diff --git "a/checkpoint-11000/trainer_state.json" "b/checkpoint-11000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-11000/trainer_state.json"
@@ -0,0 +1,6715 @@
+{
+  "best_metric": 0.7957639098167419,
+  "best_model_checkpoint": "./vit-base-beans/checkpoint-3000",
+  "epoch": 14.102564102564102,
+  "global_step": 11000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019982905982905984,
+      "loss": 1.8581,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019965811965811967,
+      "loss": 1.7758,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001994871794871795,
+      "loss": 1.7018,
+      "step": 30
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019931623931623932,
+      "loss": 1.5662,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019914529914529915,
+      "loss": 1.6721,
+      "step": 50
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019897435897435898,
+      "loss": 1.5795,
+      "step": 60
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001988034188034188,
+      "loss": 1.5471,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019863247863247864,
+      "loss": 1.5363,
+      "step": 80
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019846153846153847,
+      "loss": 1.4139,
+      "step": 90
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001982905982905983,
+      "loss": 1.4193,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019811965811965813,
+      "loss": 1.3853,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019794871794871796,
+      "loss": 1.3645,
+      "step": 120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019777777777777778,
+      "loss": 1.4095,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001976068376068376,
+      "loss": 1.4114,
+      "step": 140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00019743589743589744,
+      "loss": 1.433,
+      "step": 150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00019726495726495727,
+      "loss": 1.5094,
+      "step": 160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00019709401709401713,
+      "loss": 1.2352,
+      "step": 170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00019692307692307696,
+      "loss": 1.2916,
+      "step": 180
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00019675213675213676,
+      "loss": 1.2965,
+      "step": 190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00019658119658119659,
+      "loss": 1.3053,
+      "step": 200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00019641025641025642,
+      "loss": 1.3689,
+      "step": 210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00019623931623931624,
+      "loss": 1.3519,
+      "step": 220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00019606837606837607,
+      "loss": 1.4205,
+      "step": 230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001958974358974359,
+      "loss": 1.2702,
+      "step": 240
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00019572649572649573,
+      "loss": 1.2631,
+      "step": 250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00019555555555555556,
+      "loss": 1.2163,
+      "step": 260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001953846153846154,
+      "loss": 1.1323,
+      "step": 270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00019521367521367522,
+      "loss": 1.3013,
+      "step": 280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00019504273504273505,
+      "loss": 1.1129,
+      "step": 290
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00019487179487179487,
+      "loss": 1.4559,
+      "step": 300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001947008547008547,
+      "loss": 1.2784,
+      "step": 310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00019452991452991453,
+      "loss": 1.3402,
+      "step": 320
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00019435897435897436,
+      "loss": 1.2687,
+      "step": 330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0001941880341880342,
+      "loss": 1.1328,
+      "step": 340
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00019401709401709402,
+      "loss": 1.1216,
+      "step": 350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00019384615384615385,
+      "loss": 1.0489,
+      "step": 360
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00019367521367521368,
+      "loss": 1.0993,
+      "step": 370
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001935042735042735,
+      "loss": 1.3569,
+      "step": 380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00019333333333333333,
+      "loss": 0.9691,
+      "step": 390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00019316239316239316,
+      "loss": 1.1364,
+      "step": 400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000192991452991453,
+      "loss": 1.104,
+      "step": 410
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00019282051282051282,
+      "loss": 1.1839,
+      "step": 420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00019264957264957265,
+      "loss": 1.1365,
+      "step": 430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0001924786324786325,
+      "loss": 0.9201,
+      "step": 440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00019230769230769233,
+      "loss": 1.1012,
+      "step": 450
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00019213675213675216,
+      "loss": 1.042,
+      "step": 460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000191965811965812,
+      "loss": 0.9697,
+      "step": 470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00019179487179487182,
+      "loss": 1.4043,
+      "step": 480
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00019162393162393165,
+      "loss": 1.2395,
+      "step": 490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00019145299145299148,
+      "loss": 1.193,
+      "step": 500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0001912820512820513,
+      "loss": 1.2216,
+      "step": 510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00019111111111111114,
+      "loss": 1.0667,
+      "step": 520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00019094017094017097,
+      "loss": 1.0327,
+      "step": 530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0001907692307692308,
+      "loss": 1.0287,
+      "step": 540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001905982905982906,
+      "loss": 1.0963,
+      "step": 550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019042735042735042,
+      "loss": 1.0812,
+      "step": 560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00019025641025641025,
+      "loss": 1.0625,
+      "step": 570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00019008547008547008,
+      "loss": 0.8884,
+      "step": 580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001899145299145299,
+      "loss": 0.9661,
+      "step": 590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00018974358974358974,
+      "loss": 0.951,
+      "step": 600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00018957264957264957,
+      "loss": 0.93,
+      "step": 610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001894017094017094,
+      "loss": 0.8576,
+      "step": 620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 0.00018923076923076923,
+      "loss": 1.2505,
+      "step": 630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 0.00018905982905982906,
+      "loss": 0.9469,
+      "step": 640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 0.9869,
+      "step": 650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 0.0001887179487179487,
+      "loss": 1.1052,
+      "step": 660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 0.00018854700854700854,
+      "loss": 1.0779,
+      "step": 670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 0.00018837606837606837,
+      "loss": 1.0328,
+      "step": 680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 0.0001882051282051282,
+      "loss": 1.0377,
+      "step": 690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.00018803418803418803,
+      "loss": 0.9489,
+      "step": 700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 0.00018786324786324786,
+      "loss": 0.8715,
+      "step": 710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 0.0001876923076923077,
+      "loss": 0.9167,
+      "step": 720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 0.00018752136752136754,
+      "loss": 0.9519,
+      "step": 730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 0.00018735042735042737,
+      "loss": 0.8548,
+      "step": 740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 0.0001871794871794872,
+      "loss": 0.827,
+      "step": 750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 0.00018700854700854703,
+      "loss": 0.9922,
+      "step": 760
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 0.00018683760683760686,
+      "loss": 0.902,
+      "step": 770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0001866666666666667,
+      "loss": 0.8385,
+      "step": 780
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 0.00018649572649572652,
+      "loss": 0.5939,
+      "step": 790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 0.00018632478632478634,
+      "loss": 0.8229,
+      "step": 800
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 0.00018615384615384617,
+      "loss": 0.6577,
+      "step": 810
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 0.000185982905982906,
+      "loss": 0.7075,
+      "step": 820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 0.00018581196581196583,
+      "loss": 0.6764,
+      "step": 830
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 0.00018564102564102566,
+      "loss": 0.5858,
+      "step": 840
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 0.0001854700854700855,
+      "loss": 0.8485,
+      "step": 850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 0.00018529914529914532,
+      "loss": 0.6396,
+      "step": 860
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 0.00018512820512820515,
+      "loss": 0.6054,
+      "step": 870
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 0.00018495726495726497,
+      "loss": 0.664,
+      "step": 880
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 0.0001847863247863248,
+      "loss": 0.5736,
+      "step": 890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 0.00018461538461538463,
+      "loss": 0.6546,
+      "step": 900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 0.00018444444444444446,
+      "loss": 0.6295,
+      "step": 910
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 0.00018427350427350426,
+      "loss": 0.7309,
+      "step": 920
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 0.0001841025641025641,
+      "loss": 0.8071,
+      "step": 930
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 0.00018393162393162392,
+      "loss": 0.667,
+      "step": 940
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 0.00018376068376068375,
+      "loss": 0.7265,
+      "step": 950
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 0.00018358974358974358,
+      "loss": 0.6872,
+      "step": 960
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 0.0001834188034188034,
+      "loss": 0.8474,
+      "step": 970
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 0.00018324786324786324,
+      "loss": 0.6226,
+      "step": 980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 0.0001830769230769231,
+      "loss": 0.5453,
+      "step": 990
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 0.00018290598290598292,
+      "loss": 0.6832,
+      "step": 1000
+    },
+    {
+      "epoch": 1.28,
+      "eval_accuracy": 0.6526374859708193,
+      "eval_loss": 0.9223159551620483,
+      "eval_runtime": 34.8394,
+      "eval_samples_per_second": 51.206,
+      "eval_steps_per_second": 6.401,
+      "step": 1000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 0.00018273504273504275,
+      "loss": 0.7721,
+      "step": 1010
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 0.00018256410256410258,
+      "loss": 0.6434,
+      "step": 1020
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 0.0001823931623931624,
+      "loss": 0.7244,
+      "step": 1030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 0.00018222222222222224,
+      "loss": 0.6953,
+      "step": 1040
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 0.00018205128205128207,
+      "loss": 0.6965,
+      "step": 1050
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 0.0001818803418803419,
+      "loss": 0.7892,
+      "step": 1060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 0.00018170940170940172,
+      "loss": 0.6513,
+      "step": 1070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 0.00018153846153846155,
+      "loss": 0.6469,
+      "step": 1080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 0.00018136752136752138,
+      "loss": 0.6468,
+      "step": 1090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 0.0001811965811965812,
+      "loss": 0.7679,
+      "step": 1100
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 0.00018102564102564104,
+      "loss": 0.4376,
+      "step": 1110
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 0.00018085470085470087,
+      "loss": 0.844,
+      "step": 1120
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 0.0001806837606837607,
+      "loss": 0.7508,
+      "step": 1130
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 0.00018051282051282052,
+      "loss": 0.7154,
+      "step": 1140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 0.00018034188034188035,
+      "loss": 0.6632,
+      "step": 1150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 0.00018017094017094018,
+      "loss": 0.7109,
+      "step": 1160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 0.00018,
+      "loss": 0.6545,
+      "step": 1170
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 0.00017982905982905984,
+      "loss": 0.7316,
+      "step": 1180
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 0.00017965811965811967,
+      "loss": 0.7549,
+      "step": 1190
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 0.0001794871794871795,
+      "loss": 0.5296,
+      "step": 1200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 0.00017931623931623933,
+      "loss": 1.0734,
+      "step": 1210
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 0.00017914529914529916,
+      "loss": 0.6724,
+      "step": 1220
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 0.00017897435897435898,
+      "loss": 0.5692,
+      "step": 1230
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 0.0001788034188034188,
+      "loss": 0.6338,
+      "step": 1240
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 0.00017863247863247864,
+      "loss": 0.5972,
+      "step": 1250
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 0.00017846153846153847,
+      "loss": 0.7404,
+      "step": 1260
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 0.0001782905982905983,
+      "loss": 0.5386,
+      "step": 1270
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 0.00017811965811965813,
+      "loss": 0.775,
+      "step": 1280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 0.00017794871794871796,
+      "loss": 0.8524,
+      "step": 1290
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 0.00017777777777777779,
+      "loss": 0.6457,
+      "step": 1300
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 0.00017760683760683762,
+      "loss": 0.5095,
+      "step": 1310
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 0.00017743589743589744,
+      "loss": 0.7402,
+      "step": 1320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 0.00017726495726495727,
+      "loss": 0.4563,
+      "step": 1330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 0.0001770940170940171,
+      "loss": 0.5477,
+      "step": 1340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 0.00017692307692307693,
+      "loss": 0.7047,
+      "step": 1350
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 0.00017675213675213676,
+      "loss": 0.9488,
+      "step": 1360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 0.0001765811965811966,
+      "loss": 0.6216,
+      "step": 1370
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 0.00017641025641025642,
+      "loss": 0.6692,
+      "step": 1380
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 0.00017623931623931625,
+      "loss": 0.5363,
+      "step": 1390
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 0.00017606837606837607,
+      "loss": 0.512,
+      "step": 1400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 0.0001758974358974359,
+      "loss": 0.7907,
+      "step": 1410
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 0.00017572649572649573,
+      "loss": 0.674,
+      "step": 1420
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 0.00017555555555555556,
+      "loss": 0.6912,
+      "step": 1430
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 0.0001753846153846154,
+      "loss": 0.875,
+      "step": 1440
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 0.00017521367521367522,
+      "loss": 0.9192,
+      "step": 1450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 0.00017504273504273505,
+      "loss": 0.6845,
+      "step": 1460
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 0.00017487179487179488,
+      "loss": 0.9668,
+      "step": 1470
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 0.0001747008547008547,
+      "loss": 0.6267,
+      "step": 1480
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 0.00017452991452991453,
+      "loss": 0.7207,
+      "step": 1490
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 0.00017435897435897436,
+      "loss": 0.4895,
+      "step": 1500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 0.0001741880341880342,
+      "loss": 0.7237,
+      "step": 1510
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 0.00017401709401709402,
+      "loss": 0.702,
+      "step": 1520
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 0.00017384615384615385,
+      "loss": 0.6706,
+      "step": 1530
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 0.00017367521367521368,
+      "loss": 0.6043,
+      "step": 1540
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 0.0001735042735042735,
+      "loss": 0.7136,
+      "step": 1550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 0.00017333333333333334,
+      "loss": 0.5816,
+      "step": 1560
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 0.00017316239316239317,
+      "loss": 0.4576,
+      "step": 1570
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 0.000172991452991453,
+      "loss": 0.3339,
+      "step": 1580
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 0.00017282051282051285,
+      "loss": 0.4679,
+      "step": 1590
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 0.00017264957264957268,
+      "loss": 0.5122,
+      "step": 1600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 0.0001724786324786325,
+      "loss": 0.2515,
+      "step": 1610
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 0.00017230769230769234,
+      "loss": 0.3069,
+      "step": 1620
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 0.00017213675213675217,
+      "loss": 0.3612,
+      "step": 1630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 0.000171965811965812,
+      "loss": 0.4896,
+      "step": 1640
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 0.0001717948717948718,
+      "loss": 0.2227,
+      "step": 1650
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 0.00017162393162393162,
+      "loss": 0.3367,
+      "step": 1660
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 0.00017145299145299145,
+      "loss": 0.4207,
+      "step": 1670
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 0.00017128205128205128,
+      "loss": 0.3126,
+      "step": 1680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 0.0001711111111111111,
+      "loss": 0.2902,
+      "step": 1690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 0.00017094017094017094,
+      "loss": 0.6133,
+      "step": 1700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 0.00017076923076923077,
+      "loss": 0.4473,
+      "step": 1710
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 0.0001705982905982906,
+      "loss": 0.6531,
+      "step": 1720
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 0.00017042735042735043,
+      "loss": 0.5508,
+      "step": 1730
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 0.00017025641025641026,
+      "loss": 0.4197,
+      "step": 1740
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 0.00017008547008547008,
+      "loss": 0.4295,
+      "step": 1750
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 0.0001699145299145299,
+      "loss": 0.3714,
+      "step": 1760
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 0.00016974358974358974,
+      "loss": 0.4605,
+      "step": 1770
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 0.00016957264957264957,
+      "loss": 0.3979,
+      "step": 1780
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 0.0001694017094017094,
+      "loss": 0.3915,
+      "step": 1790
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 0.00016923076923076923,
+      "loss": 0.3594,
+      "step": 1800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 0.00016905982905982906,
+      "loss": 0.332,
+      "step": 1810
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 0.00016888888888888889,
+      "loss": 0.6412,
+      "step": 1820
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 0.00016871794871794871,
+      "loss": 0.5271,
+      "step": 1830
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 0.00016854700854700854,
+      "loss": 0.6231,
+      "step": 1840
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 0.00016837606837606837,
+      "loss": 0.4133,
+      "step": 1850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 0.00016820512820512823,
+      "loss": 0.3524,
+      "step": 1860
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 0.00016803418803418806,
+      "loss": 0.4834,
+      "step": 1870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 0.00016786324786324789,
+      "loss": 0.5658,
+      "step": 1880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 0.00016769230769230772,
+      "loss": 0.479,
+      "step": 1890
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 0.00016752136752136754,
+      "loss": 0.4138,
+      "step": 1900
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 0.00016735042735042737,
+      "loss": 0.4342,
+      "step": 1910
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 0.0001671794871794872,
+      "loss": 0.4909,
+      "step": 1920
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 0.00016700854700854703,
+      "loss": 0.4699,
+      "step": 1930
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 0.00016683760683760686,
+      "loss": 0.1936,
+      "step": 1940
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.4344,
+      "step": 1950
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 0.00016649572649572652,
+      "loss": 0.2826,
+      "step": 1960
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 0.00016632478632478635,
+      "loss": 0.3518,
+      "step": 1970
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 0.00016615384615384617,
+      "loss": 0.4507,
+      "step": 1980
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 0.000165982905982906,
+      "loss": 0.4549,
+      "step": 1990
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 0.00016581196581196583,
+      "loss": 0.348,
+      "step": 2000
+    },
+    {
+      "epoch": 2.56,
+      "eval_accuracy": 0.7008978675645342,
+      "eval_loss": 1.0017077922821045,
+      "eval_runtime": 39.0117,
+      "eval_samples_per_second": 45.73,
+      "eval_steps_per_second": 5.716,
+      "step": 2000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 0.00016564102564102566,
+      "loss": 0.7027,
+      "step": 2010
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 0.00016547008547008546,
+      "loss": 0.5353,
+      "step": 2020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 0.0001652991452991453,
+      "loss": 0.5958,
+      "step": 2030
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 0.00016512820512820512,
+      "loss": 0.4524,
+      "step": 2040
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 0.00016495726495726495,
+      "loss": 0.4626,
+      "step": 2050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 0.00016478632478632478,
+      "loss": 0.4503,
+      "step": 2060
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 0.0001646153846153846,
+      "loss": 0.5792,
+      "step": 2070
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 0.00016444444444444444,
+      "loss": 0.3757,
+      "step": 2080
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 0.00016427350427350426,
+      "loss": 0.4774,
+      "step": 2090
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 0.0001641025641025641,
+      "loss": 0.6365,
+      "step": 2100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 0.00016393162393162392,
+      "loss": 0.2231,
+      "step": 2110
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 0.00016376068376068375,
+      "loss": 0.5641,
+      "step": 2120
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 0.00016358974358974358,
+      "loss": 0.3998,
+      "step": 2130
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 0.00016341880341880344,
+      "loss": 0.4461,
+      "step": 2140
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 0.00016324786324786327,
+      "loss": 0.3812,
+      "step": 2150
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 0.0001630769230769231,
+      "loss": 0.407,
+      "step": 2160
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 0.00016290598290598292,
+      "loss": 0.5142,
+      "step": 2170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 0.00016273504273504275,
+      "loss": 0.3531,
+      "step": 2180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 0.00016256410256410258,
+      "loss": 0.4287,
+      "step": 2190
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 0.0001623931623931624,
+      "loss": 0.4242,
+      "step": 2200
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 0.00016222222222222224,
+      "loss": 0.282,
+      "step": 2210
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 0.00016205128205128207,
+      "loss": 0.5312,
+      "step": 2220
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 0.0001618803418803419,
+      "loss": 0.5925,
+      "step": 2230
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 0.00016170940170940172,
+      "loss": 0.3919,
+      "step": 2240
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 0.00016153846153846155,
+      "loss": 0.6241,
+      "step": 2250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 0.00016136752136752138,
+      "loss": 0.3143,
+      "step": 2260
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 0.0001611965811965812,
+      "loss": 0.5144,
+      "step": 2270
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 0.00016102564102564104,
+      "loss": 0.1882,
+      "step": 2280
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 0.00016085470085470087,
+      "loss": 0.3812,
+      "step": 2290
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 0.0001606837606837607,
+      "loss": 0.2284,
+      "step": 2300
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 0.00016051282051282053,
+      "loss": 0.3319,
+      "step": 2310
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 0.00016034188034188036,
+      "loss": 0.7971,
+      "step": 2320
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 0.00016017094017094018,
+      "loss": 0.3977,
+      "step": 2330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.00016,
+      "loss": 0.5652,
+      "step": 2340
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 0.00015982905982905984,
+      "loss": 0.2831,
+      "step": 2350
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 0.00015965811965811967,
+      "loss": 0.1566,
+      "step": 2360
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 0.0001594871794871795,
+      "loss": 0.1572,
+      "step": 2370
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 0.0001593162393162393,
+      "loss": 0.2303,
+      "step": 2380
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 0.00015914529914529913,
+      "loss": 0.3149,
+      "step": 2390
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00015897435897435896,
+      "loss": 0.3658,
+      "step": 2400
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 0.00015880341880341881,
+      "loss": 0.1089,
+      "step": 2410
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 0.00015863247863247864,
+      "loss": 0.3972,
+      "step": 2420
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 0.00015846153846153847,
+      "loss": 0.2249,
+      "step": 2430
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 0.0001582905982905983,
+      "loss": 0.1594,
+      "step": 2440
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 0.00015811965811965813,
+      "loss": 0.2473,
+      "step": 2450
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 0.00015794871794871796,
+      "loss": 0.3014,
+      "step": 2460
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 0.0001577777777777778,
+      "loss": 0.2009,
+      "step": 2470
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 0.00015760683760683762,
+      "loss": 0.2739,
+      "step": 2480
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 0.00015743589743589745,
+      "loss": 0.3456,
+      "step": 2490
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 0.00015726495726495727,
+      "loss": 0.201,
+      "step": 2500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 0.0001570940170940171,
+      "loss": 0.4472,
+      "step": 2510
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 0.00015692307692307693,
+      "loss": 0.3452,
+      "step": 2520
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 0.00015675213675213676,
+      "loss": 0.1802,
+      "step": 2530
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 0.0001565811965811966,
+      "loss": 0.3302,
+      "step": 2540
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 0.00015641025641025642,
+      "loss": 0.2736,
+      "step": 2550
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 0.00015623931623931625,
+      "loss": 0.3162,
+      "step": 2560
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 0.00015606837606837608,
+      "loss": 0.1153,
+      "step": 2570
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 0.0001558974358974359,
+      "loss": 0.1698,
+      "step": 2580
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 0.00015572649572649573,
+      "loss": 0.4336,
+      "step": 2590
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 0.00015555555555555556,
+      "loss": 0.1784,
+      "step": 2600
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 0.0001553846153846154,
+      "loss": 0.2714,
+      "step": 2610
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 0.00015521367521367522,
+      "loss": 0.2657,
+      "step": 2620
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 0.00015504273504273505,
+      "loss": 0.3528,
+      "step": 2630
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 0.00015487179487179488,
+      "loss": 0.2416,
+      "step": 2640
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 0.0001547008547008547,
+      "loss": 0.1516,
+      "step": 2650
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 0.00015452991452991454,
+      "loss": 0.3435,
+      "step": 2660
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 0.00015435897435897436,
+      "loss": 0.2755,
+      "step": 2670
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 0.0001541880341880342,
+      "loss": 0.3667,
+      "step": 2680
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 0.00015401709401709402,
+      "loss": 0.2273,
+      "step": 2690
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 0.00015384615384615385,
+      "loss": 0.3083,
+      "step": 2700
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 0.00015367521367521368,
+      "loss": 0.1352,
+      "step": 2710
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 0.0001535042735042735,
+      "loss": 0.2642,
+      "step": 2720
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 0.00015333333333333334,
+      "loss": 0.2947,
+      "step": 2730
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 0.0001531623931623932,
+      "loss": 0.2894,
+      "step": 2740
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 0.000152991452991453,
+      "loss": 0.3171,
+      "step": 2750
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 0.00015282051282051282,
+      "loss": 0.1399,
+      "step": 2760
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 0.00015264957264957265,
+      "loss": 0.3285,
+      "step": 2770
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 0.00015247863247863248,
+      "loss": 0.3482,
+      "step": 2780
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 0.0001523076923076923,
+      "loss": 0.2303,
+      "step": 2790
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 0.00015213675213675214,
+      "loss": 0.1022,
+      "step": 2800
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 0.00015196581196581197,
+      "loss": 0.5218,
+      "step": 2810
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 0.0001517948717948718,
+      "loss": 0.1958,
+      "step": 2820
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 0.00015162393162393163,
+      "loss": 0.2601,
+      "step": 2830
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 0.00015145299145299146,
+      "loss": 0.2364,
+      "step": 2840
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 0.00015128205128205128,
+      "loss": 0.2623,
+      "step": 2850
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 0.0001511111111111111,
+      "loss": 0.3103,
+      "step": 2860
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 0.00015094017094017094,
+      "loss": 0.5716,
+      "step": 2870
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 0.00015076923076923077,
+      "loss": 0.1348,
+      "step": 2880
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 0.0001505982905982906,
+      "loss": 0.1907,
+      "step": 2890
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 0.00015042735042735043,
+      "loss": 0.1671,
+      "step": 2900
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 0.00015025641025641026,
+      "loss": 0.4659,
+      "step": 2910
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 0.00015008547008547009,
+      "loss": 0.1819,
+      "step": 2920
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 0.00014991452991452991,
+      "loss": 0.57,
+      "step": 2930
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 0.00014974358974358974,
+      "loss": 0.4177,
+      "step": 2940
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 0.00014957264957264957,
+      "loss": 0.2709,
+      "step": 2950
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 0.0001494017094017094,
+      "loss": 0.3616,
+      "step": 2960
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 0.00014923076923076923,
+      "loss": 0.2519,
+      "step": 2970
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 0.00014905982905982906,
+      "loss": 0.3444,
+      "step": 2980
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 0.0001488888888888889,
+      "loss": 0.4528,
+      "step": 2990
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00014871794871794872,
+      "loss": 0.3906,
+      "step": 3000
+    },
+    {
+      "epoch": 3.85,
+      "eval_accuracy": 0.7536475869809203,
+      "eval_loss": 0.7957639098167419,
+      "eval_runtime": 34.4592,
+      "eval_samples_per_second": 51.771,
+      "eval_steps_per_second": 6.471,
+      "step": 3000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 0.00014854700854700857,
+      "loss": 0.3384,
+      "step": 3010
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 0.0001483760683760684,
+      "loss": 0.1817,
+      "step": 3020
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 0.00014820512820512823,
+      "loss": 0.4493,
+      "step": 3030
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 0.00014803418803418806,
+      "loss": 0.2129,
+      "step": 3040
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 0.0001478632478632479,
+      "loss": 0.3385,
+      "step": 3050
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 0.00014769230769230772,
+      "loss": 0.2875,
+      "step": 3060
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 0.00014752136752136755,
+      "loss": 0.3154,
+      "step": 3070
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 0.00014735042735042737,
+      "loss": 0.2525,
+      "step": 3080
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 0.0001471794871794872,
+      "loss": 0.3115,
+      "step": 3090
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 0.00014700854700854703,
+      "loss": 0.2375,
+      "step": 3100
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 0.00014683760683760683,
+      "loss": 0.121,
+      "step": 3110
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.00014666666666666666,
+      "loss": 0.2368,
+      "step": 3120
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 0.0001464957264957265,
+      "loss": 0.1913,
+      "step": 3130
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 0.00014632478632478632,
+      "loss": 0.2868,
+      "step": 3140
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 0.00014615384615384615,
+      "loss": 0.1729,
+      "step": 3150
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 0.00014598290598290598,
+      "loss": 0.1153,
+      "step": 3160
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 0.0001458119658119658,
+      "loss": 0.1036,
+      "step": 3170
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 0.00014564102564102564,
+      "loss": 0.0974,
+      "step": 3180
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 0.00014547008547008546,
+      "loss": 0.1176,
+      "step": 3190
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 0.0001452991452991453,
+      "loss": 0.1305,
+      "step": 3200
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 0.00014512820512820512,
+      "loss": 0.1496,
+      "step": 3210
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 0.00014495726495726495,
+      "loss": 0.143,
+      "step": 3220
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 0.00014478632478632478,
+      "loss": 0.0647,
+      "step": 3230
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 0.0001446153846153846,
+      "loss": 0.0355,
+      "step": 3240
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 0.00014444444444444444,
+      "loss": 0.0766,
+      "step": 3250
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 0.00014427350427350427,
+      "loss": 0.1553,
+      "step": 3260
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 0.0001441025641025641,
+      "loss": 0.1441,
+      "step": 3270
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 0.00014393162393162392,
+      "loss": 0.0466,
+      "step": 3280
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 0.00014376068376068378,
+      "loss": 0.0734,
+      "step": 3290
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 0.0001435897435897436,
+      "loss": 0.0801,
+      "step": 3300
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 0.00014341880341880344,
+      "loss": 0.1471,
+      "step": 3310
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 0.00014324786324786327,
+      "loss": 0.1419,
+      "step": 3320
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 0.0001430769230769231,
+      "loss": 0.1019,
+      "step": 3330
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 0.00014290598290598292,
+      "loss": 0.2108,
+      "step": 3340
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 0.00014273504273504275,
+      "loss": 0.0273,
+      "step": 3350
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 0.00014256410256410258,
+      "loss": 0.2777,
+      "step": 3360
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 0.0001423931623931624,
+      "loss": 0.085,
+      "step": 3370
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 0.00014222222222222224,
+      "loss": 0.136,
+      "step": 3380
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 0.00014205128205128207,
+      "loss": 0.3193,
+      "step": 3390
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 0.0001418803418803419,
+      "loss": 0.1982,
+      "step": 3400
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 0.00014170940170940173,
+      "loss": 0.1471,
+      "step": 3410
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 0.00014153846153846156,
+      "loss": 0.1064,
+      "step": 3420
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 0.00014136752136752138,
+      "loss": 0.2093,
+      "step": 3430
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 0.0001411965811965812,
+      "loss": 0.3454,
+      "step": 3440
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 0.00014102564102564104,
+      "loss": 0.1812,
+      "step": 3450
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 0.00014085470085470087,
+      "loss": 0.146,
+      "step": 3460
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 0.0001406837606837607,
+      "loss": 0.1797,
+      "step": 3470
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 0.0001405128205128205,
+      "loss": 0.1049,
+      "step": 3480
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 0.00014034188034188033,
+      "loss": 0.1417,
+      "step": 3490
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 0.00014017094017094016,
+      "loss": 0.1369,
+      "step": 3500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 0.00014,
+      "loss": 0.0872,
+      "step": 3510
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 0.00013982905982905982,
+      "loss": 0.0559,
+      "step": 3520
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 0.00013965811965811965,
+      "loss": 0.2372,
+      "step": 3530
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 0.00013948717948717947,
+      "loss": 0.4092,
+      "step": 3540
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 0.0001393162393162393,
+      "loss": 0.2881,
+      "step": 3550
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 0.00013914529914529916,
+      "loss": 0.3194,
+      "step": 3560
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 0.000138974358974359,
+      "loss": 0.4236,
+      "step": 3570
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 0.00013880341880341882,
+      "loss": 0.0805,
+      "step": 3580
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 0.00013863247863247865,
+      "loss": 0.1488,
+      "step": 3590
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00013846153846153847,
+      "loss": 0.1332,
+      "step": 3600
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 0.0001382905982905983,
+      "loss": 0.2034,
+      "step": 3610
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 0.00013811965811965813,
+      "loss": 0.1837,
+      "step": 3620
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 0.00013794871794871796,
+      "loss": 0.156,
+      "step": 3630
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 0.0001377777777777778,
+      "loss": 0.1857,
+      "step": 3640
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 0.00013760683760683762,
+      "loss": 0.1481,
+      "step": 3650
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 0.00013743589743589745,
+      "loss": 0.3842,
+      "step": 3660
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 0.00013726495726495728,
+      "loss": 0.1999,
+      "step": 3670
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 0.0001370940170940171,
+      "loss": 0.3166,
+      "step": 3680
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 0.00013692307692307693,
+      "loss": 0.3671,
+      "step": 3690
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 0.00013675213675213676,
+      "loss": 0.32,
+      "step": 3700
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 0.0001365811965811966,
+      "loss": 0.1638,
+      "step": 3710
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 0.00013641025641025642,
+      "loss": 0.0734,
+      "step": 3720
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 0.00013623931623931625,
+      "loss": 0.115,
+      "step": 3730
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 0.00013606837606837608,
+      "loss": 0.1742,
+      "step": 3740
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 0.0934,
+      "step": 3750
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 0.00013572649572649574,
+      "loss": 0.2705,
+      "step": 3760
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 0.00013555555555555556,
+      "loss": 0.0109,
+      "step": 3770
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 0.0001353846153846154,
+      "loss": 0.2112,
+      "step": 3780
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 0.00013521367521367522,
+      "loss": 0.0508,
+      "step": 3790
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 0.00013504273504273505,
+      "loss": 0.1096,
+      "step": 3800
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 0.00013487179487179488,
+      "loss": 0.2732,
+      "step": 3810
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 0.0001347008547008547,
+      "loss": 0.186,
+      "step": 3820
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 0.00013452991452991454,
+      "loss": 0.1218,
+      "step": 3830
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 0.00013435897435897437,
+      "loss": 0.2359,
+      "step": 3840
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 0.0001341880341880342,
+      "loss": 0.2884,
+      "step": 3850
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 0.00013401709401709402,
+      "loss": 0.286,
+      "step": 3860
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 0.00013384615384615385,
+      "loss": 0.0854,
+      "step": 3870
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 0.00013367521367521368,
+      "loss": 0.0308,
+      "step": 3880
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 0.0001335042735042735,
+      "loss": 0.1327,
+      "step": 3890
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.1869,
+      "step": 3900
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 0.00013316239316239317,
+      "loss": 0.1636,
+      "step": 3910
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 0.000132991452991453,
+      "loss": 0.079,
+      "step": 3920
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 0.00013282051282051283,
+      "loss": 0.1349,
+      "step": 3930
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 0.00013264957264957266,
+      "loss": 0.0564,
+      "step": 3940
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 0.00013247863247863248,
+      "loss": 0.1325,
+      "step": 3950
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 0.0001323076923076923,
+      "loss": 0.1156,
+      "step": 3960
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 0.00013213675213675214,
+      "loss": 0.1017,
+      "step": 3970
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 0.00013196581196581197,
+      "loss": 0.1198,
+      "step": 3980
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 0.0001317948717948718,
+      "loss": 0.1086,
+      "step": 3990
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 0.00013162393162393163,
+      "loss": 0.1966,
+      "step": 4000
+    },
+    {
+      "epoch": 5.13,
+      "eval_accuracy": 0.7727272727272727,
+      "eval_loss": 1.0012712478637695,
+      "eval_runtime": 36.9299,
+      "eval_samples_per_second": 48.308,
+      "eval_steps_per_second": 6.038,
+      "step": 4000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 0.00013145299145299146,
+      "loss": 0.0625,
+      "step": 4010
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 0.00013128205128205129,
+      "loss": 0.0244,
+      "step": 4020
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 0.00013111111111111111,
+      "loss": 0.0416,
+      "step": 4030
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 0.00013094017094017094,
+      "loss": 0.0371,
+      "step": 4040
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 0.00013076923076923077,
+      "loss": 0.075,
+      "step": 4050
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 0.0001305982905982906,
+      "loss": 0.0265,
+      "step": 4060
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 0.00013042735042735043,
+      "loss": 0.0334,
+      "step": 4070
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 0.00013025641025641026,
+      "loss": 0.1455,
+      "step": 4080
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 0.0001300854700854701,
+      "loss": 0.1084,
+      "step": 4090
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 0.00012991452991452992,
+      "loss": 0.0283,
+      "step": 4100
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 0.00012974358974358975,
+      "loss": 0.1102,
+      "step": 4110
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 0.00012957264957264957,
+      "loss": 0.0225,
+      "step": 4120
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 0.0001294017094017094,
+      "loss": 0.04,
+      "step": 4130
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 0.00012923076923076923,
+      "loss": 0.1175,
+      "step": 4140
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 0.00012905982905982906,
+      "loss": 0.0439,
+      "step": 4150
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 0.00012888888888888892,
+      "loss": 0.101,
+      "step": 4160
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 0.00012871794871794875,
+      "loss": 0.0631,
+      "step": 4170
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 0.00012854700854700857,
+      "loss": 0.0275,
+      "step": 4180
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 0.0001283760683760684,
+      "loss": 0.1865,
+      "step": 4190
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00012820512820512823,
+      "loss": 0.1747,
+      "step": 4200
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 0.00012803418803418803,
+      "loss": 0.0581,
+      "step": 4210
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 0.00012786324786324786,
+      "loss": 0.5311,
+      "step": 4220
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 0.0001276923076923077,
+      "loss": 0.019,
+      "step": 4230
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 0.00012752136752136752,
+      "loss": 0.1085,
+      "step": 4240
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 0.00012735042735042735,
+      "loss": 0.2244,
+      "step": 4250
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 0.00012717948717948718,
+      "loss": 0.2485,
+      "step": 4260
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 0.000127008547008547,
+      "loss": 0.1385,
+      "step": 4270
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 0.00012683760683760684,
+      "loss": 0.1264,
+      "step": 4280
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 0.00012666666666666666,
+      "loss": 0.1278,
+      "step": 4290
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 0.0001264957264957265,
+      "loss": 0.0181,
+      "step": 4300
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 0.00012632478632478632,
+      "loss": 0.1171,
+      "step": 4310
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 0.00012615384615384615,
+      "loss": 0.1718,
+      "step": 4320
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 0.00012598290598290598,
+      "loss": 0.0421,
+      "step": 4330
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 0.0001258119658119658,
+      "loss": 0.0212,
+      "step": 4340
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 0.00012564102564102564,
+      "loss": 0.0426,
+      "step": 4350
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 0.00012547008547008547,
+      "loss": 0.0028,
+      "step": 4360
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 0.0001252991452991453,
+      "loss": 0.0967,
+      "step": 4370
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 0.00012512820512820512,
+      "loss": 0.0351,
+      "step": 4380
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 0.00012495726495726495,
+      "loss": 0.0242,
+      "step": 4390
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 0.00012478632478632478,
+      "loss": 0.328,
+      "step": 4400
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 0.0001246153846153846,
+      "loss": 0.1125,
+      "step": 4410
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 0.00012444444444444444,
+      "loss": 0.0525,
+      "step": 4420
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 0.0001242735042735043,
+      "loss": 0.1387,
+      "step": 4430
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 0.00012410256410256412,
+      "loss": 0.22,
+      "step": 4440
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 0.00012393162393162395,
+      "loss": 0.1024,
+      "step": 4450
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 0.00012376068376068378,
+      "loss": 0.1019,
+      "step": 4460
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 0.0001235897435897436,
+      "loss": 0.0114,
+      "step": 4470
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 0.00012341880341880344,
+      "loss": 0.0043,
+      "step": 4480
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 0.00012324786324786327,
+      "loss": 0.0503,
+      "step": 4490
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 0.0001230769230769231,
+      "loss": 0.1472,
+      "step": 4500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 0.00012290598290598293,
+      "loss": 0.3344,
+      "step": 4510
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 0.00012273504273504276,
+      "loss": 0.1512,
+      "step": 4520
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 0.00012256410256410258,
+      "loss": 0.1359,
+      "step": 4530
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 0.0001223931623931624,
+      "loss": 0.0761,
+      "step": 4540
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 0.00012222222222222224,
+      "loss": 0.2204,
+      "step": 4550
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 0.00012205128205128207,
+      "loss": 0.121,
+      "step": 4560
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 0.0001218803418803419,
+      "loss": 0.0244,
+      "step": 4570
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 0.0001217094017094017,
+      "loss": 0.0192,
+      "step": 4580
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 0.00012153846153846153,
+      "loss": 0.017,
+      "step": 4590
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 0.00012136752136752136,
+      "loss": 0.1092,
+      "step": 4600
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 0.00012119658119658119,
+      "loss": 0.1026,
+      "step": 4610
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 0.00012102564102564103,
+      "loss": 0.089,
+      "step": 4620
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 0.00012085470085470086,
+      "loss": 0.0469,
+      "step": 4630
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 0.00012068376068376069,
+      "loss": 0.1308,
+      "step": 4640
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 0.00012051282051282052,
+      "loss": 0.1473,
+      "step": 4650
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 0.00012034188034188035,
+      "loss": 0.1271,
+      "step": 4660
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 0.00012017094017094017,
+      "loss": 0.2707,
+      "step": 4670
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 0.00012,
+      "loss": 0.1334,
+      "step": 4680
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 0.00011982905982905983,
+      "loss": 0.0675,
+      "step": 4690
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 0.00011965811965811966,
+      "loss": 0.025,
+      "step": 4700
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 0.00011948717948717949,
+      "loss": 0.0571,
+      "step": 4710
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 0.00011931623931623932,
+      "loss": 0.131,
+      "step": 4720
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 0.00011914529914529915,
+      "loss": 0.0201,
+      "step": 4730
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 0.00011897435897435898,
+      "loss": 0.1559,
+      "step": 4740
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 0.0001188034188034188,
+      "loss": 0.0599,
+      "step": 4750
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 0.00011863247863247863,
+      "loss": 0.0502,
+      "step": 4760
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 0.00011846153846153846,
+      "loss": 0.0041,
+      "step": 4770
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 0.00011829059829059829,
+      "loss": 0.0315,
+      "step": 4780
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 0.00011811965811965813,
+      "loss": 0.0132,
+      "step": 4790
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00011794871794871796,
+      "loss": 0.0755,
+      "step": 4800
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 0.00011777777777777779,
+      "loss": 0.1349,
+      "step": 4810
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 0.00011760683760683762,
+      "loss": 0.0186,
+      "step": 4820
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 0.00011743589743589745,
+      "loss": 0.0224,
+      "step": 4830
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 0.00011726495726495728,
+      "loss": 0.0728,
+      "step": 4840
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 0.00011709401709401711,
+      "loss": 0.0094,
+      "step": 4850
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 0.00011692307692307694,
+      "loss": 0.1172,
+      "step": 4860
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 0.00011675213675213676,
+      "loss": 0.0078,
+      "step": 4870
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 0.0001165811965811966,
+      "loss": 0.1227,
+      "step": 4880
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 0.00011641025641025642,
+      "loss": 0.0078,
+      "step": 4890
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 0.00011623931623931625,
+      "loss": 0.0053,
+      "step": 4900
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 0.00011606837606837608,
+      "loss": 0.0513,
+      "step": 4910
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 0.00011589743589743591,
+      "loss": 0.243,
+      "step": 4920
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 0.00011572649572649574,
+      "loss": 0.0159,
+      "step": 4930
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 0.00011555555555555555,
+      "loss": 0.0781,
+      "step": 4940
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 0.00011538461538461538,
+      "loss": 0.0939,
+      "step": 4950
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 0.00011521367521367521,
+      "loss": 0.0423,
+      "step": 4960
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 0.00011504273504273504,
+      "loss": 0.0027,
+      "step": 4970
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 0.00011487179487179487,
+      "loss": 0.0013,
+      "step": 4980
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 0.0001147008547008547,
+      "loss": 0.0022,
+      "step": 4990
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 0.00011452991452991453,
+      "loss": 0.0226,
+      "step": 5000
+    },
+    {
+      "epoch": 6.41,
+      "eval_accuracy": 0.7716049382716049,
+      "eval_loss": 1.2063742876052856,
+      "eval_runtime": 38.8135,
+      "eval_samples_per_second": 45.963,
+      "eval_steps_per_second": 5.745,
+      "step": 5000
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 0.00011435897435897435,
+      "loss": 0.0954,
+      "step": 5010
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 0.00011418803418803418,
+      "loss": 0.175,
+      "step": 5020
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 0.00011401709401709401,
+      "loss": 0.1029,
+      "step": 5030
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 0.00011384615384615384,
+      "loss": 0.0504,
+      "step": 5040
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 0.00011367521367521367,
+      "loss": 0.0413,
+      "step": 5050
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 0.00011350427350427351,
+      "loss": 0.0692,
+      "step": 5060
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 0.00011333333333333334,
+      "loss": 0.0874,
+      "step": 5070
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 0.00011316239316239317,
+      "loss": 0.0858,
+      "step": 5080
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 0.000112991452991453,
+      "loss": 0.0158,
+      "step": 5090
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 0.00011282051282051283,
+      "loss": 0.0103,
+      "step": 5100
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 0.00011264957264957266,
+      "loss": 0.088,
+      "step": 5110
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 0.00011247863247863249,
+      "loss": 0.0014,
+      "step": 5120
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 0.00011230769230769231,
+      "loss": 0.0473,
+      "step": 5130
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 0.00011213675213675214,
+      "loss": 0.103,
+      "step": 5140
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 0.00011196581196581197,
+      "loss": 0.0271,
+      "step": 5150
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 0.0001117948717948718,
+      "loss": 0.1189,
+      "step": 5160
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 0.00011162393162393163,
+      "loss": 0.2419,
+      "step": 5170
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 0.00011145299145299146,
+      "loss": 0.0575,
+      "step": 5180
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 0.00011128205128205129,
+      "loss": 0.0331,
+      "step": 5190
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 0.00011111111111111112,
+      "loss": 0.1995,
+      "step": 5200
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 0.00011094017094017095,
+      "loss": 0.0383,
+      "step": 5210
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 0.00011076923076923077,
+      "loss": 0.0024,
+      "step": 5220
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 0.00011059829059829062,
+      "loss": 0.2491,
+      "step": 5230
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 0.00011042735042735045,
+      "loss": 0.0031,
+      "step": 5240
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 0.00011025641025641027,
+      "loss": 0.0082,
+      "step": 5250
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 0.0001100854700854701,
+      "loss": 0.0264,
+      "step": 5260
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 0.00010991452991452993,
+      "loss": 0.0792,
+      "step": 5270
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 0.00010974358974358976,
+      "loss": 0.1843,
+      "step": 5280
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 0.00010957264957264959,
+      "loss": 0.1755,
+      "step": 5290
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 0.00010940170940170942,
+      "loss": 0.2766,
+      "step": 5300
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 0.00010923076923076922,
+      "loss": 0.1326,
+      "step": 5310
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 0.00010905982905982905,
+      "loss": 0.005,
+      "step": 5320
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 0.00010888888888888889,
+      "loss": 0.0611,
+      "step": 5330
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 0.00010871794871794872,
+      "loss": 0.2345,
+      "step": 5340
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 0.00010854700854700855,
+      "loss": 0.3693,
+      "step": 5350
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 0.00010837606837606838,
+      "loss": 0.0654,
+      "step": 5360
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 0.0001082051282051282,
+      "loss": 0.0929,
+      "step": 5370
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 0.00010803418803418804,
+      "loss": 0.087,
+      "step": 5380
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 0.00010786324786324786,
+      "loss": 0.0359,
+      "step": 5390
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0001076923076923077,
+      "loss": 0.0754,
+      "step": 5400
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 0.00010752136752136752,
+      "loss": 0.0039,
+      "step": 5410
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 0.00010735042735042735,
+      "loss": 0.0733,
+      "step": 5420
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 0.00010717948717948718,
+      "loss": 0.0426,
+      "step": 5430
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 0.00010700854700854701,
+      "loss": 0.0943,
+      "step": 5440
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 0.00010683760683760684,
+      "loss": 0.163,
+      "step": 5450
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 0.00010666666666666667,
+      "loss": 0.0653,
+      "step": 5460
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 0.0001064957264957265,
+      "loss": 0.0054,
+      "step": 5470
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 0.00010632478632478632,
+      "loss": 0.0179,
+      "step": 5480
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 0.00010615384615384615,
+      "loss": 0.0124,
+      "step": 5490
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 0.000105982905982906,
+      "loss": 0.0439,
+      "step": 5500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 0.00010581196581196582,
+      "loss": 0.0015,
+      "step": 5510
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 0.00010564102564102565,
+      "loss": 0.0434,
+      "step": 5520
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 0.00010547008547008548,
+      "loss": 0.0041,
+      "step": 5530
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 0.00010529914529914531,
+      "loss": 0.0189,
+      "step": 5540
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 0.00010512820512820514,
+      "loss": 0.0511,
+      "step": 5550
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 0.00010495726495726497,
+      "loss": 0.0013,
+      "step": 5560
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 0.0001047863247863248,
+      "loss": 0.0756,
+      "step": 5570
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 0.00010461538461538463,
+      "loss": 0.018,
+      "step": 5580
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 0.00010444444444444445,
+      "loss": 0.0416,
+      "step": 5590
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 0.00010427350427350428,
+      "loss": 0.0718,
+      "step": 5600
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 0.00010410256410256411,
+      "loss": 0.0846,
+      "step": 5610
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 0.00010393162393162394,
+      "loss": 0.0753,
+      "step": 5620
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 0.00010376068376068377,
+      "loss": 0.0051,
+      "step": 5630
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 0.0001035897435897436,
+      "loss": 0.0789,
+      "step": 5640
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 0.00010341880341880343,
+      "loss": 0.089,
+      "step": 5650
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 0.00010324786324786326,
+      "loss": 0.0561,
+      "step": 5660
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 0.00010307692307692307,
+      "loss": 0.0184,
+      "step": 5670
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 0.0001029059829059829,
+      "loss": 0.0023,
+      "step": 5680
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 0.00010273504273504273,
+      "loss": 0.016,
+      "step": 5690
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 0.00010256410256410256,
+      "loss": 0.0604,
+      "step": 5700
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 0.00010239316239316239,
+      "loss": 0.138,
+      "step": 5710
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 0.00010222222222222222,
+      "loss": 0.0214,
+      "step": 5720
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 0.00010205128205128205,
+      "loss": 0.1264,
+      "step": 5730
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 0.00010188034188034187,
+      "loss": 0.0757,
+      "step": 5740
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 0.0001017094017094017,
+      "loss": 0.0039,
+      "step": 5750
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 0.00010153846153846153,
+      "loss": 0.0066,
+      "step": 5760
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 0.00010136752136752137,
+      "loss": 0.005,
+      "step": 5770
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 0.0001011965811965812,
+      "loss": 0.0012,
+      "step": 5780
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 0.00010102564102564103,
+      "loss": 0.0018,
+      "step": 5790
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 0.00010085470085470086,
+      "loss": 0.0032,
+      "step": 5800
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 0.00010068376068376069,
+      "loss": 0.0578,
+      "step": 5810
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 0.00010051282051282052,
+      "loss": 0.0013,
+      "step": 5820
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 0.00010034188034188035,
+      "loss": 0.0016,
+      "step": 5830
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 0.00010017094017094018,
+      "loss": 0.0051,
+      "step": 5840
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 0.0001,
+      "loss": 0.147,
+      "step": 5850
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 9.982905982905983e-05,
+      "loss": 0.0103,
+      "step": 5860
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 9.965811965811966e-05,
+      "loss": 0.0011,
+      "step": 5870
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 9.948717948717949e-05,
+      "loss": 0.0829,
+      "step": 5880
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 9.931623931623932e-05,
+      "loss": 0.0065,
+      "step": 5890
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 9.914529914529915e-05,
+      "loss": 0.0598,
+      "step": 5900
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 9.897435897435898e-05,
+      "loss": 0.0473,
+      "step": 5910
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 9.88034188034188e-05,
+      "loss": 0.0007,
+      "step": 5920
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 9.863247863247864e-05,
+      "loss": 0.012,
+      "step": 5930
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 9.846153846153848e-05,
+      "loss": 0.093,
+      "step": 5940
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 9.829059829059829e-05,
+      "loss": 0.0074,
+      "step": 5950
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 9.811965811965812e-05,
+      "loss": 0.0175,
+      "step": 5960
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 9.794871794871795e-05,
+      "loss": 0.0363,
+      "step": 5970
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 9.777777777777778e-05,
+      "loss": 0.0488,
+      "step": 5980
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 9.760683760683761e-05,
+      "loss": 0.0007,
+      "step": 5990
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 9.743589743589744e-05,
+      "loss": 0.0844,
+      "step": 6000
+    },
+    {
+      "epoch": 7.69,
+      "eval_accuracy": 0.7946127946127947,
+      "eval_loss": 1.2210915088653564,
+      "eval_runtime": 34.2874,
+      "eval_samples_per_second": 52.031,
+      "eval_steps_per_second": 6.504,
+      "step": 6000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 9.726495726495727e-05,
+      "loss": 0.1933,
+      "step": 6010
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 9.70940170940171e-05,
+      "loss": 0.0448,
+      "step": 6020
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.692307692307692e-05,
+      "loss": 0.2629,
+      "step": 6030
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.675213675213675e-05,
+      "loss": 0.0007,
+      "step": 6040
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.658119658119658e-05,
+      "loss": 0.0125,
+      "step": 6050
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.641025641025641e-05,
+      "loss": 0.0129,
+      "step": 6060
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.623931623931625e-05,
+      "loss": 0.0008,
+      "step": 6070
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.606837606837608e-05,
+      "loss": 0.0009,
+      "step": 6080
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.589743589743591e-05,
+      "loss": 0.0448,
+      "step": 6090
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.572649572649574e-05,
+      "loss": 0.002,
+      "step": 6100
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.555555555555557e-05,
+      "loss": 0.0514,
+      "step": 6110
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.53846153846154e-05,
+      "loss": 0.0057,
+      "step": 6120
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.521367521367521e-05,
+      "loss": 0.0145,
+      "step": 6130
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.504273504273504e-05,
+      "loss": 0.001,
+      "step": 6140
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.487179487179487e-05,
+      "loss": 0.0009,
+      "step": 6150
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.47008547008547e-05,
+      "loss": 0.0953,
+      "step": 6160
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.452991452991453e-05,
+      "loss": 0.0011,
+      "step": 6170
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.435897435897436e-05,
+      "loss": 0.0007,
+      "step": 6180
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.418803418803419e-05,
+      "loss": 0.0034,
+      "step": 6190
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.401709401709401e-05,
+      "loss": 0.0738,
+      "step": 6200
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.384615384615386e-05,
+      "loss": 0.0482,
+      "step": 6210
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.367521367521369e-05,
+      "loss": 0.0014,
+      "step": 6220
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 9.350427350427351e-05,
+      "loss": 0.0605,
+      "step": 6230
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 0.0622,
+      "step": 6240
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.316239316239317e-05,
+      "loss": 0.0623,
+      "step": 6250
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.2991452991453e-05,
+      "loss": 0.0084,
+      "step": 6260
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.282051282051283e-05,
+      "loss": 0.0061,
+      "step": 6270
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.264957264957266e-05,
+      "loss": 0.0011,
+      "step": 6280
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.247863247863249e-05,
+      "loss": 0.0051,
+      "step": 6290
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.230769230769232e-05,
+      "loss": 0.0658,
+      "step": 6300
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.213675213675213e-05,
+      "loss": 0.0007,
+      "step": 6310
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.196581196581196e-05,
+      "loss": 0.0954,
+      "step": 6320
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 9.179487179487179e-05,
+      "loss": 0.0012,
+      "step": 6330
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.162393162393162e-05,
+      "loss": 0.1042,
+      "step": 6340
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.145299145299146e-05,
+      "loss": 0.0006,
+      "step": 6350
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.128205128205129e-05,
+      "loss": 0.0329,
+      "step": 6360
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.111111111111112e-05,
+      "loss": 0.0551,
+      "step": 6370
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.094017094017095e-05,
+      "loss": 0.0011,
+      "step": 6380
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.076923076923078e-05,
+      "loss": 0.0012,
+      "step": 6390
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 9.05982905982906e-05,
+      "loss": 0.0817,
+      "step": 6400
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 9.042735042735043e-05,
+      "loss": 0.0014,
+      "step": 6410
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 9.025641025641026e-05,
+      "loss": 0.0292,
+      "step": 6420
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 9.008547008547009e-05,
+      "loss": 0.0098,
+      "step": 6430
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.991452991452992e-05,
+      "loss": 0.0011,
+      "step": 6440
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.974358974358975e-05,
+      "loss": 0.004,
+      "step": 6450
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.957264957264958e-05,
+      "loss": 0.0988,
+      "step": 6460
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.94017094017094e-05,
+      "loss": 0.0373,
+      "step": 6470
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.923076923076924e-05,
+      "loss": 0.0431,
+      "step": 6480
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.905982905982906e-05,
+      "loss": 0.0084,
+      "step": 6490
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 0.0023,
+      "step": 6500
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 8.871794871794872e-05,
+      "loss": 0.086,
+      "step": 6510
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.854700854700855e-05,
+      "loss": 0.0997,
+      "step": 6520
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.837606837606838e-05,
+      "loss": 0.0008,
+      "step": 6530
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.820512820512821e-05,
+      "loss": 0.0011,
+      "step": 6540
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.803418803418804e-05,
+      "loss": 0.0637,
+      "step": 6550
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 8.786324786324787e-05,
+      "loss": 0.0503,
+      "step": 6560
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.76923076923077e-05,
+      "loss": 0.0587,
+      "step": 6570
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 8.752136752136752e-05,
+      "loss": 0.0858,
+      "step": 6580
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.735042735042735e-05,
+      "loss": 0.0006,
+      "step": 6590
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 8.717948717948718e-05,
+      "loss": 0.0007,
+      "step": 6600
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.700854700854701e-05,
+      "loss": 0.0172,
+      "step": 6610
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.683760683760684e-05,
+      "loss": 0.0009,
+      "step": 6620
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 8.666666666666667e-05,
+      "loss": 0.0006,
+      "step": 6630
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 8.64957264957265e-05,
+      "loss": 0.184,
+      "step": 6640
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 8.632478632478634e-05,
+      "loss": 0.167,
+      "step": 6650
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.615384615384617e-05,
+      "loss": 0.0084,
+      "step": 6660
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 8.5982905982906e-05,
+      "loss": 0.0274,
+      "step": 6670
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 8.581196581196581e-05,
+      "loss": 0.0008,
+      "step": 6680
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 8.564102564102564e-05,
+      "loss": 0.0596,
+      "step": 6690
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 8.547008547008547e-05,
+      "loss": 0.0083,
+      "step": 6700
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 8.52991452991453e-05,
+      "loss": 0.0212,
+      "step": 6710
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 8.512820512820513e-05,
+      "loss": 0.0172,
+      "step": 6720
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 8.495726495726496e-05,
+      "loss": 0.0042,
+      "step": 6730
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 8.478632478632479e-05,
+      "loss": 0.1231,
+      "step": 6740
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 8.461538461538461e-05,
+      "loss": 0.0012,
+      "step": 6750
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 8.444444444444444e-05,
+      "loss": 0.0206,
+      "step": 6760
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 8.427350427350427e-05,
+      "loss": 0.0012,
+      "step": 6770
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 8.410256410256411e-05,
+      "loss": 0.0061,
+      "step": 6780
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 8.393162393162394e-05,
+      "loss": 0.1411,
+      "step": 6790
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.376068376068377e-05,
+      "loss": 0.0489,
+      "step": 6800
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 8.35897435897436e-05,
+      "loss": 0.0044,
+      "step": 6810
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 8.341880341880343e-05,
+      "loss": 0.0262,
+      "step": 6820
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 8.324786324786326e-05,
+      "loss": 0.0793,
+      "step": 6830
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 8.307692307692309e-05,
+      "loss": 0.0054,
+      "step": 6840
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 8.290598290598292e-05,
+      "loss": 0.1281,
+      "step": 6850
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 8.273504273504273e-05,
+      "loss": 0.0657,
+      "step": 6860
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 8.256410256410256e-05,
+      "loss": 0.016,
+      "step": 6870
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 8.239316239316239e-05,
+      "loss": 0.0133,
+      "step": 6880
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 8.222222222222222e-05,
+      "loss": 0.0391,
+      "step": 6890
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 8.205128205128205e-05,
+      "loss": 0.1083,
+      "step": 6900
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 8.188034188034188e-05,
+      "loss": 0.1051,
+      "step": 6910
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 8.170940170940172e-05,
+      "loss": 0.0368,
+      "step": 6920
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 8.153846153846155e-05,
+      "loss": 0.0038,
+      "step": 6930
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 8.136752136752138e-05,
+      "loss": 0.0129,
+      "step": 6940
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 8.11965811965812e-05,
+      "loss": 0.1158,
+      "step": 6950
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 8.102564102564103e-05,
+      "loss": 0.1172,
+      "step": 6960
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 8.085470085470086e-05,
+      "loss": 0.0009,
+      "step": 6970
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 8.068376068376069e-05,
+      "loss": 0.1745,
+      "step": 6980
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 8.051282051282052e-05,
+      "loss": 0.0434,
+      "step": 6990
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 8.034188034188035e-05,
+      "loss": 0.0006,
+      "step": 7000
+    },
+    {
+      "epoch": 8.97,
+      "eval_accuracy": 0.7884399551066218,
+      "eval_loss": 1.2867748737335205,
+      "eval_runtime": 36.1409,
+      "eval_samples_per_second": 49.362,
+      "eval_steps_per_second": 6.17,
+      "step": 7000
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 8.017094017094018e-05,
+      "loss": 0.0005,
+      "step": 7010
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 8e-05,
+      "loss": 0.0017,
+      "step": 7020
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 7.982905982905984e-05,
+      "loss": 0.0526,
+      "step": 7030
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 7.965811965811965e-05,
+      "loss": 0.0812,
+      "step": 7040
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 7.948717948717948e-05,
+      "loss": 0.0004,
+      "step": 7050
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 7.931623931623932e-05,
+      "loss": 0.205,
+      "step": 7060
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 7.914529914529915e-05,
+      "loss": 0.1026,
+      "step": 7070
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 7.897435897435898e-05,
+      "loss": 0.0089,
+      "step": 7080
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 7.880341880341881e-05,
+      "loss": 0.135,
+      "step": 7090
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 7.863247863247864e-05,
+      "loss": 0.0046,
+      "step": 7100
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 7.846153846153847e-05,
+      "loss": 0.1972,
+      "step": 7110
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 7.82905982905983e-05,
+      "loss": 0.0195,
+      "step": 7120
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 7.811965811965812e-05,
+      "loss": 0.0022,
+      "step": 7130
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 7.794871794871795e-05,
+      "loss": 0.0015,
+      "step": 7140
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 7.777777777777778e-05,
+      "loss": 0.0006,
+      "step": 7150
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 7.760683760683761e-05,
+      "loss": 0.116,
+      "step": 7160
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 7.743589743589744e-05,
+      "loss": 0.0126,
+      "step": 7170
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 7.726495726495727e-05,
+      "loss": 0.0509,
+      "step": 7180
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 7.70940170940171e-05,
+      "loss": 0.0017,
+      "step": 7190
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 7.692307692307693e-05,
+      "loss": 0.1155,
+      "step": 7200
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 7.675213675213675e-05,
+      "loss": 0.0006,
+      "step": 7210
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 7.65811965811966e-05,
+      "loss": 0.0807,
+      "step": 7220
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 7.641025641025641e-05,
+      "loss": 0.0763,
+      "step": 7230
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 7.623931623931624e-05,
+      "loss": 0.0361,
+      "step": 7240
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 7.606837606837607e-05,
+      "loss": 0.0531,
+      "step": 7250
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 7.58974358974359e-05,
+      "loss": 0.1016,
+      "step": 7260
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 7.572649572649573e-05,
+      "loss": 0.0008,
+      "step": 7270
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 7.555555555555556e-05,
+      "loss": 0.0055,
+      "step": 7280
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 7.538461538461539e-05,
+      "loss": 0.1142,
+      "step": 7290
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 7.521367521367521e-05,
+      "loss": 0.0007,
+      "step": 7300
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 7.504273504273504e-05,
+      "loss": 0.0008,
+      "step": 7310
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 7.487179487179487e-05,
+      "loss": 0.0006,
+      "step": 7320
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 7.47008547008547e-05,
+      "loss": 0.0006,
+      "step": 7330
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 7.452991452991453e-05,
+      "loss": 0.0048,
+      "step": 7340
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 7.435897435897436e-05,
+      "loss": 0.0716,
+      "step": 7350
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 7.41880341880342e-05,
+      "loss": 0.0031,
+      "step": 7360
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 7.401709401709403e-05,
+      "loss": 0.1143,
+      "step": 7370
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 7.384615384615386e-05,
+      "loss": 0.0634,
+      "step": 7380
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 7.367521367521369e-05,
+      "loss": 0.0005,
+      "step": 7390
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 7.350427350427352e-05,
+      "loss": 0.1023,
+      "step": 7400
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 0.0011,
+      "step": 7410
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 7.316239316239316e-05,
+      "loss": 0.0422,
+      "step": 7420
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 7.299145299145299e-05,
+      "loss": 0.0004,
+      "step": 7430
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 7.282051282051282e-05,
+      "loss": 0.1052,
+      "step": 7440
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 7.264957264957265e-05,
+      "loss": 0.0005,
+      "step": 7450
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 7.247863247863248e-05,
+      "loss": 0.0243,
+      "step": 7460
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 7.23076923076923e-05,
+      "loss": 0.0006,
+      "step": 7470
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 7.213675213675213e-05,
+      "loss": 0.1699,
+      "step": 7480
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 7.196581196581196e-05,
+      "loss": 0.0004,
+      "step": 7490
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 0.0036,
+      "step": 7500
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 7.162393162393163e-05,
+      "loss": 0.0157,
+      "step": 7510
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 7.145299145299146e-05,
+      "loss": 0.0008,
+      "step": 7520
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 7.128205128205129e-05,
+      "loss": 0.0644,
+      "step": 7530
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 7.111111111111112e-05,
+      "loss": 0.0851,
+      "step": 7540
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 7.094017094017095e-05,
+      "loss": 0.139,
+      "step": 7550
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 7.076923076923078e-05,
+      "loss": 0.0009,
+      "step": 7560
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 7.05982905982906e-05,
+      "loss": 0.063,
+      "step": 7570
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 7.042735042735044e-05,
+      "loss": 0.0027,
+      "step": 7580
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 7.025641025641025e-05,
+      "loss": 0.227,
+      "step": 7590
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 7.008547008547008e-05,
+      "loss": 0.0005,
+      "step": 7600
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 6.991452991452991e-05,
+      "loss": 0.1772,
+      "step": 7610
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 6.974358974358974e-05,
+      "loss": 0.0007,
+      "step": 7620
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 6.957264957264958e-05,
+      "loss": 0.0034,
+      "step": 7630
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 6.940170940170941e-05,
+      "loss": 0.027,
+      "step": 7640
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 6.923076923076924e-05,
+      "loss": 0.0009,
+      "step": 7650
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 6.905982905982907e-05,
+      "loss": 0.0663,
+      "step": 7660
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 6.88888888888889e-05,
+      "loss": 0.095,
+      "step": 7670
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 6.871794871794872e-05,
+      "loss": 0.0704,
+      "step": 7680
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.854700854700855e-05,
+      "loss": 0.001,
+      "step": 7690
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.837606837606838e-05,
+      "loss": 0.0015,
+      "step": 7700
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 6.820512820512821e-05,
+      "loss": 0.0022,
+      "step": 7710
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 6.803418803418804e-05,
+      "loss": 0.0419,
+      "step": 7720
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 6.786324786324787e-05,
+      "loss": 0.0022,
+      "step": 7730
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 6.76923076923077e-05,
+      "loss": 0.0007,
+      "step": 7740
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 6.752136752136753e-05,
+      "loss": 0.215,
+      "step": 7750
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 6.735042735042735e-05,
+      "loss": 0.0006,
+      "step": 7760
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 6.717948717948718e-05,
+      "loss": 0.0018,
+      "step": 7770
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 6.700854700854701e-05,
+      "loss": 0.0006,
+      "step": 7780
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 6.683760683760684e-05,
+      "loss": 0.0007,
+      "step": 7790
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.0023,
+      "step": 7800
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 6.64957264957265e-05,
+      "loss": 0.0004,
+      "step": 7810
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 6.632478632478633e-05,
+      "loss": 0.0606,
+      "step": 7820
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 6.615384615384616e-05,
+      "loss": 0.0005,
+      "step": 7830
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 6.598290598290599e-05,
+      "loss": 0.0006,
+      "step": 7840
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 6.581196581196581e-05,
+      "loss": 0.0335,
+      "step": 7850
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 6.564102564102564e-05,
+      "loss": 0.0004,
+      "step": 7860
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 6.547008547008547e-05,
+      "loss": 0.0004,
+      "step": 7870
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 6.52991452991453e-05,
+      "loss": 0.0037,
+      "step": 7880
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 6.512820512820513e-05,
+      "loss": 0.1206,
+      "step": 7890
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 6.495726495726496e-05,
+      "loss": 0.0153,
+      "step": 7900
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 6.478632478632479e-05,
+      "loss": 0.0006,
+      "step": 7910
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 6.461538461538462e-05,
+      "loss": 0.0083,
+      "step": 7920
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 6.444444444444446e-05,
+      "loss": 0.0032,
+      "step": 7930
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 6.427350427350429e-05,
+      "loss": 0.0004,
+      "step": 7940
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 6.410256410256412e-05,
+      "loss": 0.0084,
+      "step": 7950
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 6.393162393162393e-05,
+      "loss": 0.0353,
+      "step": 7960
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 6.376068376068376e-05,
+      "loss": 0.0019,
+      "step": 7970
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 6.358974358974359e-05,
+      "loss": 0.0004,
+      "step": 7980
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 6.341880341880342e-05,
+      "loss": 0.0117,
+      "step": 7990
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 6.324786324786325e-05,
+      "loss": 0.0914,
+      "step": 8000
+    },
+    {
+      "epoch": 10.26,
+      "eval_accuracy": 0.7940516273849607,
+      "eval_loss": 1.275172233581543,
+      "eval_runtime": 35.1577,
+      "eval_samples_per_second": 50.743,
+      "eval_steps_per_second": 6.343,
+      "step": 8000
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 6.307692307692308e-05,
+      "loss": 0.0618,
+      "step": 8010
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 6.29059829059829e-05,
+      "loss": 0.0707,
+      "step": 8020
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 6.273504273504273e-05,
+      "loss": 0.0011,
+      "step": 8030
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 6.256410256410256e-05,
+      "loss": 0.0968,
+      "step": 8040
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 6.239316239316239e-05,
+      "loss": 0.0003,
+      "step": 8050
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 6.222222222222222e-05,
+      "loss": 0.0003,
+      "step": 8060
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 6.205128205128206e-05,
+      "loss": 0.0041,
+      "step": 8070
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 6.188034188034189e-05,
+      "loss": 0.0003,
+      "step": 8080
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 6.170940170940172e-05,
+      "loss": 0.0024,
+      "step": 8090
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 6.153846153846155e-05,
+      "loss": 0.0006,
+      "step": 8100
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 6.136752136752138e-05,
+      "loss": 0.0006,
+      "step": 8110
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 6.11965811965812e-05,
+      "loss": 0.0005,
+      "step": 8120
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 6.1025641025641035e-05,
+      "loss": 0.0006,
+      "step": 8130
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 6.085470085470085e-05,
+      "loss": 0.091,
+      "step": 8140
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 6.068376068376068e-05,
+      "loss": 0.0003,
+      "step": 8150
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 6.0512820512820515e-05,
+      "loss": 0.0004,
+      "step": 8160
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 6.0341880341880344e-05,
+      "loss": 0.0175,
+      "step": 8170
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 6.017094017094017e-05,
+      "loss": 0.0003,
+      "step": 8180
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 6e-05,
+      "loss": 0.0003,
+      "step": 8190
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 5.982905982905983e-05,
+      "loss": 0.0003,
+      "step": 8200
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 5.965811965811966e-05,
+      "loss": 0.0005,
+      "step": 8210
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 5.948717948717949e-05,
+      "loss": 0.0008,
+      "step": 8220
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 5.931623931623932e-05,
+      "loss": 0.0005,
+      "step": 8230
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 5.9145299145299146e-05,
+      "loss": 0.0006,
+      "step": 8240
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 5.897435897435898e-05,
+      "loss": 0.0004,
+      "step": 8250
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 5.880341880341881e-05,
+      "loss": 0.0003,
+      "step": 8260
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 5.863247863247864e-05,
+      "loss": 0.174,
+      "step": 8270
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 5.846153846153847e-05,
+      "loss": 0.0006,
+      "step": 8280
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 5.82905982905983e-05,
+      "loss": 0.0003,
+      "step": 8290
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 5.8119658119658126e-05,
+      "loss": 0.0004,
+      "step": 8300
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 5.7948717948717954e-05,
+      "loss": 0.0003,
+      "step": 8310
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 5.7777777777777776e-05,
+      "loss": 0.0003,
+      "step": 8320
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 5.7606837606837605e-05,
+      "loss": 0.0004,
+      "step": 8330
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 5.7435897435897434e-05,
+      "loss": 0.0281,
+      "step": 8340
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 5.726495726495726e-05,
+      "loss": 0.0003,
+      "step": 8350
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 5.709401709401709e-05,
+      "loss": 0.0004,
+      "step": 8360
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 5.692307692307692e-05,
+      "loss": 0.0321,
+      "step": 8370
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 5.6752136752136756e-05,
+      "loss": 0.0013,
+      "step": 8380
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 5.6581196581196585e-05,
+      "loss": 0.0004,
+      "step": 8390
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 5.6410256410256414e-05,
+      "loss": 0.1839,
+      "step": 8400
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 5.623931623931624e-05,
+      "loss": 0.0004,
+      "step": 8410
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 5.606837606837607e-05,
+      "loss": 0.0019,
+      "step": 8420
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 5.58974358974359e-05,
+      "loss": 0.0005,
+      "step": 8430
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 5.572649572649573e-05,
+      "loss": 0.0444,
+      "step": 8440
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 5.555555555555556e-05,
+      "loss": 0.0005,
+      "step": 8450
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 5.538461538461539e-05,
+      "loss": 0.0009,
+      "step": 8460
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 5.521367521367522e-05,
+      "loss": 0.0004,
+      "step": 8470
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 5.504273504273505e-05,
+      "loss": 0.0004,
+      "step": 8480
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 5.487179487179488e-05,
+      "loss": 0.0005,
+      "step": 8490
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 5.470085470085471e-05,
+      "loss": 0.0003,
+      "step": 8500
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 5.4529914529914525e-05,
+      "loss": 0.0003,
+      "step": 8510
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 5.435897435897436e-05,
+      "loss": 0.0004,
+      "step": 8520
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 5.418803418803419e-05,
+      "loss": 0.0003,
+      "step": 8530
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 5.401709401709402e-05,
+      "loss": 0.0004,
+      "step": 8540
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 5.384615384615385e-05,
+      "loss": 0.0346,
+      "step": 8550
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 5.3675213675213675e-05,
+      "loss": 0.0035,
+      "step": 8560
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 5.3504273504273504e-05,
+      "loss": 0.0415,
+      "step": 8570
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 0.0003,
+      "step": 8580
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 5.316239316239316e-05,
+      "loss": 0.0003,
+      "step": 8590
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 5.2991452991453e-05,
+      "loss": 0.0013,
+      "step": 8600
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 5.2820512820512826e-05,
+      "loss": 0.0126,
+      "step": 8610
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 5.2649572649572655e-05,
+      "loss": 0.0003,
+      "step": 8620
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 5.2478632478632484e-05,
+      "loss": 0.0003,
+      "step": 8630
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 5.230769230769231e-05,
+      "loss": 0.0003,
+      "step": 8640
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 5.213675213675214e-05,
+      "loss": 0.0003,
+      "step": 8650
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 5.196581196581197e-05,
+      "loss": 0.0003,
+      "step": 8660
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 5.17948717948718e-05,
+      "loss": 0.0004,
+      "step": 8670
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 5.162393162393163e-05,
+      "loss": 0.0002,
+      "step": 8680
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 5.145299145299145e-05,
+      "loss": 0.0002,
+      "step": 8690
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 5.128205128205128e-05,
+      "loss": 0.0002,
+      "step": 8700
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 5.111111111111111e-05,
+      "loss": 0.0003,
+      "step": 8710
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 5.094017094017094e-05,
+      "loss": 0.0003,
+      "step": 8720
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 5.0769230769230766e-05,
+      "loss": 0.0004,
+      "step": 8730
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 5.05982905982906e-05,
+      "loss": 0.028,
+      "step": 8740
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 5.042735042735043e-05,
+      "loss": 0.0002,
+      "step": 8750
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 5.025641025641026e-05,
+      "loss": 0.0003,
+      "step": 8760
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 5.008547008547009e-05,
+      "loss": 0.0025,
+      "step": 8770
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.991452991452992e-05,
+      "loss": 0.0003,
+      "step": 8780
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.9743589743589746e-05,
+      "loss": 0.0002,
+      "step": 8790
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.9572649572649575e-05,
+      "loss": 0.0003,
+      "step": 8800
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.94017094017094e-05,
+      "loss": 0.0002,
+      "step": 8810
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.923076923076924e-05,
+      "loss": 0.0736,
+      "step": 8820
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.905982905982906e-05,
+      "loss": 0.0002,
+      "step": 8830
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.888888888888889e-05,
+      "loss": 0.0007,
+      "step": 8840
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.871794871794872e-05,
+      "loss": 0.015,
+      "step": 8850
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.854700854700855e-05,
+      "loss": 0.0067,
+      "step": 8860
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.8376068376068376e-05,
+      "loss": 0.0003,
+      "step": 8870
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.8205128205128205e-05,
+      "loss": 0.0002,
+      "step": 8880
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.803418803418804e-05,
+      "loss": 0.0002,
+      "step": 8890
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.786324786324787e-05,
+      "loss": 0.0002,
+      "step": 8900
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.76923076923077e-05,
+      "loss": 0.0002,
+      "step": 8910
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.752136752136752e-05,
+      "loss": 0.0004,
+      "step": 8920
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.735042735042735e-05,
+      "loss": 0.0002,
+      "step": 8930
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.717948717948718e-05,
+      "loss": 0.0003,
+      "step": 8940
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.700854700854701e-05,
+      "loss": 0.0002,
+      "step": 8950
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.683760683760684e-05,
+      "loss": 0.0002,
+      "step": 8960
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 0.0002,
+      "step": 8970
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.64957264957265e-05,
+      "loss": 0.0002,
+      "step": 8980
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.632478632478633e-05,
+      "loss": 0.0002,
+      "step": 8990
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.615384615384616e-05,
+      "loss": 0.0003,
+      "step": 9000
+    },
+    {
+      "epoch": 11.54,
+      "eval_accuracy": 0.7951739618406285,
+      "eval_loss": 1.3951935768127441,
+      "eval_runtime": 35.2372,
+      "eval_samples_per_second": 50.628,
+      "eval_steps_per_second": 6.329,
+      "step": 9000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.598290598290598e-05,
+      "loss": 0.0002,
+      "step": 9010
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.581196581196581e-05,
+      "loss": 0.0003,
+      "step": 9020
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.5641025641025645e-05,
+      "loss": 0.0002,
+      "step": 9030
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.5470085470085474e-05,
+      "loss": 0.0002,
+      "step": 9040
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.52991452991453e-05,
+      "loss": 0.0002,
+      "step": 9050
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.512820512820513e-05,
+      "loss": 0.0002,
+      "step": 9060
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.495726495726496e-05,
+      "loss": 0.0002,
+      "step": 9070
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.478632478632479e-05,
+      "loss": 0.0002,
+      "step": 9080
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.461538461538462e-05,
+      "loss": 0.0002,
+      "step": 9090
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.0002,
+      "step": 9100
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4273504273504275e-05,
+      "loss": 0.0014,
+      "step": 9110
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.4102564102564104e-05,
+      "loss": 0.0002,
+      "step": 9120
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.393162393162393e-05,
+      "loss": 0.0005,
+      "step": 9130
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.376068376068376e-05,
+      "loss": 0.0002,
+      "step": 9140
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.358974358974359e-05,
+      "loss": 0.0002,
+      "step": 9150
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.341880341880342e-05,
+      "loss": 0.0102,
+      "step": 9160
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.324786324786325e-05,
+      "loss": 0.0002,
+      "step": 9170
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.3076923076923084e-05,
+      "loss": 0.0002,
+      "step": 9180
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.2905982905982906e-05,
+      "loss": 0.0002,
+      "step": 9190
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.2735042735042735e-05,
+      "loss": 0.0006,
+      "step": 9200
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.2564102564102564e-05,
+      "loss": 0.0002,
+      "step": 9210
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.239316239316239e-05,
+      "loss": 0.0002,
+      "step": 9220
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.222222222222222e-05,
+      "loss": 0.0002,
+      "step": 9230
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.205128205128206e-05,
+      "loss": 0.0281,
+      "step": 9240
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.1880341880341886e-05,
+      "loss": 0.0002,
+      "step": 9250
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.1709401709401715e-05,
+      "loss": 0.0002,
+      "step": 9260
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.1538461538461544e-05,
+      "loss": 0.0035,
+      "step": 9270
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.1367521367521366e-05,
+      "loss": 0.0002,
+      "step": 9280
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.1196581196581195e-05,
+      "loss": 0.0003,
+      "step": 9290
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.1025641025641023e-05,
+      "loss": 0.0002,
+      "step": 9300
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.085470085470086e-05,
+      "loss": 0.0008,
+      "step": 9310
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.068376068376069e-05,
+      "loss": 0.0002,
+      "step": 9320
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.051282051282052e-05,
+      "loss": 0.0002,
+      "step": 9330
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.0341880341880346e-05,
+      "loss": 0.0002,
+      "step": 9340
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.0170940170940174e-05,
+      "loss": 0.0002,
+      "step": 9350
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4e-05,
+      "loss": 0.0003,
+      "step": 9360
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.9829059829059825e-05,
+      "loss": 0.0002,
+      "step": 9370
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 3.965811965811966e-05,
+      "loss": 0.0002,
+      "step": 9380
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 3.948717948717949e-05,
+      "loss": 0.0002,
+      "step": 9390
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 3.931623931623932e-05,
+      "loss": 0.0002,
+      "step": 9400
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 3.914529914529915e-05,
+      "loss": 0.0002,
+      "step": 9410
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 3.8974358974358976e-05,
+      "loss": 0.0002,
+      "step": 9420
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 3.8803418803418805e-05,
+      "loss": 0.0002,
+      "step": 9430
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 3.8632478632478634e-05,
+      "loss": 0.0002,
+      "step": 9440
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 3.846153846153846e-05,
+      "loss": 0.0002,
+      "step": 9450
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 3.82905982905983e-05,
+      "loss": 0.0002,
+      "step": 9460
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 3.811965811965812e-05,
+      "loss": 0.0002,
+      "step": 9470
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 3.794871794871795e-05,
+      "loss": 0.0002,
+      "step": 9480
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 3.777777777777778e-05,
+      "loss": 0.0002,
+      "step": 9490
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 3.760683760683761e-05,
+      "loss": 0.0002,
+      "step": 9500
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 3.7435897435897436e-05,
+      "loss": 0.0002,
+      "step": 9510
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 3.7264957264957265e-05,
+      "loss": 0.0003,
+      "step": 9520
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 3.70940170940171e-05,
+      "loss": 0.0033,
+      "step": 9530
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 3.692307692307693e-05,
+      "loss": 0.0002,
+      "step": 9540
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 3.675213675213676e-05,
+      "loss": 0.0002,
+      "step": 9550
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 3.658119658119658e-05,
+      "loss": 0.0002,
+      "step": 9560
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 3.641025641025641e-05,
+      "loss": 0.0002,
+      "step": 9570
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 3.623931623931624e-05,
+      "loss": 0.0002,
+      "step": 9580
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 3.606837606837607e-05,
+      "loss": 0.0002,
+      "step": 9590
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 3.58974358974359e-05,
+      "loss": 0.0002,
+      "step": 9600
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 3.572649572649573e-05,
+      "loss": 0.0002,
+      "step": 9610
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 3.555555555555556e-05,
+      "loss": 0.0169,
+      "step": 9620
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 3.538461538461539e-05,
+      "loss": 0.0001,
+      "step": 9630
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 3.521367521367522e-05,
+      "loss": 0.0002,
+      "step": 9640
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 3.504273504273504e-05,
+      "loss": 0.0002,
+      "step": 9650
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 3.487179487179487e-05,
+      "loss": 0.0002,
+      "step": 9660
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 3.4700854700854704e-05,
+      "loss": 0.0002,
+      "step": 9670
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 3.452991452991453e-05,
+      "loss": 0.0002,
+      "step": 9680
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 3.435897435897436e-05,
+      "loss": 0.0002,
+      "step": 9690
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 3.418803418803419e-05,
+      "loss": 0.0002,
+      "step": 9700
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 3.401709401709402e-05,
+      "loss": 0.0004,
+      "step": 9710
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 3.384615384615385e-05,
+      "loss": 0.0002,
+      "step": 9720
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 3.367521367521368e-05,
+      "loss": 0.0002,
+      "step": 9730
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 3.3504273504273506e-05,
+      "loss": 0.0002,
+      "step": 9740
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.0002,
+      "step": 9750
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 3.3162393162393164e-05,
+      "loss": 0.0001,
+      "step": 9760
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 3.299145299145299e-05,
+      "loss": 0.0002,
+      "step": 9770
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 3.282051282051282e-05,
+      "loss": 0.0001,
+      "step": 9780
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 3.264957264957265e-05,
+      "loss": 0.0002,
+      "step": 9790
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 3.247863247863248e-05,
+      "loss": 0.0001,
+      "step": 9800
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 3.230769230769231e-05,
+      "loss": 0.0001,
+      "step": 9810
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 3.2136752136752144e-05,
+      "loss": 0.0002,
+      "step": 9820
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 3.1965811965811966e-05,
+      "loss": 0.0001,
+      "step": 9830
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 3.1794871794871795e-05,
+      "loss": 0.0002,
+      "step": 9840
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 3.162393162393162e-05,
+      "loss": 0.0002,
+      "step": 9850
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 3.145299145299145e-05,
+      "loss": 0.0002,
+      "step": 9860
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 3.128205128205128e-05,
+      "loss": 0.0001,
+      "step": 9870
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 3.111111111111111e-05,
+      "loss": 0.0001,
+      "step": 9880
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 3.0940170940170946e-05,
+      "loss": 0.0001,
+      "step": 9890
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 3.0769230769230774e-05,
+      "loss": 0.0001,
+      "step": 9900
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 3.05982905982906e-05,
+      "loss": 0.0001,
+      "step": 9910
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 3.0427350427350425e-05,
+      "loss": 0.0001,
+      "step": 9920
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 3.0256410256410257e-05,
+      "loss": 0.0001,
+      "step": 9930
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 3.0085470085470086e-05,
+      "loss": 0.0002,
+      "step": 9940
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.9914529914529915e-05,
+      "loss": 0.0001,
+      "step": 9950
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.9743589743589744e-05,
+      "loss": 0.0001,
+      "step": 9960
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.9572649572649573e-05,
+      "loss": 0.0001,
+      "step": 9970
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.9401709401709405e-05,
+      "loss": 0.0001,
+      "step": 9980
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.9230769230769234e-05,
+      "loss": 0.0001,
+      "step": 9990
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.9059829059829063e-05,
+      "loss": 0.0001,
+      "step": 10000
+    },
+    {
+      "epoch": 12.82,
+      "eval_accuracy": 0.7912457912457912,
+      "eval_loss": 1.479121208190918,
+      "eval_runtime": 38.2692,
+      "eval_samples_per_second": 46.617,
+      "eval_steps_per_second": 5.827,
+      "step": 10000
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8888888888888888e-05,
+      "loss": 0.0001,
+      "step": 10010
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8717948717948717e-05,
+      "loss": 0.0001,
+      "step": 10020
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8547008547008546e-05,
+      "loss": 0.0001,
+      "step": 10030
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8376068376068378e-05,
+      "loss": 0.0001,
+      "step": 10040
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8205128205128207e-05,
+      "loss": 0.0002,
+      "step": 10050
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8034188034188036e-05,
+      "loss": 0.0002,
+      "step": 10060
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.7863247863247865e-05,
+      "loss": 0.0001,
+      "step": 10070
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.7692307692307694e-05,
+      "loss": 0.0001,
+      "step": 10080
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.7521367521367526e-05,
+      "loss": 0.0001,
+      "step": 10090
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.7350427350427355e-05,
+      "loss": 0.0001,
+      "step": 10100
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.717948717948718e-05,
+      "loss": 0.0001,
+      "step": 10110
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.700854700854701e-05,
+      "loss": 0.0002,
+      "step": 10120
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.6837606837606838e-05,
+      "loss": 0.0001,
+      "step": 10130
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.0001,
+      "step": 10140
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.64957264957265e-05,
+      "loss": 0.0001,
+      "step": 10150
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.6324786324786328e-05,
+      "loss": 0.0001,
+      "step": 10160
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.6153846153846157e-05,
+      "loss": 0.0001,
+      "step": 10170
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.5982905982905985e-05,
+      "loss": 0.0001,
+      "step": 10180
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.5811965811965814e-05,
+      "loss": 0.0001,
+      "step": 10190
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.564102564102564e-05,
+      "loss": 0.0001,
+      "step": 10200
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.547008547008547e-05,
+      "loss": 0.0001,
+      "step": 10210
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.52991452991453e-05,
+      "loss": 0.0001,
+      "step": 10220
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.512820512820513e-05,
+      "loss": 0.0001,
+      "step": 10230
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.495726495726496e-05,
+      "loss": 0.0001,
+      "step": 10240
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.4786324786324787e-05,
+      "loss": 0.0001,
+      "step": 10250
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.461538461538462e-05,
+      "loss": 0.0001,
+      "step": 10260
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.4444444444444445e-05,
+      "loss": 0.0001,
+      "step": 10270
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.4273504273504274e-05,
+      "loss": 0.0001,
+      "step": 10280
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.4102564102564103e-05,
+      "loss": 0.0001,
+      "step": 10290
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.3931623931623935e-05,
+      "loss": 0.0001,
+      "step": 10300
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.376068376068376e-05,
+      "loss": 0.0001,
+      "step": 10310
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.358974358974359e-05,
+      "loss": 0.0001,
+      "step": 10320
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.341880341880342e-05,
+      "loss": 0.0001,
+      "step": 10330
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.324786324786325e-05,
+      "loss": 0.0001,
+      "step": 10340
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.307692307692308e-05,
+      "loss": 0.0001,
+      "step": 10350
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.2905982905982905e-05,
+      "loss": 0.0001,
+      "step": 10360
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.2735042735042737e-05,
+      "loss": 0.0001,
+      "step": 10370
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.2564102564102566e-05,
+      "loss": 0.0001,
+      "step": 10380
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.2393162393162394e-05,
+      "loss": 0.0001,
+      "step": 10390
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.0001,
+      "step": 10400
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.2051282051282052e-05,
+      "loss": 0.0001,
+      "step": 10410
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.188034188034188e-05,
+      "loss": 0.0001,
+      "step": 10420
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.170940170940171e-05,
+      "loss": 0.0001,
+      "step": 10430
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.1538461538461542e-05,
+      "loss": 0.0001,
+      "step": 10440
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.1367521367521368e-05,
+      "loss": 0.0001,
+      "step": 10450
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.1196581196581196e-05,
+      "loss": 0.0001,
+      "step": 10460
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.102564102564103e-05,
+      "loss": 0.0001,
+      "step": 10470
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.0854700854700857e-05,
+      "loss": 0.0001,
+      "step": 10480
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.0683760683760683e-05,
+      "loss": 0.0001,
+      "step": 10490
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.0512820512820512e-05,
+      "loss": 0.0001,
+      "step": 10500
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.0341880341880344e-05,
+      "loss": 0.0001,
+      "step": 10510
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.0170940170940173e-05,
+      "loss": 0.0001,
+      "step": 10520
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2e-05,
+      "loss": 0.0001,
+      "step": 10530
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.982905982905983e-05,
+      "loss": 0.0001,
+      "step": 10540
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.965811965811966e-05,
+      "loss": 0.0001,
+      "step": 10550
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 0.0001,
+      "step": 10560
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 1.9316239316239317e-05,
+      "loss": 0.0001,
+      "step": 10570
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.914529914529915e-05,
+      "loss": 0.0001,
+      "step": 10580
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 0.0001,
+      "step": 10590
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 1.8803418803418804e-05,
+      "loss": 0.0001,
+      "step": 10600
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.8632478632478632e-05,
+      "loss": 0.0001,
+      "step": 10610
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 0.0001,
+      "step": 10620
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 1.829059829059829e-05,
+      "loss": 0.0001,
+      "step": 10630
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 1.811965811965812e-05,
+      "loss": 0.0001,
+      "step": 10640
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 0.0001,
+      "step": 10650
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.777777777777778e-05,
+      "loss": 0.0001,
+      "step": 10660
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 1.760683760683761e-05,
+      "loss": 0.0043,
+      "step": 10670
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.7435897435897434e-05,
+      "loss": 0.0001,
+      "step": 10680
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.7264957264957267e-05,
+      "loss": 0.0001,
+      "step": 10690
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 1.7094017094017095e-05,
+      "loss": 0.0001,
+      "step": 10700
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 0.0003,
+      "step": 10710
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.6752136752136753e-05,
+      "loss": 0.0001,
+      "step": 10720
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.6581196581196582e-05,
+      "loss": 0.0001,
+      "step": 10730
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 0.0002,
+      "step": 10740
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.623931623931624e-05,
+      "loss": 0.0001,
+      "step": 10750
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 1.6068376068376072e-05,
+      "loss": 0.0001,
+      "step": 10760
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 0.0001,
+      "step": 10770
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 1.5726495726495726e-05,
+      "loss": 0.0001,
+      "step": 10780
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.5555555555555555e-05,
+      "loss": 0.0001,
+      "step": 10790
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 0.0001,
+      "step": 10800
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 1.5213675213675213e-05,
+      "loss": 0.0001,
+      "step": 10810
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.5042735042735043e-05,
+      "loss": 0.0001,
+      "step": 10820
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 1.4871794871794872e-05,
+      "loss": 0.0001,
+      "step": 10830
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.4700854700854703e-05,
+      "loss": 0.0001,
+      "step": 10840
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 1.4529914529914531e-05,
+      "loss": 0.0001,
+      "step": 10850
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.4358974358974359e-05,
+      "loss": 0.0001,
+      "step": 10860
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.4188034188034189e-05,
+      "loss": 0.0001,
+      "step": 10870
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 1.4017094017094018e-05,
+      "loss": 0.0001,
+      "step": 10880
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 0.0001,
+      "step": 10890
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 1.3675213675213677e-05,
+      "loss": 0.0001,
+      "step": 10900
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.3504273504273504e-05,
+      "loss": 0.0001,
+      "step": 10910
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0001,
+      "step": 10920
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.3162393162393164e-05,
+      "loss": 0.0001,
+      "step": 10930
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.2991452991452993e-05,
+      "loss": 0.0001,
+      "step": 10940
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 1.282051282051282e-05,
+      "loss": 0.0001,
+      "step": 10950
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.264957264957265e-05,
+      "loss": 0.0001,
+      "step": 10960
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 1.247863247863248e-05,
+      "loss": 0.0001,
+      "step": 10970
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 0.0001,
+      "step": 10980
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 1.2136752136752137e-05,
+      "loss": 0.0001,
+      "step": 10990
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.1965811965811967e-05,
+      "loss": 0.0001,
+      "step": 11000
+    },
+    {
+      "epoch": 14.1,
+      "eval_accuracy": 0.7962962962962963,
+      "eval_loss": 1.4692177772521973,
+      "eval_runtime": 34.3664,
+      "eval_samples_per_second": 51.911,
+      "eval_steps_per_second": 6.489,
+      "step": 11000
+    }
+  ],
+  "max_steps": 11700,
+  "num_train_epochs": 15,
+  "total_flos": 6.815260936120025e+18,
+  "trial_name": null,
+  "trial_params": null
+}