 7.511595079653157e-06,
+      "loss": 0.0666,
+      "step": 9530
+    },
+    {
+      "epoch": 17.31,
+      "learning_rate": 7.461181689856826e-06,
+      "loss": 0.0531,
+      "step": 9540
+    },
+    {
+      "epoch": 17.33,
+      "learning_rate": 7.410768300060496e-06,
+      "loss": 0.0625,
+      "step": 9550
+    },
+    {
+      "epoch": 17.35,
+      "learning_rate": 7.360354910264167e-06,
+      "loss": 0.0663,
+      "step": 9560
+    },
+    {
+      "epoch": 17.37,
+      "learning_rate": 7.3099415204678366e-06,
+      "loss": 0.0598,
+      "step": 9570
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 7.259528130671507e-06,
+      "loss": 0.0705,
+      "step": 9580
+    },
+    {
+      "epoch": 17.4,
+      "learning_rate": 7.209114740875176e-06,
+      "loss": 0.046,
+      "step": 9590
+    },
+    {
+      "epoch": 17.42,
+      "learning_rate": 7.158701351078848e-06,
+      "loss": 0.0566,
+      "step": 9600
+    },
+    {
+      "epoch": 17.44,
+      "learning_rate": 7.108287961282517e-06,
+      "loss": 0.0573,
+      "step": 9610
+    },
+    {
+      "epoch": 17.46,
+      "learning_rate": 7.057874571486187e-06,
+      "loss": 0.0675,
+      "step": 9620
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 7.0074611816898574e-06,
+      "loss": 0.0772,
+      "step": 9630
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 6.957047791893527e-06,
+      "loss": 0.0582,
+      "step": 9640
+    },
+    {
+      "epoch": 17.51,
+      "learning_rate": 6.906634402097198e-06,
+      "loss": 0.0791,
+      "step": 9650
+    },
+    {
+      "epoch": 17.53,
+      "learning_rate": 6.856221012300867e-06,
+      "loss": 0.0666,
+      "step": 9660
+    },
+    {
+      "epoch": 17.55,
+      "learning_rate": 6.805807622504537e-06,
+      "loss": 0.0619,
+      "step": 9670
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 6.755394232708208e-06,
+      "loss": 0.0541,
+      "step": 9680
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 6.7049808429118775e-06,
+      "loss": 0.0731,
+      "step": 9690
+    },
+    {
+      "epoch": 17.6,
+      "learning_rate": 6.654567453115548e-06,
+      "loss": 0.0631,
+      "step": 9700
+    },
+    {
+      "epoch": 17.62,
+      "learning_rate": 6.604154063319218e-06,
+      "loss": 0.0604,
+      "step": 9710
+    },
+    {
+      "epoch": 17.64,
+      "learning_rate": 6.553740673522887e-06,
+      "loss": 0.0678,
+      "step": 9720
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 6.503327283726559e-06,
+      "loss": 0.0666,
+      "step": 9730
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 6.452913893930228e-06,
+      "loss": 0.0589,
+      "step": 9740
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 6.402500504133898e-06,
+      "loss": 0.071,
+      "step": 9750
+    },
+    {
+      "epoch": 17.71,
+      "learning_rate": 6.352087114337568e-06,
+      "loss": 0.052,
+      "step": 9760
+    },
+    {
+      "epoch": 17.73,
+      "learning_rate": 6.301673724541239e-06,
+      "loss": 0.0705,
+      "step": 9770
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 6.251260334744909e-06,
+      "loss": 0.0705,
+      "step": 9780
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 6.200846944948579e-06,
+      "loss": 0.0523,
+      "step": 9790
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 6.150433555152249e-06,
+      "loss": 0.0731,
+      "step": 9800
+    },
+    {
+      "epoch": 17.8,
+      "learning_rate": 6.100020165355918e-06,
+      "loss": 0.0537,
+      "step": 9810
+    },
+    {
+      "epoch": 17.82,
+      "learning_rate": 6.049606775559589e-06,
+      "loss": 0.0637,
+      "step": 9820
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 5.999193385763259e-06,
+      "loss": 0.0541,
+      "step": 9830
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 5.948779995966929e-06,
+      "loss": 0.0513,
+      "step": 9840
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 5.8983666061705996e-06,
+      "loss": 0.0574,
+      "step": 9850
+    },
+    {
+      "epoch": 17.89,
+      "learning_rate": 5.8479532163742686e-06,
+      "loss": 0.0467,
+      "step": 9860
+    },
+    {
+      "epoch": 17.91,
+      "learning_rate": 5.797539826577939e-06,
+      "loss": 0.067,
+      "step": 9870
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 5.747126436781609e-06,
+      "loss": 0.0708,
+      "step": 9880
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 5.69671304698528e-06,
+      "loss": 0.0701,
+      "step": 9890
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 5.64629965718895e-06,
+      "loss": 0.0695,
+      "step": 9900
+    },
+    {
+      "epoch": 17.98,
+      "learning_rate": 5.5958862673926204e-06,
+      "loss": 0.0765,
+      "step": 9910
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.9908163265306122,
+      "eval_loss": 0.031277142465114594,
+      "eval_runtime": 89.2183,
+      "eval_samples_per_second": 131.811,
+      "eval_steps_per_second": 4.125,
+      "step": 9918
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 5.5454728775962895e-06,
+      "loss": 0.0623,
+      "step": 9920
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 5.49505948779996e-06,
+      "loss": 0.0653,
+      "step": 9930
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 5.44464609800363e-06,
+      "loss": 0.0611,
+      "step": 9940
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 5.3942327082073e-06,
+      "loss": 0.0752,
+      "step": 9950
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 5.343819318410971e-06,
+      "loss": 0.0572,
+      "step": 9960
+    },
+    {
+      "epoch": 18.09,
+      "learning_rate": 5.2934059286146405e-06,
+      "loss": 0.0538,
+      "step": 9970
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 5.24299253881831e-06,
+      "loss": 0.079,
+      "step": 9980
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 5.19257914902198e-06,
+      "loss": 0.0707,
+      "step": 9990
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 5.142165759225651e-06,
+      "loss": 0.0574,
+      "step": 10000
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 5.091752369429321e-06,
+      "loss": 0.0665,
+      "step": 10010
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 5.041338979632991e-06,
+      "loss": 0.0417,
+      "step": 10020
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 4.9909255898366605e-06,
+      "loss": 0.0556,
+      "step": 10030
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 4.94051220004033e-06,
+      "loss": 0.0651,
+      "step": 10040
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 4.890098810244001e-06,
+      "loss": 0.0578,
+      "step": 10050
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 4.839685420447671e-06,
+      "loss": 0.0605,
+      "step": 10060
+    },
+    {
+      "epoch": 18.28,
+      "learning_rate": 4.789272030651342e-06,
+      "loss": 0.0679,
+      "step": 10070
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 4.7388586408550115e-06,
+      "loss": 0.0718,
+      "step": 10080
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 4.688445251058681e-06,
+      "loss": 0.0643,
+      "step": 10090
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 4.638031861262351e-06,
+      "loss": 0.0686,
+      "step": 10100
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 4.587618471466022e-06,
+      "loss": 0.0607,
+      "step": 10110
+    },
+    {
+      "epoch": 18.37,
+      "learning_rate": 4.537205081669692e-06,
+      "loss": 0.0625,
+      "step": 10120
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 4.486791691873362e-06,
+      "loss": 0.0573,
+      "step": 10130
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 4.436378302077032e-06,
+      "loss": 0.0862,
+      "step": 10140
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 4.3859649122807014e-06,
+      "loss": 0.0572,
+      "step": 10150
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 4.335551522484372e-06,
+      "loss": 0.0617,
+      "step": 10160
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 4.285138132688042e-06,
+      "loss": 0.0685,
+      "step": 10170
+    },
+    {
+      "epoch": 18.48,
+      "learning_rate": 4.234724742891713e-06,
+      "loss": 0.0581,
+      "step": 10180
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 4.184311353095383e-06,
+      "loss": 0.0562,
+      "step": 10190
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 4.1338979632990524e-06,
+      "loss": 0.0641,
+      "step": 10200
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 4.083484573502722e-06,
+      "loss": 0.0586,
+      "step": 10210
+    },
+    {
+      "epoch": 18.55,
+      "learning_rate": 4.033071183706392e-06,
+      "loss": 0.0764,
+      "step": 10220
+    },
+    {
+      "epoch": 18.57,
+      "learning_rate": 3.982657793910063e-06,
+      "loss": 0.0588,
+      "step": 10230
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 3.932244404113733e-06,
+      "loss": 0.0639,
+      "step": 10240
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 3.8818310143174035e-06,
+      "loss": 0.0644,
+      "step": 10250
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 3.8314176245210725e-06,
+      "loss": 0.0771,
+      "step": 10260
+    },
+    {
+      "epoch": 18.64,
+      "learning_rate": 3.7810042347247428e-06,
+      "loss": 0.0577,
+      "step": 10270
+    },
+    {
+      "epoch": 18.66,
+      "learning_rate": 3.730590844928413e-06,
+      "loss": 0.0539,
+      "step": 10280
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 3.6801774551320833e-06,
+      "loss": 0.0658,
+      "step": 10290
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 3.6297640653357536e-06,
+      "loss": 0.0574,
+      "step": 10300
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 3.579350675539424e-06,
+      "loss": 0.0584,
+      "step": 10310
+    },
+    {
+      "epoch": 18.73,
+      "learning_rate": 3.5289372857430934e-06,
+      "loss": 0.0453,
+      "step": 10320
+    },
+    {
+      "epoch": 18.75,
+      "learning_rate": 3.4785238959467637e-06,
+      "loss": 0.062,
+      "step": 10330
+    },
+    {
+      "epoch": 18.77,
+      "learning_rate": 3.4281105061504335e-06,
+      "loss": 0.0525,
+      "step": 10340
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 3.377697116354104e-06,
+      "loss": 0.0565,
+      "step": 10350
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 3.327283726557774e-06,
+      "loss": 0.058,
+      "step": 10360
+    },
+    {
+      "epoch": 18.82,
+      "learning_rate": 3.2768703367614435e-06,
+      "loss": 0.0611,
+      "step": 10370
+    },
+    {
+      "epoch": 18.84,
+      "learning_rate": 3.226456946965114e-06,
+      "loss": 0.0498,
+      "step": 10380
+    },
+    {
+      "epoch": 18.86,
+      "learning_rate": 3.176043557168784e-06,
+      "loss": 0.0527,
+      "step": 10390
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 3.1256301673724544e-06,
+      "loss": 0.0671,
+      "step": 10400
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 3.0752167775761247e-06,
+      "loss": 0.0642,
+      "step": 10410
+    },
+    {
+      "epoch": 18.91,
+      "learning_rate": 3.0248033877797946e-06,
+      "loss": 0.0431,
+      "step": 10420
+    },
+    {
+      "epoch": 18.93,
+      "learning_rate": 2.9743899979834644e-06,
+      "loss": 0.0485,
+      "step": 10430
+    },
+    {
+      "epoch": 18.95,
+      "learning_rate": 2.9239766081871343e-06,
+      "loss": 0.0501,
+      "step": 10440
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 2.8735632183908046e-06,
+      "loss": 0.074,
+      "step": 10450
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 2.823149828594475e-06,
+      "loss": 0.0696,
+      "step": 10460
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.991156462585034,
+      "eval_loss": 0.02969362773001194,
+      "eval_runtime": 87.7437,
+      "eval_samples_per_second": 134.027,
+      "eval_steps_per_second": 4.194,
+      "step": 10469
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 2.7727364387981447e-06,
+      "loss": 0.061,
+      "step": 10470
+    },
+    {
+      "epoch": 19.02,
+      "learning_rate": 2.722323049001815e-06,
+      "loss": 0.057,
+      "step": 10480
+    },
+    {
+      "epoch": 19.04,
+      "learning_rate": 2.6719096592054853e-06,
+      "loss": 0.0473,
+      "step": 10490
+    },
+    {
+      "epoch": 19.06,
+      "learning_rate": 2.621496269409155e-06,
+      "loss": 0.0577,
+      "step": 10500
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 2.5710828796128255e-06,
+      "loss": 0.0691,
+      "step": 10510
+    },
+    {
+      "epoch": 19.09,
+      "learning_rate": 2.5206694898164953e-06,
+      "loss": 0.0582,
+      "step": 10520
+    },
+    {
+      "epoch": 19.11,
+      "learning_rate": 2.470256100020165e-06,
+      "loss": 0.0545,
+      "step": 10530
+    },
+    {
+      "epoch": 19.13,
+      "learning_rate": 2.4198427102238355e-06,
+      "loss": 0.0701,
+      "step": 10540
+    },
+    {
+      "epoch": 19.15,
+      "learning_rate": 2.3694293204275058e-06,
+      "loss": 0.0548,
+      "step": 10550
+    },
+    {
+      "epoch": 19.17,
+      "learning_rate": 2.3190159306311756e-06,
+      "loss": 0.0662,
+      "step": 10560
+    },
+    {
+      "epoch": 19.18,
+      "learning_rate": 2.268602540834846e-06,
+      "loss": 0.0593,
+      "step": 10570
+    },
+    {
+      "epoch": 19.2,
+      "learning_rate": 2.218189151038516e-06,
+      "loss": 0.0596,
+      "step": 10580
+    },
+    {
+      "epoch": 19.22,
+      "learning_rate": 2.167775761242186e-06,
+      "loss": 0.0527,
+      "step": 10590
+    },
+    {
+      "epoch": 19.24,
+      "learning_rate": 2.1173623714458564e-06,
+      "loss": 0.0566,
+      "step": 10600
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 2.0669489816495262e-06,
+      "loss": 0.0572,
+      "step": 10610
+    },
+    {
+      "epoch": 19.27,
+      "learning_rate": 2.016535591853196e-06,
+      "loss": 0.0445,
+      "step": 10620
+    },
+    {
+      "epoch": 19.29,
+      "learning_rate": 1.9661222020568664e-06,
+      "loss": 0.0726,
+      "step": 10630
+    },
+    {
+      "epoch": 19.31,
+      "learning_rate": 1.9157088122605362e-06,
+      "loss": 0.0668,
+      "step": 10640
+    },
+    {
+      "epoch": 19.33,
+      "learning_rate": 1.8652954224642065e-06,
+      "loss": 0.0705,
+      "step": 10650
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.8148820326678768e-06,
+      "loss": 0.0678,
+      "step": 10660
+    },
+    {
+      "epoch": 19.36,
+      "learning_rate": 1.7644686428715467e-06,
+      "loss": 0.0708,
+      "step": 10670
+    },
+    {
+      "epoch": 19.38,
+      "learning_rate": 1.7140552530752168e-06,
+      "loss": 0.068,
+      "step": 10680
+    },
+    {
+      "epoch": 19.4,
+      "learning_rate": 1.663641863278887e-06,
+      "loss": 0.0547,
+      "step": 10690
+    },
+    {
+      "epoch": 19.42,
+      "learning_rate": 1.613228473482557e-06,
+      "loss": 0.056,
+      "step": 10700
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.5628150836862272e-06,
+      "loss": 0.0583,
+      "step": 10710
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.5124016938898973e-06,
+      "loss": 0.0478,
+      "step": 10720
+    },
+    {
+      "epoch": 19.47,
+      "learning_rate": 1.4619883040935671e-06,
+      "loss": 0.0628,
+      "step": 10730
+    },
+    {
+      "epoch": 19.49,
+      "learning_rate": 1.4115749142972374e-06,
+      "loss": 0.068,
+      "step": 10740
+    },
+    {
+      "epoch": 19.51,
+      "learning_rate": 1.3611615245009075e-06,
+      "loss": 0.059,
+      "step": 10750
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 1.3107481347045776e-06,
+      "loss": 0.0569,
+      "step": 10760
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 1.2603347449082477e-06,
+      "loss": 0.0542,
+      "step": 10770
+    },
+    {
+      "epoch": 19.56,
+      "learning_rate": 1.2099213551119177e-06,
+      "loss": 0.0704,
+      "step": 10780
+    },
+    {
+      "epoch": 19.58,
+      "learning_rate": 1.1595079653155878e-06,
+      "loss": 0.06,
+      "step": 10790
+    },
+    {
+      "epoch": 19.6,
+      "learning_rate": 1.109094575519258e-06,
+      "loss": 0.0607,
+      "step": 10800
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 1.0586811857229282e-06,
+      "loss": 0.0545,
+      "step": 10810
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.008267795926598e-06,
+      "loss": 0.0564,
+      "step": 10820
+    },
+    {
+      "epoch": 19.65,
+      "learning_rate": 9.578544061302681e-07,
+      "loss": 0.0601,
+      "step": 10830
+    },
+    {
+      "epoch": 19.67,
+      "learning_rate": 9.074410163339384e-07,
+      "loss": 0.0705,
+      "step": 10840
+    },
+    {
+      "epoch": 19.69,
+      "learning_rate": 8.570276265376084e-07,
+      "loss": 0.0502,
+      "step": 10850
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 8.066142367412785e-07,
+      "loss": 0.053,
+      "step": 10860
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 7.562008469449486e-07,
+      "loss": 0.0389,
+      "step": 10870
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 7.057874571486187e-07,
+      "loss": 0.0462,
+      "step": 10880
+    },
+    {
+      "epoch": 19.76,
+      "learning_rate": 6.553740673522888e-07,
+      "loss": 0.0458,
+      "step": 10890
+    },
+    {
+      "epoch": 19.78,
+      "learning_rate": 6.049606775559589e-07,
+      "loss": 0.0658,
+      "step": 10900
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 5.54547287759629e-07,
+      "loss": 0.0629,
+      "step": 10910
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 5.04133897963299e-07,
+      "loss": 0.0397,
+      "step": 10920
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 4.537205081669692e-07,
+      "loss": 0.0508,
+      "step": 10930
+    },
+    {
+      "epoch": 19.85,
+      "learning_rate": 4.0330711837063923e-07,
+      "loss": 0.0545,
+      "step": 10940
+    },
+    {
+      "epoch": 19.87,
+      "learning_rate": 3.5289372857430936e-07,
+      "loss": 0.0522,
+      "step": 10950
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 3.0248033877797943e-07,
+      "loss": 0.067,
+      "step": 10960
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 2.520669489816495e-07,
+      "loss": 0.0611,
+      "step": 10970
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 2.0165355918531961e-07,
+      "loss": 0.0603,
+      "step": 10980
+    },
+    {
+      "epoch": 19.94,
+      "learning_rate": 1.5124016938898972e-07,
+      "loss": 0.0574,
+      "step": 10990
+    },
+    {
+      "epoch": 19.96,
+      "learning_rate": 1.0082677959265981e-07,
+      "loss": 0.0433,
+      "step": 11000
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 5.0413389796329904e-08,
+      "loss": 0.0533,
+      "step": 11010
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0,
+      "loss": 0.0596,
+      "step": 11020
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.9909863945578231,
+      "eval_loss": 0.028533633798360825,
+      "eval_runtime": 86.8566,
+      "eval_samples_per_second": 135.396,
+      "eval_steps_per_second": 4.237,
+      "step": 11020
+    },
+    {
+      "epoch": 20.0,
+      "step": 11020,
+      "total_flos": 1.0935764858913754e+20,
+      "train_loss": 0.15794360369119367,
+      "train_runtime": 22556.9928,
+      "train_samples_per_second": 62.562,
+      "train_steps_per_second": 0.489
+    }
+  ],
+  "max_steps": 11020,
+  "num_train_epochs": 20,
+  "total_flos": 1.0935764858913754e+20,
+  "trial_name": null,
+  "trial_params": null
+}