diff --git "a/checkpoint-1950/trainer_state.json" "b/checkpoint-1950/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-1950/trainer_state.json"
@@ -0,0 +1,11836 @@
+{
+  "best_metric": 0.13513699173927307,
+  "best_model_checkpoint": "text-emotion-classifier-distilbert/checkpoint-1950",
+  "epoch": 13.0,
+  "eval_steps": 1,
+  "global_step": 1950,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-08,
+      "loss": 1.0825,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8e-08,
+      "loss": 1.1225,
+      "step": 2
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-07,
+      "loss": 1.0905,
+      "step": 3
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6e-07,
+      "loss": 1.1366,
+      "step": 4
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-07,
+      "loss": 1.1014,
+      "step": 5
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4e-07,
+      "loss": 1.0931,
+      "step": 6
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8e-07,
+      "loss": 1.0652,
+      "step": 7
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.2e-07,
+      "loss": 1.1165,
+      "step": 8
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6e-07,
+      "loss": 1.1065,
+      "step": 9
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4e-07,
+      "loss": 1.1256,
+      "step": 10
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.3999999999999997e-07,
+      "loss": 1.1515,
+      "step": 11
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8e-07,
+      "loss": 1.103,
+      "step": 12
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.2e-07,
+      "loss": 1.0769,
+      "step": 13
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.6e-07,
+      "loss": 1.101,
+      "step": 14
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6e-07,
+      "loss": 1.1066,
+      "step": 15
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.4e-07,
+      "loss": 1.1365,
+      "step": 16
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 1.063,
+      "step": 17
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.2e-07,
+      "loss": 1.1014,
+      "step": 18
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 7.599999999999999e-07,
+      "loss": 1.1079,
+      "step": 19
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8e-07,
+      "loss": 1.0915,
+      "step": 20
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.399999999999999e-07,
+      "loss": 1.1342,
+      "step": 21
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 8.799999999999999e-07,
+      "loss": 1.1188,
+      "step": 22
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.2e-07,
+      "loss": 1.0889,
+      "step": 23
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.6e-07,
+      "loss": 1.0921,
+      "step": 24
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1e-06,
+      "loss": 1.1061,
+      "step": 25
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.04e-06,
+      "loss": 1.1493,
+      "step": 26
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.08e-06,
+      "loss": 1.0383,
+      "step": 27
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.12e-06,
+      "loss": 1.1648,
+      "step": 28
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.16e-06,
+      "loss": 1.131,
+      "step": 29
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.2e-06,
+      "loss": 1.0875,
+      "step": 30
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.24e-06,
+      "loss": 1.1012,
+      "step": 31
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.28e-06,
+      "loss": 1.1321,
+      "step": 32
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.32e-06,
+      "loss": 1.0796,
+      "step": 33
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 1.0892,
+      "step": 34
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.4e-06,
+      "loss": 1.1048,
+      "step": 35
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.44e-06,
+      "loss": 1.0683,
+      "step": 36
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.48e-06,
+      "loss": 1.09,
+      "step": 37
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5199999999999998e-06,
+      "loss": 1.1109,
+      "step": 38
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 1.1116,
+      "step": 39
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6e-06,
+      "loss": 1.1195,
+      "step": 40
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.6399999999999998e-06,
+      "loss": 1.0677,
+      "step": 41
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799999999999998e-06,
+      "loss": 1.1448,
+      "step": 42
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7199999999999998e-06,
+      "loss": 1.0935,
+      "step": 43
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.7599999999999999e-06,
+      "loss": 1.077,
+      "step": 44
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8e-06,
+      "loss": 1.1167,
+      "step": 45
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.84e-06,
+      "loss": 1.0701,
+      "step": 46
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8799999999999998e-06,
+      "loss": 1.1005,
+      "step": 47
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.92e-06,
+      "loss": 1.0671,
+      "step": 48
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96e-06,
+      "loss": 1.0934,
+      "step": 49
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2e-06,
+      "loss": 1.1116,
+      "step": 50
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.999322033898305e-06,
+      "loss": 1.1154,
+      "step": 51
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.99864406779661e-06,
+      "loss": 1.0956,
+      "step": 52
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.997966101694915e-06,
+      "loss": 1.0935,
+      "step": 53
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.99728813559322e-06,
+      "loss": 1.0894,
+      "step": 54
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.996610169491525e-06,
+      "loss": 1.0766,
+      "step": 55
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9959322033898307e-06,
+      "loss": 1.1362,
+      "step": 56
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9952542372881357e-06,
+      "loss": 1.0796,
+      "step": 57
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9945762711864408e-06,
+      "loss": 1.1127,
+      "step": 58
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.993898305084746e-06,
+      "loss": 1.1046,
+      "step": 59
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.993220338983051e-06,
+      "loss": 1.1021,
+      "step": 60
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.992542372881356e-06,
+      "loss": 1.12,
+      "step": 61
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.991864406779661e-06,
+      "loss": 1.0773,
+      "step": 62
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.991186440677966e-06,
+      "loss": 1.0872,
+      "step": 63
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.990508474576271e-06,
+      "loss": 1.0911,
+      "step": 64
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.989830508474576e-06,
+      "loss": 1.1071,
+      "step": 65
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9891525423728812e-06,
+      "loss": 1.1356,
+      "step": 66
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9884745762711863e-06,
+      "loss": 1.1178,
+      "step": 67
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9877966101694913e-06,
+      "loss": 1.1216,
+      "step": 68
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9871186440677964e-06,
+      "loss": 1.1022,
+      "step": 69
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9864406779661014e-06,
+      "loss": 1.0726,
+      "step": 70
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9857627118644065e-06,
+      "loss": 1.1405,
+      "step": 71
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.985084745762712e-06,
+      "loss": 1.0563,
+      "step": 72
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.984406779661017e-06,
+      "loss": 1.1061,
+      "step": 73
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.983728813559322e-06,
+      "loss": 1.1059,
+      "step": 74
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.983050847457627e-06,
+      "loss": 1.1053,
+      "step": 75
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.982372881355932e-06,
+      "loss": 1.076,
+      "step": 76
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9816949152542373e-06,
+      "loss": 1.0533,
+      "step": 77
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9810169491525423e-06,
+      "loss": 1.0698,
+      "step": 78
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9803389830508474e-06,
+      "loss": 1.0742,
+      "step": 79
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9796610169491524e-06,
+      "loss": 1.0577,
+      "step": 80
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9789830508474575e-06,
+      "loss": 1.1312,
+      "step": 81
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9783050847457625e-06,
+      "loss": 1.0969,
+      "step": 82
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9776271186440676e-06,
+      "loss": 1.108,
+      "step": 83
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9769491525423726e-06,
+      "loss": 1.0715,
+      "step": 84
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9762711864406777e-06,
+      "loss": 1.1095,
+      "step": 85
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9755932203389827e-06,
+      "loss": 1.1169,
+      "step": 86
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.974915254237288e-06,
+      "loss": 1.1105,
+      "step": 87
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9742372881355933e-06,
+      "loss": 1.1455,
+      "step": 88
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9735593220338983e-06,
+      "loss": 1.1055,
+      "step": 89
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9728813559322034e-06,
+      "loss": 1.1118,
+      "step": 90
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9722033898305084e-06,
+      "loss": 1.0656,
+      "step": 91
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9715254237288135e-06,
+      "loss": 1.1251,
+      "step": 92
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9708474576271186e-06,
+      "loss": 1.0446,
+      "step": 93
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9701694915254236e-06,
+      "loss": 1.0961,
+      "step": 94
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9694915254237287e-06,
+      "loss": 1.1304,
+      "step": 95
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9688135593220337e-06,
+      "loss": 1.0913,
+      "step": 96
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9681355932203388e-06,
+      "loss": 1.0938,
+      "step": 97
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.967457627118644e-06,
+      "loss": 1.0958,
+      "step": 98
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9667796610169493e-06,
+      "loss": 1.0847,
+      "step": 99
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9661016949152544e-06,
+      "loss": 1.0685,
+      "step": 100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9654237288135594e-06,
+      "loss": 1.0825,
+      "step": 101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.964745762711864e-06,
+      "loss": 1.0761,
+      "step": 102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.964067796610169e-06,
+      "loss": 1.0941,
+      "step": 103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9633898305084746e-06,
+      "loss": 1.1172,
+      "step": 104
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9627118644067796e-06,
+      "loss": 1.0578,
+      "step": 105
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9620338983050847e-06,
+      "loss": 1.1347,
+      "step": 106
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9613559322033898e-06,
+      "loss": 1.0605,
+      "step": 107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.960677966101695e-06,
+      "loss": 1.0932,
+      "step": 108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.96e-06,
+      "loss": 1.071,
+      "step": 109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.959322033898305e-06,
+      "loss": 1.0858,
+      "step": 110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.95864406779661e-06,
+      "loss": 1.0561,
+      "step": 111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.957966101694915e-06,
+      "loss": 1.0732,
+      "step": 112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.95728813559322e-06,
+      "loss": 1.0731,
+      "step": 113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.956610169491525e-06,
+      "loss": 1.085,
+      "step": 114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9559322033898306e-06,
+      "loss": 1.0778,
+      "step": 115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9552542372881357e-06,
+      "loss": 1.0489,
+      "step": 116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9545762711864407e-06,
+      "loss": 1.0719,
+      "step": 117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.953898305084746e-06,
+      "loss": 1.0752,
+      "step": 118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.953220338983051e-06,
+      "loss": 1.0763,
+      "step": 119
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952542372881356e-06,
+      "loss": 1.0847,
+      "step": 120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.951864406779661e-06,
+      "loss": 1.0884,
+      "step": 121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.951186440677966e-06,
+      "loss": 1.0877,
+      "step": 122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.950508474576271e-06,
+      "loss": 1.074,
+      "step": 123
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.949830508474576e-06,
+      "loss": 1.0542,
+      "step": 124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.949152542372881e-06,
+      "loss": 1.0796,
+      "step": 125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9484745762711862e-06,
+      "loss": 1.0593,
+      "step": 126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9477966101694913e-06,
+      "loss": 1.0454,
+      "step": 127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9471186440677963e-06,
+      "loss": 1.117,
+      "step": 128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9464406779661014e-06,
+      "loss": 1.0852,
+      "step": 129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9457627118644065e-06,
+      "loss": 1.059,
+      "step": 130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.945084745762712e-06,
+      "loss": 1.0474,
+      "step": 131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.944406779661017e-06,
+      "loss": 1.0996,
+      "step": 132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.943728813559322e-06,
+      "loss": 1.0735,
+      "step": 133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.943050847457627e-06,
+      "loss": 1.0683,
+      "step": 134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.942372881355932e-06,
+      "loss": 1.063,
+      "step": 135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.941694915254237e-06,
+      "loss": 1.063,
+      "step": 136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9410169491525423e-06,
+      "loss": 1.0754,
+      "step": 137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9403389830508473e-06,
+      "loss": 1.0264,
+      "step": 138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9396610169491524e-06,
+      "loss": 1.045,
+      "step": 139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9389830508474574e-06,
+      "loss": 1.0713,
+      "step": 140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.938305084745763e-06,
+      "loss": 1.0788,
+      "step": 141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.937627118644068e-06,
+      "loss": 1.0502,
+      "step": 142
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.936949152542373e-06,
+      "loss": 1.0731,
+      "step": 143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9362711864406777e-06,
+      "loss": 1.1199,
+      "step": 144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9355932203389827e-06,
+      "loss": 1.0752,
+      "step": 145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9349152542372878e-06,
+      "loss": 1.0377,
+      "step": 146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9342372881355932e-06,
+      "loss": 1.0764,
+      "step": 147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9335593220338983e-06,
+      "loss": 1.0315,
+      "step": 148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9328813559322034e-06,
+      "loss": 1.076,
+      "step": 149
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9322033898305084e-06,
+      "loss": 1.0424,
+      "step": 150
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.485,
+      "eval_loss": 1.060792088508606,
+      "eval_runtime": 2.1186,
+      "eval_samples_per_second": 566.416,
+      "eval_steps_per_second": 4.72,
+      "step": 150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9315254237288135e-06,
+      "loss": 1.1058,
+      "step": 151
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9308474576271185e-06,
+      "loss": 1.0661,
+      "step": 152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9301694915254236e-06,
+      "loss": 1.0798,
+      "step": 153
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9294915254237286e-06,
+      "loss": 1.0944,
+      "step": 154
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9288135593220337e-06,
+      "loss": 1.0704,
+      "step": 155
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9281355932203387e-06,
+      "loss": 1.0725,
+      "step": 156
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9274576271186442e-06,
+      "loss": 0.9912,
+      "step": 157
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9267796610169493e-06,
+      "loss": 1.032,
+      "step": 158
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9261016949152543e-06,
+      "loss": 1.0375,
+      "step": 159
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9254237288135594e-06,
+      "loss": 1.072,
+      "step": 160
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9247457627118644e-06,
+      "loss": 1.0051,
+      "step": 161
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9240677966101695e-06,
+      "loss": 1.0478,
+      "step": 162
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9233898305084746e-06,
+      "loss": 1.0166,
+      "step": 163
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9227118644067796e-06,
+      "loss": 1.077,
+      "step": 164
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9220338983050847e-06,
+      "loss": 1.0317,
+      "step": 165
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9213559322033897e-06,
+      "loss": 1.0481,
+      "step": 166
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9206779661016948e-06,
+      "loss": 1.0667,
+      "step": 167
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.92e-06,
+      "loss": 1.0739,
+      "step": 168
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.919322033898305e-06,
+      "loss": 1.05,
+      "step": 169
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.91864406779661e-06,
+      "loss": 1.0537,
+      "step": 170
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.917966101694915e-06,
+      "loss": 1.0969,
+      "step": 171
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.91728813559322e-06,
+      "loss": 1.0318,
+      "step": 172
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9166101694915255e-06,
+      "loss": 1.056,
+      "step": 173
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9159322033898306e-06,
+      "loss": 1.0636,
+      "step": 174
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9152542372881356e-06,
+      "loss": 1.0538,
+      "step": 175
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9145762711864407e-06,
+      "loss": 1.0352,
+      "step": 176
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9138983050847458e-06,
+      "loss": 1.0684,
+      "step": 177
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.913220338983051e-06,
+      "loss": 1.0111,
+      "step": 178
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.912542372881356e-06,
+      "loss": 1.0422,
+      "step": 179
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.911864406779661e-06,
+      "loss": 0.9925,
+      "step": 180
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.911186440677966e-06,
+      "loss": 0.9835,
+      "step": 181
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.910508474576271e-06,
+      "loss": 1.0539,
+      "step": 182
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.909830508474576e-06,
+      "loss": 1.0679,
+      "step": 183
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9091525423728816e-06,
+      "loss": 1.0667,
+      "step": 184
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9084745762711866e-06,
+      "loss": 1.089,
+      "step": 185
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.9077966101694912e-06,
+      "loss": 1.0636,
+      "step": 186
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9071186440677965e-06,
+      "loss": 1.0225,
+      "step": 187
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9064406779661016e-06,
+      "loss": 1.1044,
+      "step": 188
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9057627118644066e-06,
+      "loss": 1.0323,
+      "step": 189
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9050847457627117e-06,
+      "loss": 1.066,
+      "step": 190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9044067796610167e-06,
+      "loss": 1.072,
+      "step": 191
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.903728813559322e-06,
+      "loss": 0.9809,
+      "step": 192
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.903050847457627e-06,
+      "loss": 1.0569,
+      "step": 193
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9023728813559321e-06,
+      "loss": 1.036,
+      "step": 194
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.9016949152542372e-06,
+      "loss": 1.0333,
+      "step": 195
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9010169491525422e-06,
+      "loss": 1.071,
+      "step": 196
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9003389830508475e-06,
+      "loss": 1.0687,
+      "step": 197
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.8996610169491525e-06,
+      "loss": 1.0969,
+      "step": 198
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8989830508474576e-06,
+      "loss": 1.0175,
+      "step": 199
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.8983050847457627e-06,
+      "loss": 1.0223,
+      "step": 200
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.8976271186440677e-06,
+      "loss": 0.9534,
+      "step": 201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.8969491525423728e-06,
+      "loss": 1.0213,
+      "step": 202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.896271186440678e-06,
+      "loss": 1.013,
+      "step": 203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.895593220338983e-06,
+      "loss": 1.0241,
+      "step": 204
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.8949152542372881e-06,
+      "loss": 1.0355,
+      "step": 205
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.894237288135593e-06,
+      "loss": 1.0511,
+      "step": 206
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.893559322033898e-06,
+      "loss": 1.0394,
+      "step": 207
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8928813559322033e-06,
+      "loss": 0.953,
+      "step": 208
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.8922033898305084e-06,
+      "loss": 1.002,
+      "step": 209
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.8915254237288134e-06,
+      "loss": 1.0233,
+      "step": 210
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8908474576271185e-06,
+      "loss": 1.0208,
+      "step": 211
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.8901694915254235e-06,
+      "loss": 1.0506,
+      "step": 212
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.8894915254237286e-06,
+      "loss": 1.046,
+      "step": 213
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.8888135593220339e-06,
+      "loss": 1.0002,
+      "step": 214
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.888135593220339e-06,
+      "loss": 0.9716,
+      "step": 215
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.887457627118644e-06,
+      "loss": 1.0353,
+      "step": 216
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.886779661016949e-06,
+      "loss": 0.9895,
+      "step": 217
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.886101694915254e-06,
+      "loss": 1.0655,
+      "step": 218
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.8854237288135593e-06,
+      "loss": 0.9634,
+      "step": 219
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.8847457627118644e-06,
+      "loss": 0.9876,
+      "step": 220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.8840677966101695e-06,
+      "loss": 0.9716,
+      "step": 221
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.8833898305084745e-06,
+      "loss": 1.0419,
+      "step": 222
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8827118644067796e-06,
+      "loss": 1.0206,
+      "step": 223
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.8820338983050848e-06,
+      "loss": 1.0537,
+      "step": 224
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.8813559322033899e-06,
+      "loss": 1.0073,
+      "step": 225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.880677966101695e-06,
+      "loss": 1.0212,
+      "step": 226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.8799999999999998e-06,
+      "loss": 0.998,
+      "step": 227
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.8793220338983048e-06,
+      "loss": 0.9832,
+      "step": 228
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.87864406779661e-06,
+      "loss": 0.9789,
+      "step": 229
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.8779661016949152e-06,
+      "loss": 1.0204,
+      "step": 230
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.8772881355932202e-06,
+      "loss": 1.0199,
+      "step": 231
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.8766101694915253e-06,
+      "loss": 1.0431,
+      "step": 232
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.8759322033898303e-06,
+      "loss": 0.9954,
+      "step": 233
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.8752542372881354e-06,
+      "loss": 0.979,
+      "step": 234
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8745762711864407e-06,
+      "loss": 1.0617,
+      "step": 235
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.8738983050847457e-06,
+      "loss": 0.9424,
+      "step": 236
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.8732203389830508e-06,
+      "loss": 0.9906,
+      "step": 237
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8725423728813558e-06,
+      "loss": 0.9598,
+      "step": 238
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.8718644067796609e-06,
+      "loss": 0.9742,
+      "step": 239
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.8711864406779661e-06,
+      "loss": 0.9749,
+      "step": 240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8705084745762712e-06,
+      "loss": 0.9873,
+      "step": 241
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.8698305084745763e-06,
+      "loss": 1.023,
+      "step": 242
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.8691525423728813e-06,
+      "loss": 0.9876,
+      "step": 243
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8684745762711864e-06,
+      "loss": 0.993,
+      "step": 244
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.8677966101694916e-06,
+      "loss": 0.933,
+      "step": 245
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.8671186440677967e-06,
+      "loss": 0.997,
+      "step": 246
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8664406779661017e-06,
+      "loss": 0.9932,
+      "step": 247
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.8657627118644066e-06,
+      "loss": 0.975,
+      "step": 248
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.8650847457627116e-06,
+      "loss": 0.9366,
+      "step": 249
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.8644067796610167e-06,
+      "loss": 0.9404,
+      "step": 250
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.863728813559322e-06,
+      "loss": 0.9429,
+      "step": 251
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.863050847457627e-06,
+      "loss": 0.9718,
+      "step": 252
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.862372881355932e-06,
+      "loss": 0.9689,
+      "step": 253
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8616949152542371e-06,
+      "loss": 0.9405,
+      "step": 254
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8610169491525422e-06,
+      "loss": 0.9855,
+      "step": 255
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8603389830508475e-06,
+      "loss": 0.9498,
+      "step": 256
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8596610169491525e-06,
+      "loss": 0.9855,
+      "step": 257
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8589830508474576e-06,
+      "loss": 0.9612,
+      "step": 258
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8583050847457626e-06,
+      "loss": 0.9946,
+      "step": 259
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8576271186440677e-06,
+      "loss": 0.9795,
+      "step": 260
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8569491525423727e-06,
+      "loss": 0.9826,
+      "step": 261
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.856271186440678e-06,
+      "loss": 0.9288,
+      "step": 262
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.855593220338983e-06,
+      "loss": 0.9907,
+      "step": 263
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8549152542372881e-06,
+      "loss": 0.9485,
+      "step": 264
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8542372881355932e-06,
+      "loss": 0.9506,
+      "step": 265
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8535593220338982e-06,
+      "loss": 1.0025,
+      "step": 266
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8528813559322035e-06,
+      "loss": 0.904,
+      "step": 267
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8522033898305085e-06,
+      "loss": 0.9143,
+      "step": 268
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8515254237288134e-06,
+      "loss": 0.9907,
+      "step": 269
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8508474576271184e-06,
+      "loss": 0.9071,
+      "step": 270
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8501694915254235e-06,
+      "loss": 0.8728,
+      "step": 271
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.8494915254237288e-06,
+      "loss": 0.886,
+      "step": 272
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8488135593220338e-06,
+      "loss": 0.9176,
+      "step": 273
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8481355932203389e-06,
+      "loss": 0.9395,
+      "step": 274
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.847457627118644e-06,
+      "loss": 1.0047,
+      "step": 275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.846779661016949e-06,
+      "loss": 0.9278,
+      "step": 276
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.846101694915254e-06,
+      "loss": 0.8953,
+      "step": 277
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8454237288135593e-06,
+      "loss": 0.8891,
+      "step": 278
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8447457627118644e-06,
+      "loss": 0.9794,
+      "step": 279
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8440677966101694e-06,
+      "loss": 0.9192,
+      "step": 280
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8433898305084745e-06,
+      "loss": 0.8931,
+      "step": 281
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8427118644067795e-06,
+      "loss": 0.9,
+      "step": 282
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8420338983050848e-06,
+      "loss": 0.928,
+      "step": 283
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8413559322033898e-06,
+      "loss": 0.9631,
+      "step": 284
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.840677966101695e-06,
+      "loss": 0.9365,
+      "step": 285
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.84e-06,
+      "loss": 0.8809,
+      "step": 286
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.839322033898305e-06,
+      "loss": 0.9445,
+      "step": 287
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8386440677966103e-06,
+      "loss": 0.8565,
+      "step": 288
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8379661016949151e-06,
+      "loss": 0.9079,
+      "step": 289
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8372881355932202e-06,
+      "loss": 0.8988,
+      "step": 290
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8366101694915252e-06,
+      "loss": 0.9305,
+      "step": 291
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8359322033898303e-06,
+      "loss": 0.8694,
+      "step": 292
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8352542372881353e-06,
+      "loss": 0.91,
+      "step": 293
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8345762711864406e-06,
+      "loss": 0.9222,
+      "step": 294
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8338983050847457e-06,
+      "loss": 0.8507,
+      "step": 295
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.8332203389830507e-06,
+      "loss": 0.8134,
+      "step": 296
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.8325423728813558e-06,
+      "loss": 0.8757,
+      "step": 297
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8318644067796608e-06,
+      "loss": 0.9181,
+      "step": 298
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.831186440677966e-06,
+      "loss": 0.9024,
+      "step": 299
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8305084745762712e-06,
+      "loss": 0.9063,
+      "step": 300
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.7133333333333334,
+      "eval_loss": 0.8667357563972473,
+      "eval_runtime": 2.1333,
+      "eval_samples_per_second": 562.519,
+      "eval_steps_per_second": 4.688,
+      "step": 300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8298305084745762e-06,
+      "loss": 0.8751,
+      "step": 301
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8291525423728813e-06,
+      "loss": 0.8468,
+      "step": 302
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8284745762711863e-06,
+      "loss": 0.8219,
+      "step": 303
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8277966101694916e-06,
+      "loss": 0.9336,
+      "step": 304
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8271186440677966e-06,
+      "loss": 0.7976,
+      "step": 305
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8264406779661017e-06,
+      "loss": 0.8596,
+      "step": 306
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8257627118644068e-06,
+      "loss": 0.9149,
+      "step": 307
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8250847457627118e-06,
+      "loss": 0.8966,
+      "step": 308
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.8244067796610169e-06,
+      "loss": 0.8069,
+      "step": 309
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.823728813559322e-06,
+      "loss": 0.7944,
+      "step": 310
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.823050847457627e-06,
+      "loss": 0.8648,
+      "step": 311
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.822372881355932e-06,
+      "loss": 0.8707,
+      "step": 312
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.821694915254237e-06,
+      "loss": 0.9647,
+      "step": 313
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.8210169491525421e-06,
+      "loss": 0.8766,
+      "step": 314
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.8203389830508474e-06,
+      "loss": 0.8581,
+      "step": 315
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8196610169491525e-06,
+      "loss": 0.8455,
+      "step": 316
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8189830508474575e-06,
+      "loss": 0.8239,
+      "step": 317
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8183050847457626e-06,
+      "loss": 0.8324,
+      "step": 318
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8176271186440676e-06,
+      "loss": 0.8952,
+      "step": 319
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.816949152542373e-06,
+      "loss": 0.784,
+      "step": 320
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.816271186440678e-06,
+      "loss": 0.7909,
+      "step": 321
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.815593220338983e-06,
+      "loss": 0.8393,
+      "step": 322
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.814915254237288e-06,
+      "loss": 0.7742,
+      "step": 323
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.8142372881355931e-06,
+      "loss": 0.8125,
+      "step": 324
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8135593220338982e-06,
+      "loss": 0.8642,
+      "step": 325
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.8128813559322034e-06,
+      "loss": 0.8029,
+      "step": 326
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8122033898305085e-06,
+      "loss": 0.7896,
+      "step": 327
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8115254237288136e-06,
+      "loss": 0.7172,
+      "step": 328
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.8108474576271186e-06,
+      "loss": 0.7582,
+      "step": 329
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.8101694915254237e-06,
+      "loss": 0.925,
+      "step": 330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8094915254237287e-06,
+      "loss": 0.7542,
+      "step": 331
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8088135593220338e-06,
+      "loss": 0.7409,
+      "step": 332
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8081355932203388e-06,
+      "loss": 0.8217,
+      "step": 333
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8074576271186439e-06,
+      "loss": 0.7942,
+      "step": 334
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.806779661016949e-06,
+      "loss": 0.798,
+      "step": 335
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8061016949152542e-06,
+      "loss": 0.7975,
+      "step": 336
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8054237288135593e-06,
+      "loss": 0.7211,
+      "step": 337
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8047457627118643e-06,
+      "loss": 0.8354,
+      "step": 338
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8040677966101694e-06,
+      "loss": 0.8051,
+      "step": 339
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8033898305084744e-06,
+      "loss": 0.8015,
+      "step": 340
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8027118644067795e-06,
+      "loss": 0.8058,
+      "step": 341
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8020338983050848e-06,
+      "loss": 0.8007,
+      "step": 342
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8013559322033898e-06,
+      "loss": 0.8089,
+      "step": 343
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.8006779661016949e-06,
+      "loss": 0.7677,
+      "step": 344
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.8e-06,
+      "loss": 0.7971,
+      "step": 345
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.799322033898305e-06,
+      "loss": 0.7407,
+      "step": 346
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7986440677966102e-06,
+      "loss": 0.7246,
+      "step": 347
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7979661016949153e-06,
+      "loss": 0.7908,
+      "step": 348
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7972881355932204e-06,
+      "loss": 0.7823,
+      "step": 349
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7966101694915254e-06,
+      "loss": 0.8502,
+      "step": 350
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7959322033898305e-06,
+      "loss": 0.7582,
+      "step": 351
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7952542372881353e-06,
+      "loss": 0.7133,
+      "step": 352
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.7945762711864406e-06,
+      "loss": 0.7181,
+      "step": 353
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.7938983050847456e-06,
+      "loss": 0.7364,
+      "step": 354
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7932203389830507e-06,
+      "loss": 0.812,
+      "step": 355
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7925423728813557e-06,
+      "loss": 0.7124,
+      "step": 356
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7918644067796608e-06,
+      "loss": 0.7259,
+      "step": 357
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.791186440677966e-06,
+      "loss": 0.7452,
+      "step": 358
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7905084745762711e-06,
+      "loss": 0.8349,
+      "step": 359
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7898305084745762e-06,
+      "loss": 0.633,
+      "step": 360
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7891525423728812e-06,
+      "loss": 0.6854,
+      "step": 361
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7884745762711863e-06,
+      "loss": 0.7067,
+      "step": 362
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7877966101694916e-06,
+      "loss": 0.6715,
+      "step": 363
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7871186440677966e-06,
+      "loss": 0.7378,
+      "step": 364
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7864406779661017e-06,
+      "loss": 0.6822,
+      "step": 365
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7857627118644067e-06,
+      "loss": 0.6902,
+      "step": 366
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7850847457627118e-06,
+      "loss": 0.6769,
+      "step": 367
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.784406779661017e-06,
+      "loss": 0.6375,
+      "step": 368
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.783728813559322e-06,
+      "loss": 0.6882,
+      "step": 369
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7830508474576271e-06,
+      "loss": 0.6189,
+      "step": 370
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7823728813559322e-06,
+      "loss": 0.7674,
+      "step": 371
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7816949152542373e-06,
+      "loss": 0.693,
+      "step": 372
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.781016949152542e-06,
+      "loss": 0.6153,
+      "step": 373
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7803389830508474e-06,
+      "loss": 0.7488,
+      "step": 374
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7796610169491524e-06,
+      "loss": 0.6897,
+      "step": 375
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7789830508474575e-06,
+      "loss": 0.7201,
+      "step": 376
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7783050847457625e-06,
+      "loss": 0.6733,
+      "step": 377
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7776271186440676e-06,
+      "loss": 0.696,
+      "step": 378
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7769491525423729e-06,
+      "loss": 0.5612,
+      "step": 379
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.776271186440678e-06,
+      "loss": 0.6796,
+      "step": 380
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.775593220338983e-06,
+      "loss": 0.7384,
+      "step": 381
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.774915254237288e-06,
+      "loss": 0.6982,
+      "step": 382
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.774237288135593e-06,
+      "loss": 0.7704,
+      "step": 383
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.7735593220338983e-06,
+      "loss": 0.6095,
+      "step": 384
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7728813559322034e-06,
+      "loss": 0.6375,
+      "step": 385
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7722033898305085e-06,
+      "loss": 0.5513,
+      "step": 386
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7715254237288135e-06,
+      "loss": 0.6198,
+      "step": 387
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7708474576271186e-06,
+      "loss": 0.7227,
+      "step": 388
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7701694915254236e-06,
+      "loss": 0.5921,
+      "step": 389
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7694915254237289e-06,
+      "loss": 0.6074,
+      "step": 390
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.768813559322034e-06,
+      "loss": 0.5944,
+      "step": 391
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.768135593220339e-06,
+      "loss": 0.7306,
+      "step": 392
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.7674576271186438e-06,
+      "loss": 0.7679,
+      "step": 393
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.766779661016949e-06,
+      "loss": 0.6495,
+      "step": 394
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7661016949152542e-06,
+      "loss": 0.6567,
+      "step": 395
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7654237288135592e-06,
+      "loss": 0.6356,
+      "step": 396
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7647457627118643e-06,
+      "loss": 0.6007,
+      "step": 397
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.7640677966101693e-06,
+      "loss": 0.5883,
+      "step": 398
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7633898305084744e-06,
+      "loss": 0.6513,
+      "step": 399
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7627118644067794e-06,
+      "loss": 0.6602,
+      "step": 400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7620338983050847e-06,
+      "loss": 0.5831,
+      "step": 401
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.7613559322033898e-06,
+      "loss": 0.6705,
+      "step": 402
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7606779661016948e-06,
+      "loss": 0.7556,
+      "step": 403
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7599999999999999e-06,
+      "loss": 0.6033,
+      "step": 404
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.759322033898305e-06,
+      "loss": 0.7712,
+      "step": 405
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7586440677966102e-06,
+      "loss": 0.6089,
+      "step": 406
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7579661016949153e-06,
+      "loss": 0.6834,
+      "step": 407
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7572881355932203e-06,
+      "loss": 0.6176,
+      "step": 408
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7566101694915254e-06,
+      "loss": 0.6615,
+      "step": 409
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7559322033898304e-06,
+      "loss": 0.5539,
+      "step": 410
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7552542372881357e-06,
+      "loss": 0.5066,
+      "step": 411
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7545762711864407e-06,
+      "loss": 0.5326,
+      "step": 412
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7538983050847458e-06,
+      "loss": 0.6448,
+      "step": 413
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7532203389830506e-06,
+      "loss": 0.6307,
+      "step": 414
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7525423728813557e-06,
+      "loss": 0.5408,
+      "step": 415
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7518644067796608e-06,
+      "loss": 0.5494,
+      "step": 416
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.751186440677966e-06,
+      "loss": 0.5943,
+      "step": 417
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.750508474576271e-06,
+      "loss": 0.7255,
+      "step": 418
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.7498305084745761e-06,
+      "loss": 0.5502,
+      "step": 419
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.7491525423728812e-06,
+      "loss": 0.5747,
+      "step": 420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.7484745762711862e-06,
+      "loss": 0.7147,
+      "step": 421
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.7477966101694915e-06,
+      "loss": 0.5059,
+      "step": 422
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7471186440677966e-06,
+      "loss": 0.5937,
+      "step": 423
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7464406779661016e-06,
+      "loss": 0.5577,
+      "step": 424
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7457627118644067e-06,
+      "loss": 0.6441,
+      "step": 425
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.7450847457627117e-06,
+      "loss": 0.6555,
+      "step": 426
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.744406779661017e-06,
+      "loss": 0.5725,
+      "step": 427
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.743728813559322e-06,
+      "loss": 0.7748,
+      "step": 428
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.7430508474576271e-06,
+      "loss": 0.6581,
+      "step": 429
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.7423728813559322e-06,
+      "loss": 0.7125,
+      "step": 430
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.7416949152542372e-06,
+      "loss": 0.5245,
+      "step": 431
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.7410169491525423e-06,
+      "loss": 0.672,
+      "step": 432
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7403389830508475e-06,
+      "loss": 0.5478,
+      "step": 433
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7396610169491526e-06,
+      "loss": 0.5794,
+      "step": 434
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7389830508474574e-06,
+      "loss": 0.5407,
+      "step": 435
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7383050847457625e-06,
+      "loss": 0.7596,
+      "step": 436
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.7376271186440676e-06,
+      "loss": 0.5474,
+      "step": 437
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.7369491525423728e-06,
+      "loss": 0.5168,
+      "step": 438
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.7362711864406779e-06,
+      "loss": 0.5813,
+      "step": 439
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.735593220338983e-06,
+      "loss": 0.6007,
+      "step": 440
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.734915254237288e-06,
+      "loss": 0.6333,
+      "step": 441
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.734237288135593e-06,
+      "loss": 0.6517,
+      "step": 442
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7335593220338983e-06,
+      "loss": 0.708,
+      "step": 443
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.7328813559322034e-06,
+      "loss": 0.5086,
+      "step": 444
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7322033898305084e-06,
+      "loss": 0.6153,
+      "step": 445
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.7315254237288135e-06,
+      "loss": 0.5577,
+      "step": 446
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7308474576271185e-06,
+      "loss": 0.5158,
+      "step": 447
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7301694915254236e-06,
+      "loss": 0.588,
+      "step": 448
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7294915254237289e-06,
+      "loss": 0.5367,
+      "step": 449
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.728813559322034e-06,
+      "loss": 0.5797,
+      "step": 450
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.855,
+      "eval_loss": 0.5155936479568481,
+      "eval_runtime": 2.1495,
+      "eval_samples_per_second": 558.264,
+      "eval_steps_per_second": 4.652,
+      "step": 450
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.728135593220339e-06,
+      "loss": 0.613,
+      "step": 451
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.727457627118644e-06,
+      "loss": 0.5857,
+      "step": 452
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.726779661016949e-06,
+      "loss": 0.5536,
+      "step": 453
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7261016949152543e-06,
+      "loss": 0.5302,
+      "step": 454
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.7254237288135594e-06,
+      "loss": 0.4908,
+      "step": 455
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.7247457627118642e-06,
+      "loss": 0.4933,
+      "step": 456
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7240677966101693e-06,
+      "loss": 0.5044,
+      "step": 457
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.7233898305084743e-06,
+      "loss": 0.5629,
+      "step": 458
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.7227118644067796e-06,
+      "loss": 0.4751,
+      "step": 459
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7220338983050847e-06,
+      "loss": 0.7558,
+      "step": 460
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.7213559322033897e-06,
+      "loss": 0.6268,
+      "step": 461
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.7206779661016948e-06,
+      "loss": 0.5394,
+      "step": 462
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7199999999999998e-06,
+      "loss": 0.4676,
+      "step": 463
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.7193220338983049e-06,
+      "loss": 0.5127,
+      "step": 464
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.7186440677966102e-06,
+      "loss": 0.6788,
+      "step": 465
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7179661016949152e-06,
+      "loss": 0.464,
+      "step": 466
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.7172881355932203e-06,
+      "loss": 0.5642,
+      "step": 467
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.7166101694915253e-06,
+      "loss": 0.4097,
+      "step": 468
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.7159322033898304e-06,
+      "loss": 0.7187,
+      "step": 469
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.7152542372881356e-06,
+      "loss": 0.5063,
+      "step": 470
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.7145762711864407e-06,
+      "loss": 0.4078,
+      "step": 471
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7138983050847458e-06,
+      "loss": 0.6129,
+      "step": 472
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.7132203389830508e-06,
+      "loss": 0.5158,
+      "step": 473
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.7125423728813559e-06,
+      "loss": 0.6836,
+      "step": 474
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.7118644067796611e-06,
+      "loss": 0.4219,
+      "step": 475
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.711186440677966e-06,
+      "loss": 0.503,
+      "step": 476
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.710508474576271e-06,
+      "loss": 0.5149,
+      "step": 477
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.709830508474576e-06,
+      "loss": 0.5265,
+      "step": 478
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.7091525423728811e-06,
+      "loss": 0.3986,
+      "step": 479
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.7084745762711862e-06,
+      "loss": 0.6311,
+      "step": 480
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7077966101694915e-06,
+      "loss": 0.4982,
+      "step": 481
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.7071186440677965e-06,
+      "loss": 0.4812,
+      "step": 482
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.7064406779661016e-06,
+      "loss": 0.4371,
+      "step": 483
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7057627118644066e-06,
+      "loss": 0.2828,
+      "step": 484
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.7050847457627117e-06,
+      "loss": 0.4966,
+      "step": 485
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.704406779661017e-06,
+      "loss": 0.487,
+      "step": 486
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.703728813559322e-06,
+      "loss": 0.48,
+      "step": 487
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.703050847457627e-06,
+      "loss": 0.53,
+      "step": 488
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7023728813559321e-06,
+      "loss": 0.4308,
+      "step": 489
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7016949152542372e-06,
+      "loss": 0.4344,
+      "step": 490
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7010169491525424e-06,
+      "loss": 0.4264,
+      "step": 491
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.7003389830508475e-06,
+      "loss": 0.3625,
+      "step": 492
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6996610169491526e-06,
+      "loss": 0.469,
+      "step": 493
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6989830508474576e-06,
+      "loss": 0.5119,
+      "step": 494
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.6983050847457627e-06,
+      "loss": 0.5313,
+      "step": 495
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6976271186440677e-06,
+      "loss": 0.4774,
+      "step": 496
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.6969491525423728e-06,
+      "loss": 0.6185,
+      "step": 497
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.6962711864406778e-06,
+      "loss": 0.5242,
+      "step": 498
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.6955932203389829e-06,
+      "loss": 0.589,
+      "step": 499
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.694915254237288e-06,
+      "loss": 0.3651,
+      "step": 500
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.694237288135593e-06,
+      "loss": 0.5152,
+      "step": 501
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6935593220338983e-06,
+      "loss": 0.3403,
+      "step": 502
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.6928813559322033e-06,
+      "loss": 0.4487,
+      "step": 503
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.6922033898305084e-06,
+      "loss": 0.5165,
+      "step": 504
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6915254237288134e-06,
+      "loss": 0.3588,
+      "step": 505
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.6908474576271185e-06,
+      "loss": 0.36,
+      "step": 506
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.6901694915254238e-06,
+      "loss": 0.3639,
+      "step": 507
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6894915254237288e-06,
+      "loss": 0.3645,
+      "step": 508
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.6888135593220339e-06,
+      "loss": 0.4184,
+      "step": 509
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.688135593220339e-06,
+      "loss": 0.3949,
+      "step": 510
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.687457627118644e-06,
+      "loss": 0.472,
+      "step": 511
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.686779661016949e-06,
+      "loss": 0.5901,
+      "step": 512
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.6861016949152543e-06,
+      "loss": 0.5605,
+      "step": 513
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6854237288135594e-06,
+      "loss": 0.3559,
+      "step": 514
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.6847457627118644e-06,
+      "loss": 0.5166,
+      "step": 515
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.6840677966101695e-06,
+      "loss": 0.4741,
+      "step": 516
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6833898305084745e-06,
+      "loss": 0.3698,
+      "step": 517
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.6827118644067796e-06,
+      "loss": 0.3327,
+      "step": 518
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.6820338983050846e-06,
+      "loss": 0.4358,
+      "step": 519
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6813559322033897e-06,
+      "loss": 0.4175,
+      "step": 520
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.6806779661016947e-06,
+      "loss": 0.5203,
+      "step": 521
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.6799999999999998e-06,
+      "loss": 0.4134,
+      "step": 522
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6793220338983049e-06,
+      "loss": 0.4262,
+      "step": 523
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.6786440677966101e-06,
+      "loss": 0.4193,
+      "step": 524
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.6779661016949152e-06,
+      "loss": 0.4083,
+      "step": 525
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6772881355932202e-06,
+      "loss": 0.4328,
+      "step": 526
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.6766101694915253e-06,
+      "loss": 0.4179,
+      "step": 527
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.6759322033898303e-06,
+      "loss": 0.4417,
+      "step": 528
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6752542372881356e-06,
+      "loss": 0.44,
+      "step": 529
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.6745762711864407e-06,
+      "loss": 0.389,
+      "step": 530
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.6738983050847457e-06,
+      "loss": 0.4417,
+      "step": 531
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6732203389830508e-06,
+      "loss": 0.6049,
+      "step": 532
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.6725423728813558e-06,
+      "loss": 0.2886,
+      "step": 533
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.671864406779661e-06,
+      "loss": 0.362,
+      "step": 534
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6711864406779662e-06,
+      "loss": 0.4893,
+      "step": 535
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.6705084745762712e-06,
+      "loss": 0.4156,
+      "step": 536
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.6698305084745763e-06,
+      "loss": 0.4368,
+      "step": 537
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.6691525423728813e-06,
+      "loss": 0.2854,
+      "step": 538
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.6684745762711862e-06,
+      "loss": 0.3816,
+      "step": 539
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.6677966101694914e-06,
+      "loss": 0.3923,
+      "step": 540
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.6671186440677965e-06,
+      "loss": 0.3821,
+      "step": 541
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.6664406779661015e-06,
+      "loss": 0.3538,
+      "step": 542
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.6657627118644066e-06,
+      "loss": 0.4051,
+      "step": 543
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.6650847457627116e-06,
+      "loss": 0.2688,
+      "step": 544
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.664406779661017e-06,
+      "loss": 0.4024,
+      "step": 545
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.663728813559322e-06,
+      "loss": 0.4056,
+      "step": 546
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.663050847457627e-06,
+      "loss": 0.474,
+      "step": 547
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.662372881355932e-06,
+      "loss": 0.3274,
+      "step": 548
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.6616949152542371e-06,
+      "loss": 0.2781,
+      "step": 549
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.6610169491525424e-06,
+      "loss": 0.5106,
+      "step": 550
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.6603389830508475e-06,
+      "loss": 0.4192,
+      "step": 551
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.6596610169491525e-06,
+      "loss": 0.5165,
+      "step": 552
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6589830508474576e-06,
+      "loss": 0.3866,
+      "step": 553
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6583050847457626e-06,
+      "loss": 0.4865,
+      "step": 554
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6576271186440679e-06,
+      "loss": 0.4561,
+      "step": 555
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.656949152542373e-06,
+      "loss": 0.3234,
+      "step": 556
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.656271186440678e-06,
+      "loss": 0.3329,
+      "step": 557
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.655593220338983e-06,
+      "loss": 0.2898,
+      "step": 558
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.654915254237288e-06,
+      "loss": 0.2935,
+      "step": 559
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.654237288135593e-06,
+      "loss": 0.5655,
+      "step": 560
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.6535593220338982e-06,
+      "loss": 0.2198,
+      "step": 561
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6528813559322033e-06,
+      "loss": 0.3579,
+      "step": 562
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.6522033898305083e-06,
+      "loss": 0.4593,
+      "step": 563
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.6515254237288134e-06,
+      "loss": 0.4252,
+      "step": 564
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6508474576271184e-06,
+      "loss": 0.3378,
+      "step": 565
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6501694915254237e-06,
+      "loss": 0.3833,
+      "step": 566
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6494915254237288e-06,
+      "loss": 0.3362,
+      "step": 567
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.6488135593220338e-06,
+      "loss": 0.4184,
+      "step": 568
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.6481355932203389e-06,
+      "loss": 0.4318,
+      "step": 569
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.647457627118644e-06,
+      "loss": 0.4544,
+      "step": 570
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.646779661016949e-06,
+      "loss": 0.2803,
+      "step": 571
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.6461016949152543e-06,
+      "loss": 0.4824,
+      "step": 572
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.6454237288135593e-06,
+      "loss": 0.2617,
+      "step": 573
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.6447457627118644e-06,
+      "loss": 0.4356,
+      "step": 574
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.6440677966101694e-06,
+      "loss": 0.3887,
+      "step": 575
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.6433898305084745e-06,
+      "loss": 0.3377,
+      "step": 576
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.6427118644067797e-06,
+      "loss": 0.3548,
+      "step": 577
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.6420338983050848e-06,
+      "loss": 0.3864,
+      "step": 578
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.6413559322033899e-06,
+      "loss": 0.3168,
+      "step": 579
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.6406779661016947e-06,
+      "loss": 0.2914,
+      "step": 580
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.6399999999999998e-06,
+      "loss": 0.3451,
+      "step": 581
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.639322033898305e-06,
+      "loss": 0.26,
+      "step": 582
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.63864406779661e-06,
+      "loss": 0.289,
+      "step": 583
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.6379661016949151e-06,
+      "loss": 0.3175,
+      "step": 584
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.6372881355932202e-06,
+      "loss": 0.4027,
+      "step": 585
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.6366101694915252e-06,
+      "loss": 0.4006,
+      "step": 586
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.6359322033898303e-06,
+      "loss": 0.4482,
+      "step": 587
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.6352542372881356e-06,
+      "loss": 0.3256,
+      "step": 588
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.6345762711864406e-06,
+      "loss": 0.2944,
+      "step": 589
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.6338983050847457e-06,
+      "loss": 0.2837,
+      "step": 590
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.6332203389830507e-06,
+      "loss": 0.2364,
+      "step": 591
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.6325423728813558e-06,
+      "loss": 0.3801,
+      "step": 592
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.631864406779661e-06,
+      "loss": 0.3241,
+      "step": 593
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.6311864406779661e-06,
+      "loss": 0.3508,
+      "step": 594
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.6305084745762712e-06,
+      "loss": 0.4068,
+      "step": 595
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.6298305084745762e-06,
+      "loss": 0.3879,
+      "step": 596
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.6291525423728813e-06,
+      "loss": 0.4832,
+      "step": 597
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.6284745762711865e-06,
+      "loss": 0.4062,
+      "step": 598
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.6277966101694916e-06,
+      "loss": 0.3831,
+      "step": 599
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.6271186440677967e-06,
+      "loss": 0.2905,
+      "step": 600
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9041666666666667,
+      "eval_loss": 0.33033275604248047,
+      "eval_runtime": 2.1567,
+      "eval_samples_per_second": 556.397,
+      "eval_steps_per_second": 4.637,
+      "step": 600
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.6264406779661015e-06,
+      "loss": 0.3392,
+      "step": 601
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.6257627118644066e-06,
+      "loss": 0.7027,
+      "step": 602
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.6250847457627116e-06,
+      "loss": 0.2786,
+      "step": 603
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.6244067796610169e-06,
+      "loss": 0.5368,
+      "step": 604
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.623728813559322e-06,
+      "loss": 0.4069,
+      "step": 605
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.623050847457627e-06,
+      "loss": 0.2945,
+      "step": 606
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.622372881355932e-06,
+      "loss": 0.2948,
+      "step": 607
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.621694915254237e-06,
+      "loss": 0.3306,
+      "step": 608
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.6210169491525424e-06,
+      "loss": 0.3564,
+      "step": 609
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.6203389830508474e-06,
+      "loss": 0.3317,
+      "step": 610
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.6196610169491525e-06,
+      "loss": 0.3692,
+      "step": 611
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.6189830508474575e-06,
+      "loss": 0.2354,
+      "step": 612
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6183050847457626e-06,
+      "loss": 0.3354,
+      "step": 613
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6176271186440679e-06,
+      "loss": 0.4009,
+      "step": 614
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.616949152542373e-06,
+      "loss": 0.4017,
+      "step": 615
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.616271186440678e-06,
+      "loss": 0.3982,
+      "step": 616
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.615593220338983e-06,
+      "loss": 0.3335,
+      "step": 617
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.614915254237288e-06,
+      "loss": 0.3801,
+      "step": 618
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.6142372881355931e-06,
+      "loss": 0.1589,
+      "step": 619
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.6135593220338984e-06,
+      "loss": 0.3561,
+      "step": 620
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.6128813559322035e-06,
+      "loss": 0.3253,
+      "step": 621
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.6122033898305083e-06,
+      "loss": 0.3365,
+      "step": 622
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.6115254237288134e-06,
+      "loss": 0.3027,
+      "step": 623
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.6108474576271184e-06,
+      "loss": 0.3476,
+      "step": 624
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6101694915254237e-06,
+      "loss": 0.3315,
+      "step": 625
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.6094915254237287e-06,
+      "loss": 0.2791,
+      "step": 626
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.6088135593220338e-06,
+      "loss": 0.4971,
+      "step": 627
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.6081355932203388e-06,
+      "loss": 0.3441,
+      "step": 628
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.607457627118644e-06,
+      "loss": 0.2838,
+      "step": 629
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.6067796610169492e-06,
+      "loss": 0.4489,
+      "step": 630
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6061016949152542e-06,
+      "loss": 0.3462,
+      "step": 631
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.6054237288135593e-06,
+      "loss": 0.3824,
+      "step": 632
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.6047457627118643e-06,
+      "loss": 0.3284,
+      "step": 633
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6040677966101694e-06,
+      "loss": 0.3177,
+      "step": 634
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.6033898305084744e-06,
+      "loss": 0.2726,
+      "step": 635
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.6027118644067797e-06,
+      "loss": 0.285,
+      "step": 636
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6020338983050848e-06,
+      "loss": 0.1887,
+      "step": 637
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.6013559322033898e-06,
+      "loss": 0.3477,
+      "step": 638
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.6006779661016949e-06,
+      "loss": 0.4365,
+      "step": 639
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.6e-06,
+      "loss": 0.2508,
+      "step": 640
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.5993220338983052e-06,
+      "loss": 0.3327,
+      "step": 641
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.5986440677966103e-06,
+      "loss": 0.3568,
+      "step": 642
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.597966101694915e-06,
+      "loss": 0.3114,
+      "step": 643
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.5972881355932201e-06,
+      "loss": 0.1868,
+      "step": 644
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.5966101694915252e-06,
+      "loss": 0.3603,
+      "step": 645
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.5959322033898305e-06,
+      "loss": 0.3288,
+      "step": 646
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.5952542372881355e-06,
+      "loss": 0.3535,
+      "step": 647
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.5945762711864406e-06,
+      "loss": 0.2105,
+      "step": 648
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.5938983050847456e-06,
+      "loss": 0.263,
+      "step": 649
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.5932203389830507e-06,
+      "loss": 0.3083,
+      "step": 650
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.5925423728813557e-06,
+      "loss": 0.3208,
+      "step": 651
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.591864406779661e-06,
+      "loss": 0.3202,
+      "step": 652
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.591186440677966e-06,
+      "loss": 0.1551,
+      "step": 653
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.5905084745762711e-06,
+      "loss": 0.3373,
+      "step": 654
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.5898305084745762e-06,
+      "loss": 0.1501,
+      "step": 655
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.5891525423728812e-06,
+      "loss": 0.2839,
+      "step": 656
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.5884745762711865e-06,
+      "loss": 0.223,
+      "step": 657
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.5877966101694916e-06,
+      "loss": 0.2818,
+      "step": 658
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.5871186440677966e-06,
+      "loss": 0.3686,
+      "step": 659
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.5864406779661017e-06,
+      "loss": 0.186,
+      "step": 660
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.5857627118644067e-06,
+      "loss": 0.1631,
+      "step": 661
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.585084745762712e-06,
+      "loss": 0.1846,
+      "step": 662
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.5844067796610168e-06,
+      "loss": 0.3073,
+      "step": 663
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.5837288135593219e-06,
+      "loss": 0.2635,
+      "step": 664
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.583050847457627e-06,
+      "loss": 0.2,
+      "step": 665
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.582372881355932e-06,
+      "loss": 0.2204,
+      "step": 666
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.581694915254237e-06,
+      "loss": 0.1922,
+      "step": 667
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.5810169491525423e-06,
+      "loss": 0.3237,
+      "step": 668
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.5803389830508474e-06,
+      "loss": 0.3518,
+      "step": 669
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.5796610169491524e-06,
+      "loss": 0.2762,
+      "step": 670
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.5789830508474575e-06,
+      "loss": 0.2683,
+      "step": 671
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.5783050847457625e-06,
+      "loss": 0.2838,
+      "step": 672
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.5776271186440678e-06,
+      "loss": 0.2398,
+      "step": 673
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.5769491525423729e-06,
+      "loss": 0.1706,
+      "step": 674
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.576271186440678e-06,
+      "loss": 0.3316,
+      "step": 675
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.575593220338983e-06,
+      "loss": 0.1924,
+      "step": 676
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.574915254237288e-06,
+      "loss": 0.4937,
+      "step": 677
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.5742372881355933e-06,
+      "loss": 0.1876,
+      "step": 678
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.5735593220338984e-06,
+      "loss": 0.3832,
+      "step": 679
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.5728813559322034e-06,
+      "loss": 0.1886,
+      "step": 680
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.5722033898305085e-06,
+      "loss": 0.2339,
+      "step": 681
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.5715254237288135e-06,
+      "loss": 0.322,
+      "step": 682
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.5708474576271186e-06,
+      "loss": 0.266,
+      "step": 683
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.5701694915254236e-06,
+      "loss": 0.3082,
+      "step": 684
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.5694915254237287e-06,
+      "loss": 0.2399,
+      "step": 685
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.5688135593220337e-06,
+      "loss": 0.3131,
+      "step": 686
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.5681355932203388e-06,
+      "loss": 0.3412,
+      "step": 687
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.5674576271186439e-06,
+      "loss": 0.2242,
+      "step": 688
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.5667796610169491e-06,
+      "loss": 0.1247,
+      "step": 689
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5661016949152542e-06,
+      "loss": 0.2313,
+      "step": 690
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5654237288135592e-06,
+      "loss": 0.1711,
+      "step": 691
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5647457627118643e-06,
+      "loss": 0.2257,
+      "step": 692
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5640677966101693e-06,
+      "loss": 0.1452,
+      "step": 693
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.5633898305084746e-06,
+      "loss": 0.2224,
+      "step": 694
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.5627118644067797e-06,
+      "loss": 0.2951,
+      "step": 695
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.5620338983050847e-06,
+      "loss": 0.2922,
+      "step": 696
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.5613559322033898e-06,
+      "loss": 0.1637,
+      "step": 697
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.5606779661016948e-06,
+      "loss": 0.2558,
+      "step": 698
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 0.3369,
+      "step": 699
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.5593220338983052e-06,
+      "loss": 0.1788,
+      "step": 700
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.5586440677966102e-06,
+      "loss": 0.2448,
+      "step": 701
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.5579661016949153e-06,
+      "loss": 0.2631,
+      "step": 702
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5572881355932203e-06,
+      "loss": 0.2031,
+      "step": 703
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.5566101694915254e-06,
+      "loss": 0.3571,
+      "step": 704
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.5559322033898304e-06,
+      "loss": 0.4733,
+      "step": 705
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.5552542372881355e-06,
+      "loss": 0.3868,
+      "step": 706
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.5545762711864405e-06,
+      "loss": 0.376,
+      "step": 707
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5538983050847456e-06,
+      "loss": 0.3649,
+      "step": 708
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5532203389830507e-06,
+      "loss": 0.2754,
+      "step": 709
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5525423728813557e-06,
+      "loss": 0.2406,
+      "step": 710
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.551864406779661e-06,
+      "loss": 0.2395,
+      "step": 711
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.551186440677966e-06,
+      "loss": 0.2258,
+      "step": 712
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.550508474576271e-06,
+      "loss": 0.1468,
+      "step": 713
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.5498305084745761e-06,
+      "loss": 0.2688,
+      "step": 714
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5491525423728812e-06,
+      "loss": 0.2329,
+      "step": 715
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.5484745762711865e-06,
+      "loss": 0.1794,
+      "step": 716
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.5477966101694915e-06,
+      "loss": 0.2069,
+      "step": 717
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5471186440677966e-06,
+      "loss": 0.3916,
+      "step": 718
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.5464406779661016e-06,
+      "loss": 0.1633,
+      "step": 719
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5457627118644067e-06,
+      "loss": 0.4608,
+      "step": 720
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.545084745762712e-06,
+      "loss": 0.3149,
+      "step": 721
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.544406779661017e-06,
+      "loss": 0.3074,
+      "step": 722
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.543728813559322e-06,
+      "loss": 0.1666,
+      "step": 723
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5430508474576271e-06,
+      "loss": 0.235,
+      "step": 724
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.5423728813559322e-06,
+      "loss": 0.3463,
+      "step": 725
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.541694915254237e-06,
+      "loss": 0.151,
+      "step": 726
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5410169491525423e-06,
+      "loss": 0.1855,
+      "step": 727
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.5403389830508473e-06,
+      "loss": 0.2876,
+      "step": 728
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.5396610169491524e-06,
+      "loss": 0.2788,
+      "step": 729
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5389830508474574e-06,
+      "loss": 0.2637,
+      "step": 730
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.5383050847457625e-06,
+      "loss": 0.2103,
+      "step": 731
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5376271186440678e-06,
+      "loss": 0.1924,
+      "step": 732
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5369491525423728e-06,
+      "loss": 0.2806,
+      "step": 733
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.5362711864406779e-06,
+      "loss": 0.3203,
+      "step": 734
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.535593220338983e-06,
+      "loss": 0.2389,
+      "step": 735
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.534915254237288e-06,
+      "loss": 0.3997,
+      "step": 736
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5342372881355933e-06,
+      "loss": 0.4916,
+      "step": 737
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.5335593220338983e-06,
+      "loss": 0.3362,
+      "step": 738
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5328813559322034e-06,
+      "loss": 0.2262,
+      "step": 739
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.5322033898305084e-06,
+      "loss": 0.1974,
+      "step": 740
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.5315254237288135e-06,
+      "loss": 0.2708,
+      "step": 741
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5308474576271187e-06,
+      "loss": 0.141,
+      "step": 742
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.5301694915254238e-06,
+      "loss": 0.1588,
+      "step": 743
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5294915254237289e-06,
+      "loss": 0.3345,
+      "step": 744
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.528813559322034e-06,
+      "loss": 0.2939,
+      "step": 745
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.5281355932203388e-06,
+      "loss": 0.2862,
+      "step": 746
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5274576271186438e-06,
+      "loss": 0.1724,
+      "step": 747
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.526779661016949e-06,
+      "loss": 0.2035,
+      "step": 748
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.5261016949152541e-06,
+      "loss": 0.2464,
+      "step": 749
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5254237288135592e-06,
+      "loss": 0.3118,
+      "step": 750
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9233333333333333,
+      "eval_loss": 0.24194158613681793,
+      "eval_runtime": 2.1646,
+      "eval_samples_per_second": 554.376,
+      "eval_steps_per_second": 4.62,
+      "step": 750
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5247457627118642e-06,
+      "loss": 0.3308,
+      "step": 751
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.5240677966101693e-06,
+      "loss": 0.2512,
+      "step": 752
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.5233898305084746e-06,
+      "loss": 0.1865,
+      "step": 753
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5227118644067796e-06,
+      "loss": 0.3534,
+      "step": 754
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.5220338983050847e-06,
+      "loss": 0.1939,
+      "step": 755
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.5213559322033897e-06,
+      "loss": 0.2233,
+      "step": 756
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5206779661016948e-06,
+      "loss": 0.2171,
+      "step": 757
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.5199999999999998e-06,
+      "loss": 0.1356,
+      "step": 758
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.5193220338983051e-06,
+      "loss": 0.409,
+      "step": 759
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5186440677966102e-06,
+      "loss": 0.2212,
+      "step": 760
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.5179661016949152e-06,
+      "loss": 0.2672,
+      "step": 761
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.5172881355932203e-06,
+      "loss": 0.4102,
+      "step": 762
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5166101694915253e-06,
+      "loss": 0.2248,
+      "step": 763
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.5159322033898306e-06,
+      "loss": 0.2012,
+      "step": 764
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.5152542372881357e-06,
+      "loss": 0.2082,
+      "step": 765
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.5145762711864407e-06,
+      "loss": 0.3749,
+      "step": 766
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.5138983050847456e-06,
+      "loss": 0.2099,
+      "step": 767
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.5132203389830506e-06,
+      "loss": 0.2483,
+      "step": 768
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.5125423728813559e-06,
+      "loss": 0.1525,
+      "step": 769
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.511864406779661e-06,
+      "loss": 0.1175,
+      "step": 770
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.511186440677966e-06,
+      "loss": 0.1789,
+      "step": 771
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.510508474576271e-06,
+      "loss": 0.152,
+      "step": 772
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.509830508474576e-06,
+      "loss": 0.1298,
+      "step": 773
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.5091525423728812e-06,
+      "loss": 0.2071,
+      "step": 774
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.5084745762711864e-06,
+      "loss": 0.1652,
+      "step": 775
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 1.5077966101694915e-06,
+      "loss": 0.3342,
+      "step": 776
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.5071186440677965e-06,
+      "loss": 0.22,
+      "step": 777
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.5064406779661016e-06,
+      "loss": 0.1648,
+      "step": 778
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 1.5057627118644066e-06,
+      "loss": 0.2909,
+      "step": 779
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 1.505084745762712e-06,
+      "loss": 0.1288,
+      "step": 780
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.504406779661017e-06,
+      "loss": 0.2381,
+      "step": 781
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.503728813559322e-06,
+      "loss": 0.1202,
+      "step": 782
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 1.503050847457627e-06,
+      "loss": 0.2136,
+      "step": 783
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5023728813559321e-06,
+      "loss": 0.1663,
+      "step": 784
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.5016949152542374e-06,
+      "loss": 0.2027,
+      "step": 785
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 1.5010169491525425e-06,
+      "loss": 0.2011,
+      "step": 786
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.5003389830508475e-06,
+      "loss": 0.2096,
+      "step": 787
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.4996610169491524e-06,
+      "loss": 0.256,
+      "step": 788
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 1.4989830508474574e-06,
+      "loss": 0.309,
+      "step": 789
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4983050847457625e-06,
+      "loss": 0.144,
+      "step": 790
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.4976271186440677e-06,
+      "loss": 0.1867,
+      "step": 791
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 1.4969491525423728e-06,
+      "loss": 0.2838,
+      "step": 792
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.4962711864406778e-06,
+      "loss": 0.2575,
+      "step": 793
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 1.495593220338983e-06,
+      "loss": 0.1512,
+      "step": 794
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.494915254237288e-06,
+      "loss": 0.1375,
+      "step": 795
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4942372881355932e-06,
+      "loss": 0.1081,
+      "step": 796
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 1.4935593220338983e-06,
+      "loss": 0.2312,
+      "step": 797
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.4928813559322033e-06,
+      "loss": 0.0993,
+      "step": 798
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4922033898305084e-06,
+      "loss": 0.1649,
+      "step": 799
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 1.4915254237288134e-06,
+      "loss": 0.182,
+      "step": 800
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.4908474576271187e-06,
+      "loss": 0.5477,
+      "step": 801
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4901694915254238e-06,
+      "loss": 0.0945,
+      "step": 802
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 1.4894915254237288e-06,
+      "loss": 0.1891,
+      "step": 803
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.4888135593220339e-06,
+      "loss": 0.1851,
+      "step": 804
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.488135593220339e-06,
+      "loss": 0.1563,
+      "step": 805
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 1.487457627118644e-06,
+      "loss": 0.2035,
+      "step": 806
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 1.4867796610169493e-06,
+      "loss": 0.0855,
+      "step": 807
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4861016949152543e-06,
+      "loss": 0.2168,
+      "step": 808
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.4854237288135592e-06,
+      "loss": 0.1785,
+      "step": 809
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 1.4847457627118642e-06,
+      "loss": 0.1845,
+      "step": 810
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4840677966101693e-06,
+      "loss": 0.2782,
+      "step": 811
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.4833898305084745e-06,
+      "loss": 0.1818,
+      "step": 812
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.4827118644067796e-06,
+      "loss": 0.1847,
+      "step": 813
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4820338983050846e-06,
+      "loss": 0.5401,
+      "step": 814
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.4813559322033897e-06,
+      "loss": 0.2525,
+      "step": 815
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 1.4806779661016948e-06,
+      "loss": 0.311,
+      "step": 816
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.48e-06,
+      "loss": 0.1483,
+      "step": 817
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.479322033898305e-06,
+      "loss": 0.1421,
+      "step": 818
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.4786440677966101e-06,
+      "loss": 0.26,
+      "step": 819
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.4779661016949152e-06,
+      "loss": 0.3372,
+      "step": 820
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 1.4772881355932202e-06,
+      "loss": 0.242,
+      "step": 821
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.4766101694915253e-06,
+      "loss": 0.0846,
+      "step": 822
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4759322033898306e-06,
+      "loss": 0.1841,
+      "step": 823
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 1.4752542372881356e-06,
+      "loss": 0.3118,
+      "step": 824
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.4745762711864407e-06,
+      "loss": 0.1246,
+      "step": 825
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4738983050847457e-06,
+      "loss": 0.1171,
+      "step": 826
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.4732203389830508e-06,
+      "loss": 0.1163,
+      "step": 827
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.472542372881356e-06,
+      "loss": 0.1439,
+      "step": 828
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.471864406779661e-06,
+      "loss": 0.3402,
+      "step": 829
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.471186440677966e-06,
+      "loss": 0.2218,
+      "step": 830
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.470508474576271e-06,
+      "loss": 0.0947,
+      "step": 831
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.469830508474576e-06,
+      "loss": 0.1727,
+      "step": 832
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 1.4691525423728811e-06,
+      "loss": 0.2279,
+      "step": 833
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4684745762711864e-06,
+      "loss": 0.1859,
+      "step": 834
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4677966101694914e-06,
+      "loss": 0.3139,
+      "step": 835
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.4671186440677965e-06,
+      "loss": 0.1925,
+      "step": 836
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4664406779661015e-06,
+      "loss": 0.2804,
+      "step": 837
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4657627118644066e-06,
+      "loss": 0.2243,
+      "step": 838
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.4650847457627119e-06,
+      "loss": 0.1766,
+      "step": 839
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 1.464406779661017e-06,
+      "loss": 0.1647,
+      "step": 840
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.463728813559322e-06,
+      "loss": 0.329,
+      "step": 841
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.463050847457627e-06,
+      "loss": 0.1023,
+      "step": 842
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.462372881355932e-06,
+      "loss": 0.1247,
+      "step": 843
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4616949152542374e-06,
+      "loss": 0.1611,
+      "step": 844
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.4610169491525424e-06,
+      "loss": 0.1313,
+      "step": 845
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.4603389830508475e-06,
+      "loss": 0.1791,
+      "step": 846
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4596610169491525e-06,
+      "loss": 0.1911,
+      "step": 847
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.4589830508474576e-06,
+      "loss": 0.1199,
+      "step": 848
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.4583050847457628e-06,
+      "loss": 0.2245,
+      "step": 849
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4576271186440677e-06,
+      "loss": 0.1106,
+      "step": 850
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.4569491525423727e-06,
+      "loss": 0.1827,
+      "step": 851
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.4562711864406778e-06,
+      "loss": 0.2192,
+      "step": 852
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.4555932203389829e-06,
+      "loss": 0.2889,
+      "step": 853
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.454915254237288e-06,
+      "loss": 0.2722,
+      "step": 854
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.4542372881355932e-06,
+      "loss": 0.222,
+      "step": 855
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4535593220338982e-06,
+      "loss": 0.1417,
+      "step": 856
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 1.4528813559322033e-06,
+      "loss": 0.3165,
+      "step": 857
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.4522033898305083e-06,
+      "loss": 0.1703,
+      "step": 858
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4515254237288134e-06,
+      "loss": 0.1312,
+      "step": 859
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 1.4508474576271187e-06,
+      "loss": 0.1586,
+      "step": 860
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 1.4501694915254237e-06,
+      "loss": 0.3892,
+      "step": 861
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4494915254237288e-06,
+      "loss": 0.2609,
+      "step": 862
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.4488135593220338e-06,
+      "loss": 0.1844,
+      "step": 863
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 1.4481355932203389e-06,
+      "loss": 0.2793,
+      "step": 864
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4474576271186442e-06,
+      "loss": 0.2283,
+      "step": 865
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.4467796610169492e-06,
+      "loss": 0.2165,
+      "step": 866
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 1.4461016949152543e-06,
+      "loss": 0.2505,
+      "step": 867
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4454237288135593e-06,
+      "loss": 0.1226,
+      "step": 868
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.4447457627118644e-06,
+      "loss": 0.2771,
+      "step": 869
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 1.4440677966101694e-06,
+      "loss": 0.1191,
+      "step": 870
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4433898305084745e-06,
+      "loss": 0.2561,
+      "step": 871
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 1.4427118644067795e-06,
+      "loss": 0.2066,
+      "step": 872
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.4420338983050846e-06,
+      "loss": 0.2358,
+      "step": 873
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.4413559322033897e-06,
+      "loss": 0.1005,
+      "step": 874
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 1.4406779661016947e-06,
+      "loss": 0.0813,
+      "step": 875
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.44e-06,
+      "loss": 0.1356,
+      "step": 876
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.439322033898305e-06,
+      "loss": 0.1793,
+      "step": 877
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 1.43864406779661e-06,
+      "loss": 0.116,
+      "step": 878
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.4379661016949151e-06,
+      "loss": 0.3809,
+      "step": 879
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.4372881355932202e-06,
+      "loss": 0.3438,
+      "step": 880
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 1.4366101694915253e-06,
+      "loss": 0.1055,
+      "step": 881
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.4359322033898305e-06,
+      "loss": 0.1812,
+      "step": 882
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.4352542372881356e-06,
+      "loss": 0.2359,
+      "step": 883
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 1.4345762711864406e-06,
+      "loss": 0.2917,
+      "step": 884
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 1.4338983050847457e-06,
+      "loss": 0.188,
+      "step": 885
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.4332203389830507e-06,
+      "loss": 0.2786,
+      "step": 886
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.432542372881356e-06,
+      "loss": 0.272,
+      "step": 887
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 1.431864406779661e-06,
+      "loss": 0.2009,
+      "step": 888
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.4311864406779661e-06,
+      "loss": 0.1849,
+      "step": 889
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.4305084745762712e-06,
+      "loss": 0.2114,
+      "step": 890
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 1.4298305084745762e-06,
+      "loss": 0.198,
+      "step": 891
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.4291525423728813e-06,
+      "loss": 0.2431,
+      "step": 892
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.4284745762711863e-06,
+      "loss": 0.2313,
+      "step": 893
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.4277966101694914e-06,
+      "loss": 0.4397,
+      "step": 894
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.4271186440677965e-06,
+      "loss": 0.2812,
+      "step": 895
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.4264406779661015e-06,
+      "loss": 0.3551,
+      "step": 896
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 1.4257627118644066e-06,
+      "loss": 0.2413,
+      "step": 897
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.4250847457627118e-06,
+      "loss": 0.1721,
+      "step": 898
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 1.4244067796610169e-06,
+      "loss": 0.228,
+      "step": 899
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.423728813559322e-06,
+      "loss": 0.2364,
+      "step": 900
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.9275,
+      "eval_loss": 0.20115427672863007,
+      "eval_runtime": 2.1625,
+      "eval_samples_per_second": 554.901,
+      "eval_steps_per_second": 4.624,
+      "step": 900
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.423050847457627e-06,
+      "loss": 0.0924,
+      "step": 901
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.422372881355932e-06,
+      "loss": 0.3558,
+      "step": 902
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.4216949152542373e-06,
+      "loss": 0.1723,
+      "step": 903
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.4210169491525424e-06,
+      "loss": 0.199,
+      "step": 904
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.4203389830508474e-06,
+      "loss": 0.2078,
+      "step": 905
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.4196610169491525e-06,
+      "loss": 0.0834,
+      "step": 906
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.4189830508474575e-06,
+      "loss": 0.1201,
+      "step": 907
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.4183050847457628e-06,
+      "loss": 0.1316,
+      "step": 908
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.4176271186440679e-06,
+      "loss": 0.1494,
+      "step": 909
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.416949152542373e-06,
+      "loss": 0.0932,
+      "step": 910
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.416271186440678e-06,
+      "loss": 0.2139,
+      "step": 911
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.415593220338983e-06,
+      "loss": 0.1086,
+      "step": 912
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.4149152542372879e-06,
+      "loss": 0.2021,
+      "step": 913
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.4142372881355931e-06,
+      "loss": 0.1413,
+      "step": 914
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.4135593220338982e-06,
+      "loss": 0.1648,
+      "step": 915
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.4128813559322032e-06,
+      "loss": 0.1543,
+      "step": 916
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.4122033898305083e-06,
+      "loss": 0.1439,
+      "step": 917
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.4115254237288134e-06,
+      "loss": 0.0855,
+      "step": 918
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.4108474576271186e-06,
+      "loss": 0.1591,
+      "step": 919
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.4101694915254237e-06,
+      "loss": 0.1353,
+      "step": 920
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.4094915254237287e-06,
+      "loss": 0.1186,
+      "step": 921
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.4088135593220338e-06,
+      "loss": 0.33,
+      "step": 922
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.4081355932203388e-06,
+      "loss": 0.1859,
+      "step": 923
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.4074576271186441e-06,
+      "loss": 0.2789,
+      "step": 924
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.4067796610169492e-06,
+      "loss": 0.1976,
+      "step": 925
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.4061016949152542e-06,
+      "loss": 0.2264,
+      "step": 926
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.4054237288135593e-06,
+      "loss": 0.1494,
+      "step": 927
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.4047457627118643e-06,
+      "loss": 0.2535,
+      "step": 928
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.4040677966101694e-06,
+      "loss": 0.1734,
+      "step": 929
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.4033898305084747e-06,
+      "loss": 0.0785,
+      "step": 930
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.4027118644067797e-06,
+      "loss": 0.1786,
+      "step": 931
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.4020338983050848e-06,
+      "loss": 0.1981,
+      "step": 932
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.4013559322033896e-06,
+      "loss": 0.3671,
+      "step": 933
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.4006779661016947e-06,
+      "loss": 0.133,
+      "step": 934
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.4e-06,
+      "loss": 0.1537,
+      "step": 935
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.399322033898305e-06,
+      "loss": 0.2066,
+      "step": 936
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.39864406779661e-06,
+      "loss": 0.197,
+      "step": 937
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.397966101694915e-06,
+      "loss": 0.1113,
+      "step": 938
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.3972881355932202e-06,
+      "loss": 0.196,
+      "step": 939
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.3966101694915254e-06,
+      "loss": 0.16,
+      "step": 940
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.3959322033898305e-06,
+      "loss": 0.3039,
+      "step": 941
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.3952542372881355e-06,
+      "loss": 0.3009,
+      "step": 942
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.3945762711864406e-06,
+      "loss": 0.2367,
+      "step": 943
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.3938983050847456e-06,
+      "loss": 0.1706,
+      "step": 944
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.3932203389830507e-06,
+      "loss": 0.0864,
+      "step": 945
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.392542372881356e-06,
+      "loss": 0.1151,
+      "step": 946
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.391864406779661e-06,
+      "loss": 0.1147,
+      "step": 947
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.391186440677966e-06,
+      "loss": 0.1549,
+      "step": 948
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.3905084745762711e-06,
+      "loss": 0.1189,
+      "step": 949
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.3898305084745762e-06,
+      "loss": 0.1248,
+      "step": 950
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.3891525423728815e-06,
+      "loss": 0.155,
+      "step": 951
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.3884745762711865e-06,
+      "loss": 0.1817,
+      "step": 952
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.3877966101694916e-06,
+      "loss": 0.076,
+      "step": 953
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.3871186440677964e-06,
+      "loss": 0.1318,
+      "step": 954
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.3864406779661015e-06,
+      "loss": 0.2379,
+      "step": 955
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.3857627118644067e-06,
+      "loss": 0.1208,
+      "step": 956
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.3850847457627118e-06,
+      "loss": 0.2364,
+      "step": 957
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.3844067796610168e-06,
+      "loss": 0.3038,
+      "step": 958
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.383728813559322e-06,
+      "loss": 0.1233,
+      "step": 959
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.383050847457627e-06,
+      "loss": 0.1563,
+      "step": 960
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.382372881355932e-06,
+      "loss": 0.3739,
+      "step": 961
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.3816949152542373e-06,
+      "loss": 0.1857,
+      "step": 962
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.3810169491525423e-06,
+      "loss": 0.0931,
+      "step": 963
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.3803389830508474e-06,
+      "loss": 0.3036,
+      "step": 964
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.3796610169491524e-06,
+      "loss": 0.1429,
+      "step": 965
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.3789830508474575e-06,
+      "loss": 0.3071,
+      "step": 966
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.3783050847457628e-06,
+      "loss": 0.2532,
+      "step": 967
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.3776271186440678e-06,
+      "loss": 0.05,
+      "step": 968
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.3769491525423729e-06,
+      "loss": 0.2778,
+      "step": 969
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.376271186440678e-06,
+      "loss": 0.2016,
+      "step": 970
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.375593220338983e-06,
+      "loss": 0.1147,
+      "step": 971
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.3749152542372883e-06,
+      "loss": 0.1774,
+      "step": 972
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.3742372881355933e-06,
+      "loss": 0.2507,
+      "step": 973
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.3735593220338984e-06,
+      "loss": 0.1127,
+      "step": 974
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.3728813559322032e-06,
+      "loss": 0.1885,
+      "step": 975
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.3722033898305083e-06,
+      "loss": 0.2006,
+      "step": 976
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.3715254237288133e-06,
+      "loss": 0.1305,
+      "step": 977
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.3708474576271186e-06,
+      "loss": 0.1185,
+      "step": 978
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.3701694915254236e-06,
+      "loss": 0.1554,
+      "step": 979
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.3694915254237287e-06,
+      "loss": 0.1233,
+      "step": 980
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.3688135593220338e-06,
+      "loss": 0.2261,
+      "step": 981
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.3681355932203388e-06,
+      "loss": 0.1937,
+      "step": 982
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.367457627118644e-06,
+      "loss": 0.2992,
+      "step": 983
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.3667796610169491e-06,
+      "loss": 0.099,
+      "step": 984
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.3661016949152542e-06,
+      "loss": 0.2058,
+      "step": 985
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.3654237288135592e-06,
+      "loss": 0.1922,
+      "step": 986
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.3647457627118643e-06,
+      "loss": 0.2177,
+      "step": 987
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.3640677966101696e-06,
+      "loss": 0.2719,
+      "step": 988
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.3633898305084746e-06,
+      "loss": 0.1461,
+      "step": 989
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.3627118644067797e-06,
+      "loss": 0.0626,
+      "step": 990
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.3620338983050847e-06,
+      "loss": 0.2347,
+      "step": 991
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.3613559322033898e-06,
+      "loss": 0.1569,
+      "step": 992
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.3606779661016948e-06,
+      "loss": 0.0765,
+      "step": 993
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 0.1408,
+      "step": 994
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.3593220338983052e-06,
+      "loss": 0.1158,
+      "step": 995
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.35864406779661e-06,
+      "loss": 0.1584,
+      "step": 996
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.357966101694915e-06,
+      "loss": 0.125,
+      "step": 997
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.3572881355932201e-06,
+      "loss": 0.2532,
+      "step": 998
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.3566101694915254e-06,
+      "loss": 0.2595,
+      "step": 999
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.3559322033898304e-06,
+      "loss": 0.2905,
+      "step": 1000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.3552542372881355e-06,
+      "loss": 0.2109,
+      "step": 1001
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.3545762711864406e-06,
+      "loss": 0.2787,
+      "step": 1002
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.3538983050847456e-06,
+      "loss": 0.1537,
+      "step": 1003
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.3532203389830509e-06,
+      "loss": 0.0606,
+      "step": 1004
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.352542372881356e-06,
+      "loss": 0.1591,
+      "step": 1005
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.351864406779661e-06,
+      "loss": 0.3794,
+      "step": 1006
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.351186440677966e-06,
+      "loss": 0.1112,
+      "step": 1007
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.350508474576271e-06,
+      "loss": 0.099,
+      "step": 1008
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.3498305084745761e-06,
+      "loss": 0.2185,
+      "step": 1009
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.3491525423728814e-06,
+      "loss": 0.1865,
+      "step": 1010
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.3484745762711865e-06,
+      "loss": 0.1684,
+      "step": 1011
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.3477966101694915e-06,
+      "loss": 0.1465,
+      "step": 1012
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.3471186440677966e-06,
+      "loss": 0.3698,
+      "step": 1013
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.3464406779661016e-06,
+      "loss": 0.0668,
+      "step": 1014
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.345762711864407e-06,
+      "loss": 0.0842,
+      "step": 1015
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.345084745762712e-06,
+      "loss": 0.058,
+      "step": 1016
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.3444067796610168e-06,
+      "loss": 0.2392,
+      "step": 1017
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.3437288135593219e-06,
+      "loss": 0.1301,
+      "step": 1018
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.343050847457627e-06,
+      "loss": 0.0994,
+      "step": 1019
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.342372881355932e-06,
+      "loss": 0.1115,
+      "step": 1020
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.3416949152542372e-06,
+      "loss": 0.2144,
+      "step": 1021
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.3410169491525423e-06,
+      "loss": 0.1689,
+      "step": 1022
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.3403389830508473e-06,
+      "loss": 0.0722,
+      "step": 1023
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.3396610169491524e-06,
+      "loss": 0.1058,
+      "step": 1024
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.3389830508474575e-06,
+      "loss": 0.2104,
+      "step": 1025
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.3383050847457627e-06,
+      "loss": 0.1704,
+      "step": 1026
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.3376271186440678e-06,
+      "loss": 0.3156,
+      "step": 1027
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.3369491525423728e-06,
+      "loss": 0.1713,
+      "step": 1028
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.3362711864406779e-06,
+      "loss": 0.0866,
+      "step": 1029
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.335593220338983e-06,
+      "loss": 0.1016,
+      "step": 1030
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.3349152542372882e-06,
+      "loss": 0.1027,
+      "step": 1031
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.3342372881355933e-06,
+      "loss": 0.236,
+      "step": 1032
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.3335593220338983e-06,
+      "loss": 0.1598,
+      "step": 1033
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.3328813559322034e-06,
+      "loss": 0.227,
+      "step": 1034
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.3322033898305084e-06,
+      "loss": 0.1154,
+      "step": 1035
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.3315254237288137e-06,
+      "loss": 0.1596,
+      "step": 1036
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.3308474576271185e-06,
+      "loss": 0.2688,
+      "step": 1037
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.3301694915254236e-06,
+      "loss": 0.1368,
+      "step": 1038
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.3294915254237287e-06,
+      "loss": 0.2205,
+      "step": 1039
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.3288135593220337e-06,
+      "loss": 0.2239,
+      "step": 1040
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.3281355932203388e-06,
+      "loss": 0.1772,
+      "step": 1041
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.327457627118644e-06,
+      "loss": 0.0688,
+      "step": 1042
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.326779661016949e-06,
+      "loss": 0.1931,
+      "step": 1043
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.3261016949152541e-06,
+      "loss": 0.049,
+      "step": 1044
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.3254237288135592e-06,
+      "loss": 0.2912,
+      "step": 1045
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.3247457627118643e-06,
+      "loss": 0.151,
+      "step": 1046
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.3240677966101695e-06,
+      "loss": 0.0761,
+      "step": 1047
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.3233898305084746e-06,
+      "loss": 0.1741,
+      "step": 1048
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.3227118644067796e-06,
+      "loss": 0.249,
+      "step": 1049
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.3220338983050847e-06,
+      "loss": 0.1217,
+      "step": 1050
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9391666666666667,
+      "eval_loss": 0.17867614328861237,
+      "eval_runtime": 2.1693,
+      "eval_samples_per_second": 553.168,
+      "eval_steps_per_second": 4.61,
+      "step": 1050
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.3213559322033897e-06,
+      "loss": 0.1033,
+      "step": 1051
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.320677966101695e-06,
+      "loss": 0.1672,
+      "step": 1052
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.32e-06,
+      "loss": 0.097,
+      "step": 1053
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3193220338983051e-06,
+      "loss": 0.1527,
+      "step": 1054
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.3186440677966102e-06,
+      "loss": 0.0696,
+      "step": 1055
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.3179661016949152e-06,
+      "loss": 0.0536,
+      "step": 1056
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.3172881355932203e-06,
+      "loss": 0.1639,
+      "step": 1057
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.3166101694915253e-06,
+      "loss": 0.2985,
+      "step": 1058
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.3159322033898304e-06,
+      "loss": 0.1313,
+      "step": 1059
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.3152542372881355e-06,
+      "loss": 0.1431,
+      "step": 1060
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.3145762711864405e-06,
+      "loss": 0.0903,
+      "step": 1061
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.3138983050847456e-06,
+      "loss": 0.0682,
+      "step": 1062
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.3132203389830508e-06,
+      "loss": 0.155,
+      "step": 1063
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.3125423728813559e-06,
+      "loss": 0.1242,
+      "step": 1064
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.311864406779661e-06,
+      "loss": 0.1489,
+      "step": 1065
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.311186440677966e-06,
+      "loss": 0.1878,
+      "step": 1066
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.310508474576271e-06,
+      "loss": 0.1566,
+      "step": 1067
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.3098305084745761e-06,
+      "loss": 0.0949,
+      "step": 1068
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.3091525423728814e-06,
+      "loss": 0.1652,
+      "step": 1069
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.3084745762711864e-06,
+      "loss": 0.2099,
+      "step": 1070
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.3077966101694915e-06,
+      "loss": 0.0429,
+      "step": 1071
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.3071186440677965e-06,
+      "loss": 0.1763,
+      "step": 1072
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.3064406779661016e-06,
+      "loss": 0.1139,
+      "step": 1073
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.3057627118644069e-06,
+      "loss": 0.2101,
+      "step": 1074
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.305084745762712e-06,
+      "loss": 0.1314,
+      "step": 1075
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.304406779661017e-06,
+      "loss": 0.0805,
+      "step": 1076
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.303728813559322e-06,
+      "loss": 0.0727,
+      "step": 1077
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.303050847457627e-06,
+      "loss": 0.0858,
+      "step": 1078
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.3023728813559321e-06,
+      "loss": 0.2719,
+      "step": 1079
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.3016949152542372e-06,
+      "loss": 0.1048,
+      "step": 1080
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3010169491525423e-06,
+      "loss": 0.0948,
+      "step": 1081
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.3003389830508473e-06,
+      "loss": 0.182,
+      "step": 1082
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.2996610169491524e-06,
+      "loss": 0.222,
+      "step": 1083
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.2989830508474574e-06,
+      "loss": 0.1353,
+      "step": 1084
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.2983050847457627e-06,
+      "loss": 0.0588,
+      "step": 1085
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.2976271186440677e-06,
+      "loss": 0.1803,
+      "step": 1086
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.2969491525423728e-06,
+      "loss": 0.27,
+      "step": 1087
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.2962711864406779e-06,
+      "loss": 0.1085,
+      "step": 1088
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.295593220338983e-06,
+      "loss": 0.062,
+      "step": 1089
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.2949152542372882e-06,
+      "loss": 0.0562,
+      "step": 1090
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.2942372881355932e-06,
+      "loss": 0.099,
+      "step": 1091
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.2935593220338983e-06,
+      "loss": 0.2152,
+      "step": 1092
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.2928813559322033e-06,
+      "loss": 0.3026,
+      "step": 1093
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.2922033898305084e-06,
+      "loss": 0.1544,
+      "step": 1094
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.2915254237288137e-06,
+      "loss": 0.1499,
+      "step": 1095
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.2908474576271187e-06,
+      "loss": 0.076,
+      "step": 1096
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.2901694915254238e-06,
+      "loss": 0.0523,
+      "step": 1097
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.2894915254237288e-06,
+      "loss": 0.1224,
+      "step": 1098
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.2888135593220339e-06,
+      "loss": 0.1076,
+      "step": 1099
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.2881355932203387e-06,
+      "loss": 0.1661,
+      "step": 1100
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.287457627118644e-06,
+      "loss": 0.102,
+      "step": 1101
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.286779661016949e-06,
+      "loss": 0.3861,
+      "step": 1102
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.286101694915254e-06,
+      "loss": 0.1388,
+      "step": 1103
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.2854237288135592e-06,
+      "loss": 0.0618,
+      "step": 1104
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.2847457627118642e-06,
+      "loss": 0.071,
+      "step": 1105
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.2840677966101695e-06,
+      "loss": 0.2051,
+      "step": 1106
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.2833898305084745e-06,
+      "loss": 0.0579,
+      "step": 1107
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.2827118644067796e-06,
+      "loss": 0.1756,
+      "step": 1108
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.2820338983050846e-06,
+      "loss": 0.2274,
+      "step": 1109
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.2813559322033897e-06,
+      "loss": 0.267,
+      "step": 1110
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.280677966101695e-06,
+      "loss": 0.0814,
+      "step": 1111
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.28e-06,
+      "loss": 0.1732,
+      "step": 1112
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.279322033898305e-06,
+      "loss": 0.0912,
+      "step": 1113
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.2786440677966101e-06,
+      "loss": 0.2504,
+      "step": 1114
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.2779661016949152e-06,
+      "loss": 0.2948,
+      "step": 1115
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.2772881355932202e-06,
+      "loss": 0.1412,
+      "step": 1116
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2766101694915255e-06,
+      "loss": 0.2982,
+      "step": 1117
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.2759322033898306e-06,
+      "loss": 0.0817,
+      "step": 1118
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2752542372881356e-06,
+      "loss": 0.1372,
+      "step": 1119
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2745762711864405e-06,
+      "loss": 0.2141,
+      "step": 1120
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2738983050847455e-06,
+      "loss": 0.2414,
+      "step": 1121
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2732203389830508e-06,
+      "loss": 0.0736,
+      "step": 1122
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.2725423728813558e-06,
+      "loss": 0.0489,
+      "step": 1123
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.271864406779661e-06,
+      "loss": 0.2414,
+      "step": 1124
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.271186440677966e-06,
+      "loss": 0.0823,
+      "step": 1125
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.270508474576271e-06,
+      "loss": 0.1825,
+      "step": 1126
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2698305084745763e-06,
+      "loss": 0.0772,
+      "step": 1127
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2691525423728813e-06,
+      "loss": 0.0974,
+      "step": 1128
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2684745762711864e-06,
+      "loss": 0.1226,
+      "step": 1129
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2677966101694914e-06,
+      "loss": 0.115,
+      "step": 1130
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2671186440677965e-06,
+      "loss": 0.0753,
+      "step": 1131
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2664406779661016e-06,
+      "loss": 0.0742,
+      "step": 1132
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2657627118644068e-06,
+      "loss": 0.0477,
+      "step": 1133
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.2650847457627119e-06,
+      "loss": 0.0739,
+      "step": 1134
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.264406779661017e-06,
+      "loss": 0.0984,
+      "step": 1135
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.263728813559322e-06,
+      "loss": 0.1167,
+      "step": 1136
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.263050847457627e-06,
+      "loss": 0.1581,
+      "step": 1137
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2623728813559323e-06,
+      "loss": 0.1199,
+      "step": 1138
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2616949152542374e-06,
+      "loss": 0.0638,
+      "step": 1139
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.2610169491525424e-06,
+      "loss": 0.0984,
+      "step": 1140
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.2603389830508473e-06,
+      "loss": 0.1782,
+      "step": 1141
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.2596610169491523e-06,
+      "loss": 0.1572,
+      "step": 1142
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.2589830508474576e-06,
+      "loss": 0.2107,
+      "step": 1143
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.2583050847457626e-06,
+      "loss": 0.1769,
+      "step": 1144
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.2576271186440677e-06,
+      "loss": 0.0882,
+      "step": 1145
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.2569491525423728e-06,
+      "loss": 0.1487,
+      "step": 1146
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.2562711864406778e-06,
+      "loss": 0.0576,
+      "step": 1147
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.2555932203389829e-06,
+      "loss": 0.049,
+      "step": 1148
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.2549152542372881e-06,
+      "loss": 0.1626,
+      "step": 1149
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.2542372881355932e-06,
+      "loss": 0.1257,
+      "step": 1150
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 1.2535593220338982e-06,
+      "loss": 0.0854,
+      "step": 1151
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.2528813559322033e-06,
+      "loss": 0.097,
+      "step": 1152
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.2522033898305084e-06,
+      "loss": 0.1523,
+      "step": 1153
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.2515254237288136e-06,
+      "loss": 0.184,
+      "step": 1154
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.2508474576271187e-06,
+      "loss": 0.0491,
+      "step": 1155
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.2501694915254237e-06,
+      "loss": 0.3348,
+      "step": 1156
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.2494915254237288e-06,
+      "loss": 0.0907,
+      "step": 1157
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.2488135593220338e-06,
+      "loss": 0.2645,
+      "step": 1158
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.2481355932203391e-06,
+      "loss": 0.1606,
+      "step": 1159
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.2474576271186442e-06,
+      "loss": 0.1083,
+      "step": 1160
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.2467796610169492e-06,
+      "loss": 0.1732,
+      "step": 1161
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.246101694915254e-06,
+      "loss": 0.1342,
+      "step": 1162
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.2454237288135591e-06,
+      "loss": 0.0978,
+      "step": 1163
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.2447457627118642e-06,
+      "loss": 0.1686,
+      "step": 1164
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.2440677966101694e-06,
+      "loss": 0.0807,
+      "step": 1165
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.2433898305084745e-06,
+      "loss": 0.0859,
+      "step": 1166
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.2427118644067796e-06,
+      "loss": 0.0708,
+      "step": 1167
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.2420338983050846e-06,
+      "loss": 0.1128,
+      "step": 1168
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.2413559322033897e-06,
+      "loss": 0.209,
+      "step": 1169
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.240677966101695e-06,
+      "loss": 0.0853,
+      "step": 1170
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.24e-06,
+      "loss": 0.128,
+      "step": 1171
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.239322033898305e-06,
+      "loss": 0.0852,
+      "step": 1172
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.23864406779661e-06,
+      "loss": 0.0496,
+      "step": 1173
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.2379661016949152e-06,
+      "loss": 0.1112,
+      "step": 1174
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.2372881355932204e-06,
+      "loss": 0.083,
+      "step": 1175
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.2366101694915255e-06,
+      "loss": 0.086,
+      "step": 1176
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.2359322033898305e-06,
+      "loss": 0.1494,
+      "step": 1177
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.2352542372881356e-06,
+      "loss": 0.0834,
+      "step": 1178
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.2345762711864406e-06,
+      "loss": 0.228,
+      "step": 1179
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.2338983050847457e-06,
+      "loss": 0.237,
+      "step": 1180
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.233220338983051e-06,
+      "loss": 0.2149,
+      "step": 1181
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.232542372881356e-06,
+      "loss": 0.0522,
+      "step": 1182
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.2318644067796609e-06,
+      "loss": 0.2585,
+      "step": 1183
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.231186440677966e-06,
+      "loss": 0.0892,
+      "step": 1184
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.230508474576271e-06,
+      "loss": 0.1251,
+      "step": 1185
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.2298305084745762e-06,
+      "loss": 0.0866,
+      "step": 1186
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.2291525423728813e-06,
+      "loss": 0.2315,
+      "step": 1187
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.2284745762711864e-06,
+      "loss": 0.0673,
+      "step": 1188
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.2277966101694914e-06,
+      "loss": 0.1812,
+      "step": 1189
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.2271186440677965e-06,
+      "loss": 0.0626,
+      "step": 1190
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.2264406779661015e-06,
+      "loss": 0.1914,
+      "step": 1191
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.2257627118644068e-06,
+      "loss": 0.084,
+      "step": 1192
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.2250847457627118e-06,
+      "loss": 0.1848,
+      "step": 1193
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.224406779661017e-06,
+      "loss": 0.0629,
+      "step": 1194
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.223728813559322e-06,
+      "loss": 0.053,
+      "step": 1195
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.223050847457627e-06,
+      "loss": 0.1371,
+      "step": 1196
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.2223728813559323e-06,
+      "loss": 0.1618,
+      "step": 1197
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.2216949152542373e-06,
+      "loss": 0.0804,
+      "step": 1198
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.2210169491525424e-06,
+      "loss": 0.1701,
+      "step": 1199
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.2203389830508474e-06,
+      "loss": 0.2239,
+      "step": 1200
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.9408333333333333,
+      "eval_loss": 0.16648565232753754,
+      "eval_runtime": 2.1629,
+      "eval_samples_per_second": 554.823,
+      "eval_steps_per_second": 4.624,
+      "step": 1200
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2196610169491525e-06,
+      "loss": 0.0578,
+      "step": 1201
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 1.2189830508474578e-06,
+      "loss": 0.2605,
+      "step": 1202
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.2183050847457626e-06,
+      "loss": 0.0664,
+      "step": 1203
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2176271186440677e-06,
+      "loss": 0.0592,
+      "step": 1204
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.2169491525423727e-06,
+      "loss": 0.2028,
+      "step": 1205
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.2162711864406778e-06,
+      "loss": 0.05,
+      "step": 1206
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.2155932203389828e-06,
+      "loss": 0.0889,
+      "step": 1207
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.214915254237288e-06,
+      "loss": 0.059,
+      "step": 1208
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.2142372881355931e-06,
+      "loss": 0.1146,
+      "step": 1209
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.2135593220338982e-06,
+      "loss": 0.0448,
+      "step": 1210
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.2128813559322033e-06,
+      "loss": 0.1012,
+      "step": 1211
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.2122033898305083e-06,
+      "loss": 0.1311,
+      "step": 1212
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.2115254237288136e-06,
+      "loss": 0.1254,
+      "step": 1213
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.2108474576271186e-06,
+      "loss": 0.2726,
+      "step": 1214
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.2101694915254237e-06,
+      "loss": 0.0499,
+      "step": 1215
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.2094915254237287e-06,
+      "loss": 0.1829,
+      "step": 1216
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.2088135593220338e-06,
+      "loss": 0.0689,
+      "step": 1217
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.208135593220339e-06,
+      "loss": 0.1835,
+      "step": 1218
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.2074576271186441e-06,
+      "loss": 0.1104,
+      "step": 1219
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.2067796610169492e-06,
+      "loss": 0.0912,
+      "step": 1220
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.2061016949152542e-06,
+      "loss": 0.1757,
+      "step": 1221
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.2054237288135593e-06,
+      "loss": 0.0958,
+      "step": 1222
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.2047457627118646e-06,
+      "loss": 0.1525,
+      "step": 1223
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.2040677966101694e-06,
+      "loss": 0.069,
+      "step": 1224
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.2033898305084745e-06,
+      "loss": 0.1472,
+      "step": 1225
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.2027118644067795e-06,
+      "loss": 0.1457,
+      "step": 1226
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.2020338983050846e-06,
+      "loss": 0.0579,
+      "step": 1227
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.2013559322033896e-06,
+      "loss": 0.049,
+      "step": 1228
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.2006779661016949e-06,
+      "loss": 0.2624,
+      "step": 1229
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.2e-06,
+      "loss": 0.2189,
+      "step": 1230
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.199322033898305e-06,
+      "loss": 0.0755,
+      "step": 1231
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.19864406779661e-06,
+      "loss": 0.1616,
+      "step": 1232
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.1979661016949151e-06,
+      "loss": 0.0491,
+      "step": 1233
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.1972881355932204e-06,
+      "loss": 0.1849,
+      "step": 1234
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.1966101694915254e-06,
+      "loss": 0.0913,
+      "step": 1235
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.1959322033898305e-06,
+      "loss": 0.12,
+      "step": 1236
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.1952542372881355e-06,
+      "loss": 0.0723,
+      "step": 1237
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.1945762711864406e-06,
+      "loss": 0.0674,
+      "step": 1238
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.1938983050847457e-06,
+      "loss": 0.0437,
+      "step": 1239
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.193220338983051e-06,
+      "loss": 0.0451,
+      "step": 1240
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.192542372881356e-06,
+      "loss": 0.0546,
+      "step": 1241
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.191864406779661e-06,
+      "loss": 0.1625,
+      "step": 1242
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.191186440677966e-06,
+      "loss": 0.0842,
+      "step": 1243
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.1905084745762711e-06,
+      "loss": 0.0743,
+      "step": 1244
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.1898305084745762e-06,
+      "loss": 0.1086,
+      "step": 1245
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.1891525423728813e-06,
+      "loss": 0.0691,
+      "step": 1246
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.1884745762711863e-06,
+      "loss": 0.0501,
+      "step": 1247
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.1877966101694914e-06,
+      "loss": 0.1493,
+      "step": 1248
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.1871186440677964e-06,
+      "loss": 0.1773,
+      "step": 1249
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.1864406779661017e-06,
+      "loss": 0.0371,
+      "step": 1250
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.1857627118644067e-06,
+      "loss": 0.0545,
+      "step": 1251
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.1850847457627118e-06,
+      "loss": 0.2342,
+      "step": 1252
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.1844067796610169e-06,
+      "loss": 0.0701,
+      "step": 1253
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.183728813559322e-06,
+      "loss": 0.0812,
+      "step": 1254
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.183050847457627e-06,
+      "loss": 0.0415,
+      "step": 1255
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.1823728813559322e-06,
+      "loss": 0.2034,
+      "step": 1256
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.1816949152542373e-06,
+      "loss": 0.1265,
+      "step": 1257
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.1810169491525423e-06,
+      "loss": 0.0876,
+      "step": 1258
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.1803389830508474e-06,
+      "loss": 0.111,
+      "step": 1259
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.1796610169491525e-06,
+      "loss": 0.0849,
+      "step": 1260
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.1789830508474577e-06,
+      "loss": 0.2333,
+      "step": 1261
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.1783050847457628e-06,
+      "loss": 0.0608,
+      "step": 1262
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.1776271186440678e-06,
+      "loss": 0.1015,
+      "step": 1263
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.1769491525423729e-06,
+      "loss": 0.0737,
+      "step": 1264
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.176271186440678e-06,
+      "loss": 0.1042,
+      "step": 1265
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.175593220338983e-06,
+      "loss": 0.0406,
+      "step": 1266
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.174915254237288e-06,
+      "loss": 0.1145,
+      "step": 1267
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.1742372881355931e-06,
+      "loss": 0.1783,
+      "step": 1268
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.1735593220338982e-06,
+      "loss": 0.0428,
+      "step": 1269
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.1728813559322032e-06,
+      "loss": 0.128,
+      "step": 1270
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.1722033898305083e-06,
+      "loss": 0.0431,
+      "step": 1271
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.1715254237288135e-06,
+      "loss": 0.3058,
+      "step": 1272
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.1708474576271186e-06,
+      "loss": 0.0724,
+      "step": 1273
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.1701694915254237e-06,
+      "loss": 0.1464,
+      "step": 1274
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.1694915254237287e-06,
+      "loss": 0.1469,
+      "step": 1275
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.1688135593220338e-06,
+      "loss": 0.2153,
+      "step": 1276
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.168135593220339e-06,
+      "loss": 0.1105,
+      "step": 1277
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.167457627118644e-06,
+      "loss": 0.0811,
+      "step": 1278
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.1667796610169491e-06,
+      "loss": 0.1187,
+      "step": 1279
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.1661016949152542e-06,
+      "loss": 0.1969,
+      "step": 1280
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.1654237288135593e-06,
+      "loss": 0.1847,
+      "step": 1281
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.1647457627118645e-06,
+      "loss": 0.1981,
+      "step": 1282
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.1640677966101696e-06,
+      "loss": 0.0518,
+      "step": 1283
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.1633898305084746e-06,
+      "loss": 0.1896,
+      "step": 1284
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.1627118644067797e-06,
+      "loss": 0.0507,
+      "step": 1285
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.1620338983050847e-06,
+      "loss": 0.0714,
+      "step": 1286
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.1613559322033896e-06,
+      "loss": 0.1415,
+      "step": 1287
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.1606779661016948e-06,
+      "loss": 0.0575,
+      "step": 1288
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.16e-06,
+      "loss": 0.0807,
+      "step": 1289
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.159322033898305e-06,
+      "loss": 0.0522,
+      "step": 1290
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.15864406779661e-06,
+      "loss": 0.3148,
+      "step": 1291
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.157966101694915e-06,
+      "loss": 0.1748,
+      "step": 1292
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 1.1572881355932203e-06,
+      "loss": 0.0825,
+      "step": 1293
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.1566101694915254e-06,
+      "loss": 0.2799,
+      "step": 1294
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 1.1559322033898304e-06,
+      "loss": 0.1794,
+      "step": 1295
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 1.1552542372881355e-06,
+      "loss": 0.1373,
+      "step": 1296
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.1545762711864406e-06,
+      "loss": 0.0965,
+      "step": 1297
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 1.1538983050847458e-06,
+      "loss": 0.2499,
+      "step": 1298
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 1.1532203389830509e-06,
+      "loss": 0.0718,
+      "step": 1299
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.152542372881356e-06,
+      "loss": 0.226,
+      "step": 1300
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 1.151864406779661e-06,
+      "loss": 0.0687,
+      "step": 1301
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 1.151186440677966e-06,
+      "loss": 0.1217,
+      "step": 1302
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.150508474576271e-06,
+      "loss": 0.0572,
+      "step": 1303
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 1.1498305084745764e-06,
+      "loss": 0.3177,
+      "step": 1304
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 1.1491525423728814e-06,
+      "loss": 0.0489,
+      "step": 1305
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.1484745762711865e-06,
+      "loss": 0.0623,
+      "step": 1306
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 1.1477966101694913e-06,
+      "loss": 0.0568,
+      "step": 1307
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 1.1471186440677964e-06,
+      "loss": 0.131,
+      "step": 1308
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.1464406779661016e-06,
+      "loss": 0.0687,
+      "step": 1309
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 1.1457627118644067e-06,
+      "loss": 0.1849,
+      "step": 1310
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 1.1450847457627118e-06,
+      "loss": 0.0848,
+      "step": 1311
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.1444067796610168e-06,
+      "loss": 0.1458,
+      "step": 1312
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 1.1437288135593219e-06,
+      "loss": 0.1349,
+      "step": 1313
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 1.1430508474576271e-06,
+      "loss": 0.0331,
+      "step": 1314
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.1423728813559322e-06,
+      "loss": 0.154,
+      "step": 1315
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 1.1416949152542372e-06,
+      "loss": 0.0355,
+      "step": 1316
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 1.1410169491525423e-06,
+      "loss": 0.0741,
+      "step": 1317
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.1403389830508474e-06,
+      "loss": 0.0332,
+      "step": 1318
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 1.1396610169491524e-06,
+      "loss": 0.0633,
+      "step": 1319
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 1.1389830508474577e-06,
+      "loss": 0.2599,
+      "step": 1320
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.1383050847457627e-06,
+      "loss": 0.1882,
+      "step": 1321
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 1.1376271186440678e-06,
+      "loss": 0.1333,
+      "step": 1322
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 1.1369491525423728e-06,
+      "loss": 0.1178,
+      "step": 1323
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.136271186440678e-06,
+      "loss": 0.2969,
+      "step": 1324
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 1.1355932203389832e-06,
+      "loss": 0.2684,
+      "step": 1325
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 1.1349152542372882e-06,
+      "loss": 0.0795,
+      "step": 1326
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.1342372881355933e-06,
+      "loss": 0.1673,
+      "step": 1327
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 1.1335593220338981e-06,
+      "loss": 0.1067,
+      "step": 1328
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 1.1328813559322032e-06,
+      "loss": 0.0918,
+      "step": 1329
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.1322033898305082e-06,
+      "loss": 0.1004,
+      "step": 1330
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 1.1315254237288135e-06,
+      "loss": 0.1313,
+      "step": 1331
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 1.1308474576271186e-06,
+      "loss": 0.055,
+      "step": 1332
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.1301694915254236e-06,
+      "loss": 0.1359,
+      "step": 1333
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 1.1294915254237287e-06,
+      "loss": 0.0309,
+      "step": 1334
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 1.1288135593220337e-06,
+      "loss": 0.0437,
+      "step": 1335
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.128135593220339e-06,
+      "loss": 0.1121,
+      "step": 1336
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 1.127457627118644e-06,
+      "loss": 0.0813,
+      "step": 1337
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 1.126779661016949e-06,
+      "loss": 0.0731,
+      "step": 1338
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.1261016949152542e-06,
+      "loss": 0.2019,
+      "step": 1339
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 1.1254237288135592e-06,
+      "loss": 0.4634,
+      "step": 1340
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 1.1247457627118645e-06,
+      "loss": 0.0817,
+      "step": 1341
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.1240677966101695e-06,
+      "loss": 0.1337,
+      "step": 1342
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 1.1233898305084746e-06,
+      "loss": 0.0505,
+      "step": 1343
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 1.1227118644067796e-06,
+      "loss": 0.1064,
+      "step": 1344
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.1220338983050847e-06,
+      "loss": 0.102,
+      "step": 1345
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 1.12135593220339e-06,
+      "loss": 0.0968,
+      "step": 1346
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 1.120677966101695e-06,
+      "loss": 0.0588,
+      "step": 1347
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.12e-06,
+      "loss": 0.1623,
+      "step": 1348
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 1.119322033898305e-06,
+      "loss": 0.0739,
+      "step": 1349
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 1.11864406779661e-06,
+      "loss": 0.1104,
+      "step": 1350
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9466666666666667,
+      "eval_loss": 0.15664003789424896,
+      "eval_runtime": 2.163,
+      "eval_samples_per_second": 554.797,
+      "eval_steps_per_second": 4.623,
+      "step": 1350
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.117966101694915e-06,
+      "loss": 0.1003,
+      "step": 1351
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 1.1172881355932203e-06,
+      "loss": 0.0391,
+      "step": 1352
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 1.1166101694915254e-06,
+      "loss": 0.1251,
+      "step": 1353
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1159322033898304e-06,
+      "loss": 0.1441,
+      "step": 1354
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 1.1152542372881355e-06,
+      "loss": 0.0324,
+      "step": 1355
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 1.1145762711864405e-06,
+      "loss": 0.3081,
+      "step": 1356
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.1138983050847458e-06,
+      "loss": 0.0455,
+      "step": 1357
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 1.1132203389830508e-06,
+      "loss": 0.0934,
+      "step": 1358
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 1.112542372881356e-06,
+      "loss": 0.0865,
+      "step": 1359
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.111864406779661e-06,
+      "loss": 0.2466,
+      "step": 1360
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 1.111186440677966e-06,
+      "loss": 0.0347,
+      "step": 1361
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 1.1105084745762713e-06,
+      "loss": 0.3363,
+      "step": 1362
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.1098305084745763e-06,
+      "loss": 0.0317,
+      "step": 1363
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 1.1091525423728814e-06,
+      "loss": 0.0699,
+      "step": 1364
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 1.1084745762711864e-06,
+      "loss": 0.1004,
+      "step": 1365
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.1077966101694915e-06,
+      "loss": 0.1698,
+      "step": 1366
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 1.1071186440677966e-06,
+      "loss": 0.2511,
+      "step": 1367
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 1.1064406779661018e-06,
+      "loss": 0.0314,
+      "step": 1368
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.1057627118644069e-06,
+      "loss": 0.2047,
+      "step": 1369
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 1.1050847457627117e-06,
+      "loss": 0.0611,
+      "step": 1370
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 1.1044067796610168e-06,
+      "loss": 0.0405,
+      "step": 1371
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.1037288135593218e-06,
+      "loss": 0.0874,
+      "step": 1372
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 1.103050847457627e-06,
+      "loss": 0.0307,
+      "step": 1373
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 1.1023728813559322e-06,
+      "loss": 0.1837,
+      "step": 1374
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.1016949152542372e-06,
+      "loss": 0.0504,
+      "step": 1375
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 1.1010169491525423e-06,
+      "loss": 0.0599,
+      "step": 1376
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 1.1003389830508473e-06,
+      "loss": 0.2235,
+      "step": 1377
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.0996610169491524e-06,
+      "loss": 0.0907,
+      "step": 1378
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 1.0989830508474576e-06,
+      "loss": 0.0383,
+      "step": 1379
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 1.0983050847457627e-06,
+      "loss": 0.0758,
+      "step": 1380
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.0976271186440677e-06,
+      "loss": 0.1179,
+      "step": 1381
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.0969491525423728e-06,
+      "loss": 0.185,
+      "step": 1382
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 1.0962711864406779e-06,
+      "loss": 0.0848,
+      "step": 1383
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.0955932203389831e-06,
+      "loss": 0.0453,
+      "step": 1384
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.0949152542372882e-06,
+      "loss": 0.2184,
+      "step": 1385
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 1.0942372881355932e-06,
+      "loss": 0.0544,
+      "step": 1386
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.0935593220338983e-06,
+      "loss": 0.1797,
+      "step": 1387
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 1.0928813559322033e-06,
+      "loss": 0.1055,
+      "step": 1388
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.0922033898305086e-06,
+      "loss": 0.085,
+      "step": 1389
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.0915254237288135e-06,
+      "loss": 0.0641,
+      "step": 1390
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 1.0908474576271185e-06,
+      "loss": 0.0557,
+      "step": 1391
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.0901694915254236e-06,
+      "loss": 0.151,
+      "step": 1392
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.0894915254237286e-06,
+      "loss": 0.12,
+      "step": 1393
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 1.0888135593220337e-06,
+      "loss": 0.1,
+      "step": 1394
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.088135593220339e-06,
+      "loss": 0.1123,
+      "step": 1395
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.087457627118644e-06,
+      "loss": 0.0301,
+      "step": 1396
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 1.086779661016949e-06,
+      "loss": 0.0313,
+      "step": 1397
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.0861016949152541e-06,
+      "loss": 0.2272,
+      "step": 1398
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.0854237288135592e-06,
+      "loss": 0.2015,
+      "step": 1399
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 1.0847457627118644e-06,
+      "loss": 0.1647,
+      "step": 1400
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 1.0840677966101695e-06,
+      "loss": 0.1699,
+      "step": 1401
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.0833898305084745e-06,
+      "loss": 0.1895,
+      "step": 1402
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.0827118644067796e-06,
+      "loss": 0.0597,
+      "step": 1403
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 1.0820338983050847e-06,
+      "loss": 0.0582,
+      "step": 1404
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.08135593220339e-06,
+      "loss": 0.1196,
+      "step": 1405
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.080677966101695e-06,
+      "loss": 0.1069,
+      "step": 1406
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 1.08e-06,
+      "loss": 0.1025,
+      "step": 1407
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.079322033898305e-06,
+      "loss": 0.1306,
+      "step": 1408
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.0786440677966101e-06,
+      "loss": 0.0443,
+      "step": 1409
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.0779661016949154e-06,
+      "loss": 0.2391,
+      "step": 1410
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.0772881355932203e-06,
+      "loss": 0.0477,
+      "step": 1411
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.0766101694915253e-06,
+      "loss": 0.058,
+      "step": 1412
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.0759322033898304e-06,
+      "loss": 0.0741,
+      "step": 1413
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.0752542372881354e-06,
+      "loss": 0.1817,
+      "step": 1414
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.0745762711864405e-06,
+      "loss": 0.0343,
+      "step": 1415
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.0738983050847457e-06,
+      "loss": 0.1451,
+      "step": 1416
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.0732203389830508e-06,
+      "loss": 0.1805,
+      "step": 1417
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.0725423728813559e-06,
+      "loss": 0.079,
+      "step": 1418
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.071864406779661e-06,
+      "loss": 0.222,
+      "step": 1419
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.071186440677966e-06,
+      "loss": 0.0854,
+      "step": 1420
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.0705084745762712e-06,
+      "loss": 0.0296,
+      "step": 1421
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.0698305084745763e-06,
+      "loss": 0.1856,
+      "step": 1422
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.0691525423728813e-06,
+      "loss": 0.0282,
+      "step": 1423
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.0684745762711864e-06,
+      "loss": 0.2331,
+      "step": 1424
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.0677966101694915e-06,
+      "loss": 0.0268,
+      "step": 1425
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.0671186440677965e-06,
+      "loss": 0.0301,
+      "step": 1426
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.0664406779661018e-06,
+      "loss": 0.0385,
+      "step": 1427
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.0657627118644068e-06,
+      "loss": 0.2385,
+      "step": 1428
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.0650847457627119e-06,
+      "loss": 0.1927,
+      "step": 1429
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.064406779661017e-06,
+      "loss": 0.2379,
+      "step": 1430
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.063728813559322e-06,
+      "loss": 0.0843,
+      "step": 1431
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.063050847457627e-06,
+      "loss": 0.1116,
+      "step": 1432
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.0623728813559321e-06,
+      "loss": 0.2229,
+      "step": 1433
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.0616949152542372e-06,
+      "loss": 0.2296,
+      "step": 1434
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.0610169491525422e-06,
+      "loss": 0.1429,
+      "step": 1435
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.0603389830508473e-06,
+      "loss": 0.0599,
+      "step": 1436
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 1.0596610169491525e-06,
+      "loss": 0.0424,
+      "step": 1437
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.0589830508474576e-06,
+      "loss": 0.1371,
+      "step": 1438
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.0583050847457627e-06,
+      "loss": 0.1929,
+      "step": 1439
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 1.0576271186440677e-06,
+      "loss": 0.1231,
+      "step": 1440
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.0569491525423728e-06,
+      "loss": 0.1825,
+      "step": 1441
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.0562711864406778e-06,
+      "loss": 0.0358,
+      "step": 1442
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.055593220338983e-06,
+      "loss": 0.052,
+      "step": 1443
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.0549152542372881e-06,
+      "loss": 0.1638,
+      "step": 1444
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.0542372881355932e-06,
+      "loss": 0.0853,
+      "step": 1445
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.0535593220338983e-06,
+      "loss": 0.2089,
+      "step": 1446
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0528813559322033e-06,
+      "loss": 0.0909,
+      "step": 1447
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.0522033898305086e-06,
+      "loss": 0.0385,
+      "step": 1448
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.0515254237288136e-06,
+      "loss": 0.0395,
+      "step": 1449
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0508474576271187e-06,
+      "loss": 0.0251,
+      "step": 1450
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.0501694915254237e-06,
+      "loss": 0.0471,
+      "step": 1451
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.0494915254237288e-06,
+      "loss": 0.0989,
+      "step": 1452
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.0488135593220339e-06,
+      "loss": 0.0657,
+      "step": 1453
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.048135593220339e-06,
+      "loss": 0.0321,
+      "step": 1454
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.047457627118644e-06,
+      "loss": 0.0272,
+      "step": 1455
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.046779661016949e-06,
+      "loss": 0.131,
+      "step": 1456
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.046101694915254e-06,
+      "loss": 0.1074,
+      "step": 1457
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.0454237288135591e-06,
+      "loss": 0.0996,
+      "step": 1458
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.0447457627118644e-06,
+      "loss": 0.0533,
+      "step": 1459
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.0440677966101695e-06,
+      "loss": 0.036,
+      "step": 1460
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.0433898305084745e-06,
+      "loss": 0.1837,
+      "step": 1461
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.0427118644067796e-06,
+      "loss": 0.0418,
+      "step": 1462
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.0420338983050846e-06,
+      "loss": 0.0848,
+      "step": 1463
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.0413559322033899e-06,
+      "loss": 0.1811,
+      "step": 1464
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.040677966101695e-06,
+      "loss": 0.1128,
+      "step": 1465
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.04e-06,
+      "loss": 0.1276,
+      "step": 1466
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.039322033898305e-06,
+      "loss": 0.0564,
+      "step": 1467
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0386440677966101e-06,
+      "loss": 0.0403,
+      "step": 1468
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.0379661016949154e-06,
+      "loss": 0.1912,
+      "step": 1469
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.0372881355932204e-06,
+      "loss": 0.0548,
+      "step": 1470
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.0366101694915255e-06,
+      "loss": 0.043,
+      "step": 1471
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.0359322033898305e-06,
+      "loss": 0.0497,
+      "step": 1472
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.0352542372881356e-06,
+      "loss": 0.1063,
+      "step": 1473
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.0345762711864404e-06,
+      "loss": 0.0593,
+      "step": 1474
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.0338983050847457e-06,
+      "loss": 0.1648,
+      "step": 1475
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.0332203389830508e-06,
+      "loss": 0.033,
+      "step": 1476
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.0325423728813558e-06,
+      "loss": 0.0263,
+      "step": 1477
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.0318644067796609e-06,
+      "loss": 0.0807,
+      "step": 1478
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.031186440677966e-06,
+      "loss": 0.0287,
+      "step": 1479
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.0305084745762712e-06,
+      "loss": 0.0417,
+      "step": 1480
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 1.0298305084745762e-06,
+      "loss": 0.1855,
+      "step": 1481
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 1.0291525423728813e-06,
+      "loss": 0.0682,
+      "step": 1482
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.0284745762711864e-06,
+      "loss": 0.1175,
+      "step": 1483
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.0277966101694914e-06,
+      "loss": 0.1745,
+      "step": 1484
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 1.0271186440677967e-06,
+      "loss": 0.0323,
+      "step": 1485
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.0264406779661017e-06,
+      "loss": 0.0698,
+      "step": 1486
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.0257627118644068e-06,
+      "loss": 0.118,
+      "step": 1487
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 1.0250847457627118e-06,
+      "loss": 0.0233,
+      "step": 1488
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.024406779661017e-06,
+      "loss": 0.0805,
+      "step": 1489
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.023728813559322e-06,
+      "loss": 0.1352,
+      "step": 1490
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.0230508474576272e-06,
+      "loss": 0.0255,
+      "step": 1491
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.0223728813559323e-06,
+      "loss": 0.0864,
+      "step": 1492
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.0216949152542373e-06,
+      "loss": 0.0446,
+      "step": 1493
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.0210169491525422e-06,
+      "loss": 0.1036,
+      "step": 1494
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0203389830508472e-06,
+      "loss": 0.0254,
+      "step": 1495
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.0196610169491525e-06,
+      "loss": 0.2954,
+      "step": 1496
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.0189830508474576e-06,
+      "loss": 0.2212,
+      "step": 1497
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.0183050847457626e-06,
+      "loss": 0.0817,
+      "step": 1498
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.0176271186440677e-06,
+      "loss": 0.0306,
+      "step": 1499
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.0169491525423727e-06,
+      "loss": 0.1143,
+      "step": 1500
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.945,
+      "eval_loss": 0.15177254378795624,
+      "eval_runtime": 2.1691,
+      "eval_samples_per_second": 553.236,
+      "eval_steps_per_second": 4.61,
+      "step": 1500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.016271186440678e-06,
+      "loss": 0.0557,
+      "step": 1501
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 1.015593220338983e-06,
+      "loss": 0.1567,
+      "step": 1502
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.014915254237288e-06,
+      "loss": 0.0305,
+      "step": 1503
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0142372881355932e-06,
+      "loss": 0.1316,
+      "step": 1504
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 1.0135593220338982e-06,
+      "loss": 0.1483,
+      "step": 1505
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.0128813559322033e-06,
+      "loss": 0.1625,
+      "step": 1506
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.0122033898305085e-06,
+      "loss": 0.0724,
+      "step": 1507
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 1.0115254237288136e-06,
+      "loss": 0.0909,
+      "step": 1508
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 1.0108474576271186e-06,
+      "loss": 0.035,
+      "step": 1509
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.0101694915254237e-06,
+      "loss": 0.0409,
+      "step": 1510
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.0094915254237288e-06,
+      "loss": 0.0768,
+      "step": 1511
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 1.008813559322034e-06,
+      "loss": 0.0784,
+      "step": 1512
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.008135593220339e-06,
+      "loss": 0.2329,
+      "step": 1513
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.0074576271186441e-06,
+      "loss": 0.0417,
+      "step": 1514
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 1.006779661016949e-06,
+      "loss": 0.1155,
+      "step": 1515
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.006101694915254e-06,
+      "loss": 0.0537,
+      "step": 1516
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.005423728813559e-06,
+      "loss": 0.0658,
+      "step": 1517
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 1.0047457627118644e-06,
+      "loss": 0.0241,
+      "step": 1518
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.0040677966101694e-06,
+      "loss": 0.063,
+      "step": 1519
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.0033898305084745e-06,
+      "loss": 0.0334,
+      "step": 1520
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 1.0027118644067795e-06,
+      "loss": 0.0531,
+      "step": 1521
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.0020338983050846e-06,
+      "loss": 0.075,
+      "step": 1522
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 1.0013559322033898e-06,
+      "loss": 0.1308,
+      "step": 1523
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.000677966101695e-06,
+      "loss": 0.263,
+      "step": 1524
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 1e-06,
+      "loss": 0.0646,
+      "step": 1525
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 9.99322033898305e-07,
+      "loss": 0.1697,
+      "step": 1526
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 9.9864406779661e-07,
+      "loss": 0.1115,
+      "step": 1527
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 9.979661016949153e-07,
+      "loss": 0.0688,
+      "step": 1528
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 9.972881355932204e-07,
+      "loss": 0.0401,
+      "step": 1529
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 9.966101694915254e-07,
+      "loss": 0.0223,
+      "step": 1530
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 9.959322033898305e-07,
+      "loss": 0.0776,
+      "step": 1531
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 9.952542372881356e-07,
+      "loss": 0.0407,
+      "step": 1532
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 9.945762711864406e-07,
+      "loss": 0.0726,
+      "step": 1533
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 9.938983050847457e-07,
+      "loss": 0.055,
+      "step": 1534
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 9.932203389830507e-07,
+      "loss": 0.0968,
+      "step": 1535
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 9.92542372881356e-07,
+      "loss": 0.0592,
+      "step": 1536
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 9.91864406779661e-07,
+      "loss": 0.1778,
+      "step": 1537
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 9.91186440677966e-07,
+      "loss": 0.1782,
+      "step": 1538
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 9.905084745762712e-07,
+      "loss": 0.0839,
+      "step": 1539
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 9.898305084745762e-07,
+      "loss": 0.0448,
+      "step": 1540
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 9.891525423728813e-07,
+      "loss": 0.2248,
+      "step": 1541
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 9.884745762711863e-07,
+      "loss": 0.0401,
+      "step": 1542
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 9.877966101694914e-07,
+      "loss": 0.0886,
+      "step": 1543
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 9.871186440677966e-07,
+      "loss": 0.0326,
+      "step": 1544
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 9.864406779661017e-07,
+      "loss": 0.0931,
+      "step": 1545
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 9.857627118644068e-07,
+      "loss": 0.069,
+      "step": 1546
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 9.850847457627118e-07,
+      "loss": 0.1535,
+      "step": 1547
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 9.844067796610169e-07,
+      "loss": 0.0687,
+      "step": 1548
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 9.83728813559322e-07,
+      "loss": 0.0883,
+      "step": 1549
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 9.830508474576272e-07,
+      "loss": 0.0346,
+      "step": 1550
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 9.82372881355932e-07,
+      "loss": 0.0401,
+      "step": 1551
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 9.816949152542373e-07,
+      "loss": 0.0254,
+      "step": 1552
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 9.810169491525424e-07,
+      "loss": 0.0543,
+      "step": 1553
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 9.803389830508474e-07,
+      "loss": 0.0263,
+      "step": 1554
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 9.796610169491525e-07,
+      "loss": 0.0876,
+      "step": 1555
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 9.789830508474575e-07,
+      "loss": 0.0264,
+      "step": 1556
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 9.783050847457626e-07,
+      "loss": 0.0293,
+      "step": 1557
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 9.776271186440678e-07,
+      "loss": 0.0385,
+      "step": 1558
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 9.76949152542373e-07,
+      "loss": 0.1289,
+      "step": 1559
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 9.76271186440678e-07,
+      "loss": 0.0517,
+      "step": 1560
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 9.75593220338983e-07,
+      "loss": 0.143,
+      "step": 1561
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 9.74915254237288e-07,
+      "loss": 0.0375,
+      "step": 1562
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 9.742372881355931e-07,
+      "loss": 0.0317,
+      "step": 1563
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 9.735593220338982e-07,
+      "loss": 0.1121,
+      "step": 1564
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 9.728813559322032e-07,
+      "loss": 0.0701,
+      "step": 1565
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 9.722033898305085e-07,
+      "loss": 0.2551,
+      "step": 1566
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 9.715254237288135e-07,
+      "loss": 0.0532,
+      "step": 1567
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 9.708474576271186e-07,
+      "loss": 0.1007,
+      "step": 1568
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 9.701694915254237e-07,
+      "loss": 0.0974,
+      "step": 1569
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 9.694915254237287e-07,
+      "loss": 0.0702,
+      "step": 1570
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 9.68813559322034e-07,
+      "loss": 0.0662,
+      "step": 1571
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 9.681355932203388e-07,
+      "loss": 0.1244,
+      "step": 1572
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 9.674576271186439e-07,
+      "loss": 0.0324,
+      "step": 1573
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 9.667796610169491e-07,
+      "loss": 0.0394,
+      "step": 1574
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 9.661016949152542e-07,
+      "loss": 0.0366,
+      "step": 1575
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 9.654237288135593e-07,
+      "loss": 0.0771,
+      "step": 1576
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 9.647457627118643e-07,
+      "loss": 0.2417,
+      "step": 1577
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 9.640677966101694e-07,
+      "loss": 0.0307,
+      "step": 1578
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 9.633898305084746e-07,
+      "loss": 0.0383,
+      "step": 1579
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 9.627118644067797e-07,
+      "loss": 0.0589,
+      "step": 1580
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 9.620338983050847e-07,
+      "loss": 0.2179,
+      "step": 1581
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 9.613559322033898e-07,
+      "loss": 0.0272,
+      "step": 1582
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 9.606779661016949e-07,
+      "loss": 0.0946,
+      "step": 1583
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 9.6e-07,
+      "loss": 0.1388,
+      "step": 1584
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 9.59322033898305e-07,
+      "loss": 0.0241,
+      "step": 1585
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 9.5864406779661e-07,
+      "loss": 0.1374,
+      "step": 1586
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 9.579661016949153e-07,
+      "loss": 0.1069,
+      "step": 1587
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 9.572881355932203e-07,
+      "loss": 0.0575,
+      "step": 1588
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 9.566101694915254e-07,
+      "loss": 0.0337,
+      "step": 1589
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 9.559322033898305e-07,
+      "loss": 0.0286,
+      "step": 1590
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 9.552542372881355e-07,
+      "loss": 0.0446,
+      "step": 1591
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 9.545762711864408e-07,
+      "loss": 0.0488,
+      "step": 1592
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 9.538983050847456e-07,
+      "loss": 0.0918,
+      "step": 1593
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 9.532203389830508e-07,
+      "loss": 0.1277,
+      "step": 1594
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 9.525423728813558e-07,
+      "loss": 0.0363,
+      "step": 1595
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 9.51864406779661e-07,
+      "loss": 0.2863,
+      "step": 1596
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 9.511864406779661e-07,
+      "loss": 0.0457,
+      "step": 1597
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 9.505084745762711e-07,
+      "loss": 0.0841,
+      "step": 1598
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 9.498305084745763e-07,
+      "loss": 0.0411,
+      "step": 1599
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 9.491525423728813e-07,
+      "loss": 0.0781,
+      "step": 1600
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 9.484745762711864e-07,
+      "loss": 0.29,
+      "step": 1601
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 9.477966101694915e-07,
+      "loss": 0.1361,
+      "step": 1602
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 9.471186440677965e-07,
+      "loss": 0.0296,
+      "step": 1603
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 9.464406779661017e-07,
+      "loss": 0.1135,
+      "step": 1604
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 9.457627118644067e-07,
+      "loss": 0.3699,
+      "step": 1605
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 9.450847457627118e-07,
+      "loss": 0.0626,
+      "step": 1606
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 9.444067796610169e-07,
+      "loss": 0.1147,
+      "step": 1607
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 9.43728813559322e-07,
+      "loss": 0.0588,
+      "step": 1608
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 9.43050847457627e-07,
+      "loss": 0.2042,
+      "step": 1609
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 9.423728813559322e-07,
+      "loss": 0.0297,
+      "step": 1610
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 9.416949152542373e-07,
+      "loss": 0.0562,
+      "step": 1611
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 9.410169491525424e-07,
+      "loss": 0.0372,
+      "step": 1612
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 9.403389830508475e-07,
+      "loss": 0.0276,
+      "step": 1613
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 9.396610169491524e-07,
+      "loss": 0.1468,
+      "step": 1614
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 9.389830508474576e-07,
+      "loss": 0.1258,
+      "step": 1615
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 9.383050847457626e-07,
+      "loss": 0.0655,
+      "step": 1616
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 9.376271186440677e-07,
+      "loss": 0.1057,
+      "step": 1617
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 9.369491525423729e-07,
+      "loss": 0.1058,
+      "step": 1618
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 9.362711864406779e-07,
+      "loss": 0.1868,
+      "step": 1619
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 9.355932203389831e-07,
+      "loss": 0.0221,
+      "step": 1620
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 9.349152542372881e-07,
+      "loss": 0.1945,
+      "step": 1621
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 9.342372881355932e-07,
+      "loss": 0.0807,
+      "step": 1622
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 9.335593220338983e-07,
+      "loss": 0.043,
+      "step": 1623
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 9.328813559322033e-07,
+      "loss": 0.0308,
+      "step": 1624
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 9.322033898305083e-07,
+      "loss": 0.0649,
+      "step": 1625
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 9.315254237288135e-07,
+      "loss": 0.1007,
+      "step": 1626
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 9.308474576271186e-07,
+      "loss": 0.0588,
+      "step": 1627
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 9.301694915254237e-07,
+      "loss": 0.0193,
+      "step": 1628
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 9.294915254237288e-07,
+      "loss": 0.2193,
+      "step": 1629
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 9.288135593220338e-07,
+      "loss": 0.1133,
+      "step": 1630
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 9.28135593220339e-07,
+      "loss": 0.0732,
+      "step": 1631
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 9.274576271186441e-07,
+      "loss": 0.0529,
+      "step": 1632
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 9.267796610169491e-07,
+      "loss": 0.1215,
+      "step": 1633
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 9.261016949152543e-07,
+      "loss": 0.0263,
+      "step": 1634
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 9.254237288135592e-07,
+      "loss": 0.2879,
+      "step": 1635
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 9.247457627118644e-07,
+      "loss": 0.05,
+      "step": 1636
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 9.240677966101694e-07,
+      "loss": 0.0468,
+      "step": 1637
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 9.233898305084745e-07,
+      "loss": 0.2238,
+      "step": 1638
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 9.227118644067797e-07,
+      "loss": 0.0409,
+      "step": 1639
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 9.220338983050847e-07,
+      "loss": 0.0212,
+      "step": 1640
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 9.213559322033898e-07,
+      "loss": 0.0686,
+      "step": 1641
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 9.206779661016949e-07,
+      "loss": 0.0202,
+      "step": 1642
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 9.2e-07,
+      "loss": 0.0578,
+      "step": 1643
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 9.193220338983051e-07,
+      "loss": 0.1923,
+      "step": 1644
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 9.186440677966101e-07,
+      "loss": 0.0574,
+      "step": 1645
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 9.179661016949151e-07,
+      "loss": 0.2171,
+      "step": 1646
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 9.172881355932203e-07,
+      "loss": 0.0264,
+      "step": 1647
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 9.166101694915254e-07,
+      "loss": 0.0226,
+      "step": 1648
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 9.159322033898304e-07,
+      "loss": 0.1663,
+      "step": 1649
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 9.152542372881356e-07,
+      "loss": 0.1493,
+      "step": 1650
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.9475,
+      "eval_loss": 0.14478321373462677,
+      "eval_runtime": 2.1629,
+      "eval_samples_per_second": 554.809,
+      "eval_steps_per_second": 4.623,
+      "step": 1650
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.145762711864406e-07,
+      "loss": 0.081,
+      "step": 1651
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 9.138983050847458e-07,
+      "loss": 0.0363,
+      "step": 1652
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 9.132203389830509e-07,
+      "loss": 0.225,
+      "step": 1653
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.125423728813559e-07,
+      "loss": 0.0189,
+      "step": 1654
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 9.11864406779661e-07,
+      "loss": 0.112,
+      "step": 1655
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 9.11186440677966e-07,
+      "loss": 0.0565,
+      "step": 1656
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 9.105084745762711e-07,
+      "loss": 0.0993,
+      "step": 1657
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 9.098305084745762e-07,
+      "loss": 0.0203,
+      "step": 1658
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 9.091525423728813e-07,
+      "loss": 0.1224,
+      "step": 1659
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 9.084745762711864e-07,
+      "loss": 0.0782,
+      "step": 1660
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 9.077966101694915e-07,
+      "loss": 0.0613,
+      "step": 1661
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 9.071186440677966e-07,
+      "loss": 0.1541,
+      "step": 1662
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 9.064406779661017e-07,
+      "loss": 0.0877,
+      "step": 1663
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 9.057627118644068e-07,
+      "loss": 0.0268,
+      "step": 1664
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 9.050847457627118e-07,
+      "loss": 0.0517,
+      "step": 1665
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 9.044067796610169e-07,
+      "loss": 0.029,
+      "step": 1666
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 9.037288135593219e-07,
+      "loss": 0.0643,
+      "step": 1667
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 9.030508474576271e-07,
+      "loss": 0.1341,
+      "step": 1668
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 9.023728813559322e-07,
+      "loss": 0.0244,
+      "step": 1669
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 9.016949152542372e-07,
+      "loss": 0.0228,
+      "step": 1670
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 9.010169491525424e-07,
+      "loss": 0.2094,
+      "step": 1671
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 9.003389830508474e-07,
+      "loss": 0.0346,
+      "step": 1672
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 8.996610169491525e-07,
+      "loss": 0.0725,
+      "step": 1673
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 8.989830508474576e-07,
+      "loss": 0.1164,
+      "step": 1674
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 8.983050847457627e-07,
+      "loss": 0.0226,
+      "step": 1675
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 8.976271186440677e-07,
+      "loss": 0.2077,
+      "step": 1676
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 8.969491525423728e-07,
+      "loss": 0.0591,
+      "step": 1677
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 8.962711864406779e-07,
+      "loss": 0.032,
+      "step": 1678
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 8.95593220338983e-07,
+      "loss": 0.0822,
+      "step": 1679
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 8.949152542372881e-07,
+      "loss": 0.0863,
+      "step": 1680
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 8.942372881355931e-07,
+      "loss": 0.1718,
+      "step": 1681
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 8.935593220338983e-07,
+      "loss": 0.0819,
+      "step": 1682
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 8.928813559322034e-07,
+      "loss": 0.1918,
+      "step": 1683
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 8.922033898305085e-07,
+      "loss": 0.0255,
+      "step": 1684
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 8.915254237288136e-07,
+      "loss": 0.2546,
+      "step": 1685
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 8.908474576271186e-07,
+      "loss": 0.0429,
+      "step": 1686
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 8.901694915254237e-07,
+      "loss": 0.0385,
+      "step": 1687
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 8.894915254237287e-07,
+      "loss": 0.0557,
+      "step": 1688
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 8.888135593220338e-07,
+      "loss": 0.0666,
+      "step": 1689
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 8.88135593220339e-07,
+      "loss": 0.0375,
+      "step": 1690
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 8.87457627118644e-07,
+      "loss": 0.0403,
+      "step": 1691
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 8.867796610169492e-07,
+      "loss": 0.1401,
+      "step": 1692
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 8.861016949152542e-07,
+      "loss": 0.1327,
+      "step": 1693
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 8.854237288135593e-07,
+      "loss": 0.0674,
+      "step": 1694
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 8.847457627118644e-07,
+      "loss": 0.1114,
+      "step": 1695
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 8.840677966101695e-07,
+      "loss": 0.0906,
+      "step": 1696
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 8.833898305084745e-07,
+      "loss": 0.1371,
+      "step": 1697
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 8.827118644067796e-07,
+      "loss": 0.1079,
+      "step": 1698
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 8.820338983050847e-07,
+      "loss": 0.0557,
+      "step": 1699
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 8.813559322033897e-07,
+      "loss": 0.1407,
+      "step": 1700
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 8.806779661016949e-07,
+      "loss": 0.0174,
+      "step": 1701
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 8.799999999999999e-07,
+      "loss": 0.0196,
+      "step": 1702
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 8.793220338983051e-07,
+      "loss": 0.2311,
+      "step": 1703
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 8.786440677966102e-07,
+      "loss": 0.0892,
+      "step": 1704
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 8.779661016949152e-07,
+      "loss": 0.0751,
+      "step": 1705
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 8.772881355932204e-07,
+      "loss": 0.2013,
+      "step": 1706
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 8.766101694915253e-07,
+      "loss": 0.1109,
+      "step": 1707
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 8.759322033898304e-07,
+      "loss": 0.1075,
+      "step": 1708
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 8.752542372881355e-07,
+      "loss": 0.0662,
+      "step": 1709
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 8.745762711864406e-07,
+      "loss": 0.0216,
+      "step": 1710
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 8.738983050847458e-07,
+      "loss": 0.0504,
+      "step": 1711
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 8.732203389830508e-07,
+      "loss": 0.0234,
+      "step": 1712
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 8.725423728813559e-07,
+      "loss": 0.0263,
+      "step": 1713
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 8.71864406779661e-07,
+      "loss": 0.0962,
+      "step": 1714
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 8.711864406779661e-07,
+      "loss": 0.1633,
+      "step": 1715
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 8.705084745762711e-07,
+      "loss": 0.04,
+      "step": 1716
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 8.698305084745763e-07,
+      "loss": 0.0321,
+      "step": 1717
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 8.691525423728812e-07,
+      "loss": 0.0356,
+      "step": 1718
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 8.684745762711864e-07,
+      "loss": 0.1446,
+      "step": 1719
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 8.677966101694915e-07,
+      "loss": 0.0257,
+      "step": 1720
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 8.671186440677965e-07,
+      "loss": 0.159,
+      "step": 1721
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 8.664406779661017e-07,
+      "loss": 0.0947,
+      "step": 1722
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 8.657627118644067e-07,
+      "loss": 0.0194,
+      "step": 1723
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 8.650847457627118e-07,
+      "loss": 0.0315,
+      "step": 1724
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 8.64406779661017e-07,
+      "loss": 0.1295,
+      "step": 1725
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 8.63728813559322e-07,
+      "loss": 0.0317,
+      "step": 1726
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 8.630508474576272e-07,
+      "loss": 0.0466,
+      "step": 1727
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 8.623728813559321e-07,
+      "loss": 0.2306,
+      "step": 1728
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 8.616949152542372e-07,
+      "loss": 0.2285,
+      "step": 1729
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 8.610169491525423e-07,
+      "loss": 0.0204,
+      "step": 1730
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 8.603389830508474e-07,
+      "loss": 0.0379,
+      "step": 1731
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 8.596610169491524e-07,
+      "loss": 0.025,
+      "step": 1732
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 8.589830508474576e-07,
+      "loss": 0.1142,
+      "step": 1733
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 8.583050847457627e-07,
+      "loss": 0.1409,
+      "step": 1734
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 8.576271186440678e-07,
+      "loss": 0.0855,
+      "step": 1735
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 8.569491525423729e-07,
+      "loss": 0.0196,
+      "step": 1736
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 8.562711864406779e-07,
+      "loss": 0.0458,
+      "step": 1737
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 8.55593220338983e-07,
+      "loss": 0.1197,
+      "step": 1738
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 8.54915254237288e-07,
+      "loss": 0.0594,
+      "step": 1739
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 8.542372881355931e-07,
+      "loss": 0.0677,
+      "step": 1740
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 8.535593220338983e-07,
+      "loss": 0.0647,
+      "step": 1741
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 8.528813559322033e-07,
+      "loss": 0.0388,
+      "step": 1742
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 8.522033898305085e-07,
+      "loss": 0.0583,
+      "step": 1743
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 8.515254237288135e-07,
+      "loss": 0.1218,
+      "step": 1744
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 8.508474576271186e-07,
+      "loss": 0.0576,
+      "step": 1745
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 8.501694915254238e-07,
+      "loss": 0.0614,
+      "step": 1746
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 8.494915254237288e-07,
+      "loss": 0.0291,
+      "step": 1747
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 8.488135593220339e-07,
+      "loss": 0.0214,
+      "step": 1748
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 8.481355932203389e-07,
+      "loss": 0.0295,
+      "step": 1749
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 8.47457627118644e-07,
+      "loss": 0.04,
+      "step": 1750
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 8.467796610169491e-07,
+      "loss": 0.2888,
+      "step": 1751
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 8.461016949152542e-07,
+      "loss": 0.091,
+      "step": 1752
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 8.454237288135592e-07,
+      "loss": 0.1011,
+      "step": 1753
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 8.447457627118644e-07,
+      "loss": 0.0537,
+      "step": 1754
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 8.440677966101695e-07,
+      "loss": 0.0212,
+      "step": 1755
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 8.433898305084745e-07,
+      "loss": 0.0602,
+      "step": 1756
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 8.427118644067797e-07,
+      "loss": 0.0292,
+      "step": 1757
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 8.420338983050847e-07,
+      "loss": 0.1767,
+      "step": 1758
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 8.413559322033898e-07,
+      "loss": 0.1482,
+      "step": 1759
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 8.406779661016948e-07,
+      "loss": 0.1359,
+      "step": 1760
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 8.399999999999999e-07,
+      "loss": 0.165,
+      "step": 1761
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 8.393220338983051e-07,
+      "loss": 0.0285,
+      "step": 1762
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 8.386440677966101e-07,
+      "loss": 0.1368,
+      "step": 1763
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 8.379661016949152e-07,
+      "loss": 0.0363,
+      "step": 1764
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 8.372881355932203e-07,
+      "loss": 0.0407,
+      "step": 1765
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 8.366101694915254e-07,
+      "loss": 0.1627,
+      "step": 1766
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 8.359322033898305e-07,
+      "loss": 0.1292,
+      "step": 1767
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 8.352542372881356e-07,
+      "loss": 0.0687,
+      "step": 1768
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 8.345762711864407e-07,
+      "loss": 0.0187,
+      "step": 1769
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 8.338983050847457e-07,
+      "loss": 0.0235,
+      "step": 1770
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 8.332203389830508e-07,
+      "loss": 0.0354,
+      "step": 1771
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 8.325423728813558e-07,
+      "loss": 0.0961,
+      "step": 1772
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 8.31864406779661e-07,
+      "loss": 0.0591,
+      "step": 1773
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 8.31186440677966e-07,
+      "loss": 0.0528,
+      "step": 1774
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 8.305084745762712e-07,
+      "loss": 0.0346,
+      "step": 1775
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 8.298305084745763e-07,
+      "loss": 0.053,
+      "step": 1776
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 8.291525423728813e-07,
+      "loss": 0.0571,
+      "step": 1777
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 8.284745762711865e-07,
+      "loss": 0.3169,
+      "step": 1778
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 8.277966101694915e-07,
+      "loss": 0.2546,
+      "step": 1779
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 8.271186440677965e-07,
+      "loss": 0.0486,
+      "step": 1780
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 8.264406779661016e-07,
+      "loss": 0.1539,
+      "step": 1781
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 8.257627118644067e-07,
+      "loss": 0.0455,
+      "step": 1782
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 8.250847457627119e-07,
+      "loss": 0.0232,
+      "step": 1783
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 8.244067796610169e-07,
+      "loss": 0.0457,
+      "step": 1784
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 8.23728813559322e-07,
+      "loss": 0.0172,
+      "step": 1785
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 8.230508474576271e-07,
+      "loss": 0.1517,
+      "step": 1786
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 8.223728813559322e-07,
+      "loss": 0.0472,
+      "step": 1787
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 8.216949152542372e-07,
+      "loss": 0.0212,
+      "step": 1788
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 8.210169491525424e-07,
+      "loss": 0.0432,
+      "step": 1789
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 8.203389830508474e-07,
+      "loss": 0.1246,
+      "step": 1790
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 8.196610169491525e-07,
+      "loss": 0.0208,
+      "step": 1791
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 8.189830508474576e-07,
+      "loss": 0.0466,
+      "step": 1792
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 8.183050847457626e-07,
+      "loss": 0.0909,
+      "step": 1793
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 8.176271186440678e-07,
+      "loss": 0.0204,
+      "step": 1794
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 8.169491525423728e-07,
+      "loss": 0.12,
+      "step": 1795
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 8.162711864406779e-07,
+      "loss": 0.0938,
+      "step": 1796
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 8.155932203389831e-07,
+      "loss": 0.0808,
+      "step": 1797
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 8.149152542372881e-07,
+      "loss": 0.02,
+      "step": 1798
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 8.142372881355933e-07,
+      "loss": 0.0523,
+      "step": 1799
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 8.135593220338983e-07,
+      "loss": 0.094,
+      "step": 1800
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.9525,
+      "eval_loss": 0.13722257316112518,
+      "eval_runtime": 2.1728,
+      "eval_samples_per_second": 552.291,
+      "eval_steps_per_second": 4.602,
+      "step": 1800
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.128813559322033e-07,
+      "loss": 0.0227,
+      "step": 1801
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 8.122033898305084e-07,
+      "loss": 0.0743,
+      "step": 1802
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 8.115254237288135e-07,
+      "loss": 0.0921,
+      "step": 1803
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.108474576271185e-07,
+      "loss": 0.0433,
+      "step": 1804
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 8.101694915254237e-07,
+      "loss": 0.0233,
+      "step": 1805
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 8.094915254237288e-07,
+      "loss": 0.2267,
+      "step": 1806
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.088135593220339e-07,
+      "loss": 0.0152,
+      "step": 1807
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 8.08135593220339e-07,
+      "loss": 0.1364,
+      "step": 1808
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 8.07457627118644e-07,
+      "loss": 0.0307,
+      "step": 1809
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 8.067796610169492e-07,
+      "loss": 0.0941,
+      "step": 1810
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 8.061016949152541e-07,
+      "loss": 0.0434,
+      "step": 1811
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 8.054237288135592e-07,
+      "loss": 0.0579,
+      "step": 1812
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 8.047457627118644e-07,
+      "loss": 0.0385,
+      "step": 1813
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 8.040677966101694e-07,
+      "loss": 0.209,
+      "step": 1814
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 8.033898305084746e-07,
+      "loss": 0.0209,
+      "step": 1815
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 8.027118644067796e-07,
+      "loss": 0.0451,
+      "step": 1816
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 8.020338983050847e-07,
+      "loss": 0.0444,
+      "step": 1817
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 8.013559322033899e-07,
+      "loss": 0.0255,
+      "step": 1818
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 8.006779661016949e-07,
+      "loss": 0.1247,
+      "step": 1819
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 8e-07,
+      "loss": 0.176,
+      "step": 1820
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 7.993220338983051e-07,
+      "loss": 0.0581,
+      "step": 1821
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 7.986440677966101e-07,
+      "loss": 0.1404,
+      "step": 1822
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 7.979661016949152e-07,
+      "loss": 0.1071,
+      "step": 1823
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 7.972881355932203e-07,
+      "loss": 0.0824,
+      "step": 1824
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 7.966101694915253e-07,
+      "loss": 0.0981,
+      "step": 1825
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 7.959322033898305e-07,
+      "loss": 0.0582,
+      "step": 1826
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 7.952542372881356e-07,
+      "loss": 0.0143,
+      "step": 1827
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 7.945762711864406e-07,
+      "loss": 0.0426,
+      "step": 1828
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 7.938983050847458e-07,
+      "loss": 0.1532,
+      "step": 1829
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 7.932203389830508e-07,
+      "loss": 0.0176,
+      "step": 1830
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 7.92542372881356e-07,
+      "loss": 0.0448,
+      "step": 1831
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 7.918644067796609e-07,
+      "loss": 0.0358,
+      "step": 1832
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 7.91186440677966e-07,
+      "loss": 0.0301,
+      "step": 1833
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 7.905084745762712e-07,
+      "loss": 0.025,
+      "step": 1834
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 7.898305084745762e-07,
+      "loss": 0.0306,
+      "step": 1835
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 7.891525423728813e-07,
+      "loss": 0.0499,
+      "step": 1836
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 7.884745762711864e-07,
+      "loss": 0.0214,
+      "step": 1837
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 7.877966101694915e-07,
+      "loss": 0.0269,
+      "step": 1838
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 7.871186440677967e-07,
+      "loss": 0.1555,
+      "step": 1839
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 7.864406779661017e-07,
+      "loss": 0.0176,
+      "step": 1840
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 7.857627118644068e-07,
+      "loss": 0.0696,
+      "step": 1841
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 7.850847457627118e-07,
+      "loss": 0.0516,
+      "step": 1842
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 7.844067796610169e-07,
+      "loss": 0.2515,
+      "step": 1843
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 7.837288135593219e-07,
+      "loss": 0.0157,
+      "step": 1844
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 7.830508474576271e-07,
+      "loss": 0.3365,
+      "step": 1845
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 7.823728813559321e-07,
+      "loss": 0.0659,
+      "step": 1846
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 7.816949152542373e-07,
+      "loss": 0.2075,
+      "step": 1847
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 7.810169491525424e-07,
+      "loss": 0.0214,
+      "step": 1848
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 7.803389830508474e-07,
+      "loss": 0.0191,
+      "step": 1849
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 7.796610169491526e-07,
+      "loss": 0.0968,
+      "step": 1850
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 7.789830508474576e-07,
+      "loss": 0.0258,
+      "step": 1851
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 7.783050847457627e-07,
+      "loss": 0.0192,
+      "step": 1852
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 7.776271186440677e-07,
+      "loss": 0.0697,
+      "step": 1853
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 7.769491525423728e-07,
+      "loss": 0.0173,
+      "step": 1854
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 7.762711864406779e-07,
+      "loss": 0.0179,
+      "step": 1855
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 7.75593220338983e-07,
+      "loss": 0.1792,
+      "step": 1856
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 7.749152542372881e-07,
+      "loss": 0.0428,
+      "step": 1857
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 7.742372881355932e-07,
+      "loss": 0.1118,
+      "step": 1858
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 7.735593220338983e-07,
+      "loss": 0.0186,
+      "step": 1859
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 7.728813559322033e-07,
+      "loss": 0.0226,
+      "step": 1860
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 7.722033898305085e-07,
+      "loss": 0.0414,
+      "step": 1861
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 7.715254237288136e-07,
+      "loss": 0.018,
+      "step": 1862
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 7.708474576271185e-07,
+      "loss": 0.0611,
+      "step": 1863
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 7.701694915254237e-07,
+      "loss": 0.0439,
+      "step": 1864
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 7.694915254237287e-07,
+      "loss": 0.1583,
+      "step": 1865
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 7.688135593220339e-07,
+      "loss": 0.0423,
+      "step": 1866
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 7.681355932203389e-07,
+      "loss": 0.1553,
+      "step": 1867
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 7.67457627118644e-07,
+      "loss": 0.0157,
+      "step": 1868
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 7.667796610169492e-07,
+      "loss": 0.1468,
+      "step": 1869
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 7.661016949152542e-07,
+      "loss": 0.1471,
+      "step": 1870
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 7.654237288135594e-07,
+      "loss": 0.0795,
+      "step": 1871
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 7.647457627118644e-07,
+      "loss": 0.02,
+      "step": 1872
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 7.640677966101694e-07,
+      "loss": 0.1232,
+      "step": 1873
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 7.633898305084745e-07,
+      "loss": 0.1155,
+      "step": 1874
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 7.627118644067796e-07,
+      "loss": 0.0198,
+      "step": 1875
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 7.620338983050847e-07,
+      "loss": 0.1619,
+      "step": 1876
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 7.613559322033898e-07,
+      "loss": 0.2416,
+      "step": 1877
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 7.606779661016949e-07,
+      "loss": 0.0857,
+      "step": 1878
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 7.599999999999999e-07,
+      "loss": 0.1299,
+      "step": 1879
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 7.593220338983051e-07,
+      "loss": 0.0285,
+      "step": 1880
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 7.586440677966101e-07,
+      "loss": 0.0767,
+      "step": 1881
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 7.579661016949153e-07,
+      "loss": 0.0193,
+      "step": 1882
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 7.572881355932204e-07,
+      "loss": 0.0185,
+      "step": 1883
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 7.566101694915253e-07,
+      "loss": 0.0485,
+      "step": 1884
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 7.559322033898305e-07,
+      "loss": 0.1179,
+      "step": 1885
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 7.552542372881355e-07,
+      "loss": 0.1148,
+      "step": 1886
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 7.545762711864406e-07,
+      "loss": 0.0195,
+      "step": 1887
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 7.538983050847457e-07,
+      "loss": 0.0834,
+      "step": 1888
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 7.532203389830508e-07,
+      "loss": 0.0482,
+      "step": 1889
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 7.52542372881356e-07,
+      "loss": 0.0625,
+      "step": 1890
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 7.51864406779661e-07,
+      "loss": 0.0235,
+      "step": 1891
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 7.511864406779661e-07,
+      "loss": 0.0796,
+      "step": 1892
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 7.505084745762712e-07,
+      "loss": 0.0281,
+      "step": 1893
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 7.498305084745762e-07,
+      "loss": 0.1491,
+      "step": 1894
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 7.491525423728812e-07,
+      "loss": 0.0249,
+      "step": 1895
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 7.484745762711864e-07,
+      "loss": 0.0944,
+      "step": 1896
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 7.477966101694914e-07,
+      "loss": 0.1889,
+      "step": 1897
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 7.471186440677966e-07,
+      "loss": 0.0392,
+      "step": 1898
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 7.464406779661017e-07,
+      "loss": 0.1033,
+      "step": 1899
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 7.457627118644067e-07,
+      "loss": 0.0419,
+      "step": 1900
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 7.450847457627119e-07,
+      "loss": 0.1089,
+      "step": 1901
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 7.444067796610169e-07,
+      "loss": 0.0267,
+      "step": 1902
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 7.43728813559322e-07,
+      "loss": 0.079,
+      "step": 1903
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 7.430508474576272e-07,
+      "loss": 0.0689,
+      "step": 1904
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 7.423728813559321e-07,
+      "loss": 0.1049,
+      "step": 1905
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 7.416949152542373e-07,
+      "loss": 0.0277,
+      "step": 1906
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 7.410169491525423e-07,
+      "loss": 0.0506,
+      "step": 1907
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 7.403389830508474e-07,
+      "loss": 0.1173,
+      "step": 1908
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 7.396610169491525e-07,
+      "loss": 0.0162,
+      "step": 1909
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 7.389830508474576e-07,
+      "loss": 0.0229,
+      "step": 1910
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 7.383050847457626e-07,
+      "loss": 0.0869,
+      "step": 1911
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 7.376271186440678e-07,
+      "loss": 0.033,
+      "step": 1912
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 7.369491525423729e-07,
+      "loss": 0.0501,
+      "step": 1913
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 7.36271186440678e-07,
+      "loss": 0.0581,
+      "step": 1914
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 7.35593220338983e-07,
+      "loss": 0.0955,
+      "step": 1915
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 7.34915254237288e-07,
+      "loss": 0.1402,
+      "step": 1916
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 7.342372881355932e-07,
+      "loss": 0.0487,
+      "step": 1917
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 7.335593220338982e-07,
+      "loss": 0.1935,
+      "step": 1918
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 7.328813559322033e-07,
+      "loss": 0.1322,
+      "step": 1919
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 7.322033898305085e-07,
+      "loss": 0.0488,
+      "step": 1920
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.315254237288135e-07,
+      "loss": 0.0133,
+      "step": 1921
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 7.308474576271187e-07,
+      "loss": 0.0185,
+      "step": 1922
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 7.301694915254237e-07,
+      "loss": 0.0196,
+      "step": 1923
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.294915254237288e-07,
+      "loss": 0.0639,
+      "step": 1924
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 7.288135593220338e-07,
+      "loss": 0.1526,
+      "step": 1925
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 7.281355932203389e-07,
+      "loss": 0.1469,
+      "step": 1926
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.27457627118644e-07,
+      "loss": 0.0621,
+      "step": 1927
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 7.267796610169491e-07,
+      "loss": 0.1961,
+      "step": 1928
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 7.261016949152542e-07,
+      "loss": 0.0304,
+      "step": 1929
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.254237288135593e-07,
+      "loss": 0.016,
+      "step": 1930
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 7.247457627118644e-07,
+      "loss": 0.0129,
+      "step": 1931
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 7.240677966101694e-07,
+      "loss": 0.0342,
+      "step": 1932
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.233898305084746e-07,
+      "loss": 0.1346,
+      "step": 1933
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 7.227118644067797e-07,
+      "loss": 0.0674,
+      "step": 1934
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 7.220338983050847e-07,
+      "loss": 0.1377,
+      "step": 1935
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 7.213559322033898e-07,
+      "loss": 0.0353,
+      "step": 1936
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 7.206779661016948e-07,
+      "loss": 0.083,
+      "step": 1937
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 7.2e-07,
+      "loss": 0.1333,
+      "step": 1938
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 7.19322033898305e-07,
+      "loss": 0.0248,
+      "step": 1939
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 7.186440677966101e-07,
+      "loss": 0.0488,
+      "step": 1940
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 7.179661016949153e-07,
+      "loss": 0.0335,
+      "step": 1941
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 7.172881355932203e-07,
+      "loss": 0.0201,
+      "step": 1942
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 7.166101694915254e-07,
+      "loss": 0.0566,
+      "step": 1943
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 7.159322033898305e-07,
+      "loss": 0.0592,
+      "step": 1944
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 7.152542372881356e-07,
+      "loss": 0.1423,
+      "step": 1945
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 7.145762711864406e-07,
+      "loss": 0.0194,
+      "step": 1946
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 7.138983050847457e-07,
+      "loss": 0.1888,
+      "step": 1947
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 7.132203389830508e-07,
+      "loss": 0.0184,
+      "step": 1948
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 7.125423728813559e-07,
+      "loss": 0.1541,
+      "step": 1949
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 7.11864406779661e-07,
+      "loss": 0.0269,
+      "step": 1950
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.9533333333333334,
+      "eval_loss": 0.13513699173927307,
+      "eval_runtime": 2.1693,
+      "eval_samples_per_second": 553.18,
+      "eval_steps_per_second": 4.61,
+      "step": 1950
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3000,
+  "num_train_epochs": 20,
+  "save_steps": 500,
+  "total_flos": 1664093699597952.0,
+  "trial_name": null,
+  "trial_params": null
+}