diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12652 @@
+{
+  "best_metric": 0.95,
+  "best_model_checkpoint": "timesformer-base-finetuned-k400-kinetic400-subset-epoch6real-num_frame_10_myViT2_more_data/checkpoint-750",
+  "epoch": 5.166666666666667,
+  "eval_steps": 500,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "accuracy": 0.0,
+      "epoch": 0,
+      "step": 0
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.0,
+      "step": 1
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.0,
+      "step": 2
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.0,
+      "step": 3
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.0,
+      "step": 4
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.0,
+      "step": 5
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.0,
+      "step": 6
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.0,
+      "step": 7
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.01,
+      "step": 8
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.01,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 27.48541259765625,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.4397,
+      "step": 10
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 10
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 11
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.01,
+      "step": 12
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 13
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 14
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 15
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.01,
+      "step": 16
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 17
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 18
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 29.550081253051758,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.7011,
+      "step": 20
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 20
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 21
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 22
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.02,
+      "step": 23
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 24
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 25
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.02,
+      "step": 26
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 27
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.02,
+      "step": 28
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 30.159160614013672,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.6597,
+      "step": 30
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 30
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 31
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 32
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 33
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 34
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 35
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.02,
+      "step": 36
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 37
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.03,
+      "step": 38
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 30.40444564819336,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 1.4862,
+      "step": 40
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.03,
+      "step": 40
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 41
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 42
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.03,
+      "step": 43
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.03,
+      "step": 44
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 45
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 46
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.03,
+      "step": 47
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 48
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 24.201858520507812,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.2066,
+      "step": 50
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 50
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 51
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.03,
+      "step": 52
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.04,
+      "step": 53
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.04,
+      "step": 54
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.04,
+      "step": 55
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.04,
+      "step": 56
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.04,
+      "step": 57
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.04,
+      "step": 58
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.04,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 35.361610412597656,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.3168,
+      "step": 60
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.04,
+      "step": 60
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.04,
+      "step": 61
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.04,
+      "step": 62
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.04,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.04,
+      "step": 64
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.04,
+      "step": 65
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.04,
+      "step": 66
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.04,
+      "step": 67
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 68
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 17.602001190185547,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.9897,
+      "step": 70
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 70
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 71
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 72
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.05,
+      "step": 73
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 74
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.05,
+      "step": 75
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 76
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 77
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 78
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 18.5585880279541,
+      "learning_rate": 2.1333333333333335e-05,
+      "loss": 0.724,
+      "step": 80
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 80
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.05,
+      "step": 81
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 82
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 83
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 84
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 85
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.06,
+      "step": 86
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.06,
+      "step": 87
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 88
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 89
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 7.358654022216797,
+      "learning_rate": 2.4e-05,
+      "loss": 0.9884,
+      "step": 90
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 90
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 91
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 92
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.06,
+      "step": 93
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 94
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 95
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 96
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 97
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 98
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 6.941775321960449,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.4211,
+      "step": 100
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.07,
+      "step": 100
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 101
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 102
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 103
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 104
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 105
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.07,
+      "step": 106
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 107
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 108
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.07,
+      "step": 109
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 40.621578216552734,
+      "learning_rate": 2.9333333333333333e-05,
+      "loss": 0.8335,
+      "step": 110
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 110
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.07,
+      "step": 111
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 112
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 113
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 114
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 115
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 116
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 117
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.08,
+      "step": 118
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 119
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 32.61237335205078,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.7528,
+      "step": 120
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 120
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 121
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 122
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.08,
+      "step": 123
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.08,
+      "step": 124
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 125
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 127
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.09,
+      "step": 128
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 129
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 3.2939963340759277,
+      "learning_rate": 3.466666666666667e-05,
+      "loss": 0.6304,
+      "step": 130
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 130
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 131
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.09,
+      "step": 132
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 133
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.09,
+      "step": 134
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 135
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 136
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 137
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 138
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 139
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 15.463095664978027,
+      "learning_rate": 3.733333333333334e-05,
+      "loss": 0.3755,
+      "step": 140
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.09,
+      "step": 140
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 141
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 142
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 143
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 144
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 145
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 146
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.1,
+      "step": 147
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 148
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 149
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 26.474327087402344,
+      "learning_rate": 4e-05,
+      "loss": 0.6569,
+      "step": 150
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.1,
+      "step": 150
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 151
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 152
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 153
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 154
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 155
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 156
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 157
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 158
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.11,
+      "step": 159
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 59.37823486328125,
+      "learning_rate": 3.970370370370371e-05,
+      "loss": 0.7029,
+      "step": 160
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 160
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 161
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 162
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 163
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.11,
+      "step": 164
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 165
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 166
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 167
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.11,
+      "step": 168
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.11,
+      "step": 169
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 47.64479064941406,
+      "learning_rate": 3.940740740740741e-05,
+      "loss": 0.9479,
+      "step": 170
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 170
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 171
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 172
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 173
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 174
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 175
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 176
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 177
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 178
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 179
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 6.282685279846191,
+      "learning_rate": 3.9111111111111115e-05,
+      "loss": 0.0138,
+      "step": 180
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 180
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 181
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 182
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.12,
+      "step": 183
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 184
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 185
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 186
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 187
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 188
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 189
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.2229953110218048,
+      "learning_rate": 3.8814814814814814e-05,
+      "loss": 0.1847,
+      "step": 190
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 190
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 191
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 192
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 193
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 194
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.13,
+      "step": 195
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 196
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.13,
+      "step": 197
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 198
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 199
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.057327281683683395,
+      "learning_rate": 3.851851851851852e-05,
+      "loss": 0.5499,
+      "step": 200
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 200
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 201
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 202
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 203
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 204
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 205
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 206
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 207
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 208
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 209
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.054187316447496414,
+      "learning_rate": 3.8222222222222226e-05,
+      "loss": 0.0226,
+      "step": 210
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 210
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 211
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 212
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 213
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 214
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 215
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 216
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 217
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 218
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 219
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.8997392654418945,
+      "learning_rate": 3.792592592592593e-05,
+      "loss": 0.1139,
+      "step": 220
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.15,
+      "step": 220
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 221
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 222
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 223
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 224
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 225
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.15,
+      "step": 226
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 227
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 228
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 229
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 24.185007095336914,
+      "learning_rate": 3.762962962962963e-05,
+      "loss": 0.8915,
+      "step": 230
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 230
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 231
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 232
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.16,
+      "step": 233
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 234
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 235
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 236
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 237
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 238
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 239
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.667302370071411,
+      "learning_rate": 3.733333333333334e-05,
+      "loss": 0.5073,
+      "step": 240
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 240
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 241
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 242
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 243
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.16,
+      "step": 244
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 245
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 246
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 247
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 248
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 249
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.050393581390381,
+      "learning_rate": 3.703703703703704e-05,
+      "loss": 0.4861,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "epoch": 0.17,
+      "eval_accuracy": 0.9,
+      "eval_loss": 0.3385741412639618,
+      "eval_runtime": 122.8863,
+      "eval_samples_per_second": 0.814,
+      "eval_steps_per_second": 0.814,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.0,
+      "step": 251
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.0,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.0,
+      "step": 253
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.0,
+      "step": 254
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.0,
+      "step": 255
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.0,
+      "step": 256
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.0,
+      "step": 257
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 258
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.01,
+      "step": 259
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 59.7421875,
+      "learning_rate": 3.674074074074074e-05,
+      "loss": 1.0274,
+      "step": 260
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 260
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 261
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 262
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.01,
+      "step": 263
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 264
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 265
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 266
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 267
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 268
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 269
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.36427316069602966,
+      "learning_rate": 3.644444444444445e-05,
+      "loss": 0.5578,
+      "step": 270
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 270
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 271
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 272
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 273
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 274
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 275
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 276
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 277
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 278
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 279
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.14365604519844055,
+      "learning_rate": 3.614814814814815e-05,
+      "loss": 0.0018,
+      "step": 280
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 280
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 281
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 282
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 283
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 284
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 285
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 286
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 287
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.03,
+      "step": 288
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 289
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.021687423810362816,
+      "learning_rate": 3.5851851851851854e-05,
+      "loss": 0.0945,
+      "step": 290
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 290
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 291
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 292
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 293
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 294
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 295
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 296
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 297
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 298
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 299
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.15070036053657532,
+      "learning_rate": 3.555555555555555e-05,
+      "loss": 0.0024,
+      "step": 300
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 300
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 301
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 302
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 303
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 304
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 305
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 306
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.04,
+      "step": 307
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 308
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 309
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.14499500393867493,
+      "learning_rate": 3.5259259259259266e-05,
+      "loss": 0.104,
+      "step": 310
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 310
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 311
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 312
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.04,
+      "step": 313
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 314
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 316
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 317
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.05,
+      "step": 318
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 319
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.022930795326828957,
+      "learning_rate": 3.4962962962962965e-05,
+      "loss": 0.9185,
+      "step": 320
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 320
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 321
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 322
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 323
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 324
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 325
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 326
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 327
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 328
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 329
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.4176673889160156,
+      "learning_rate": 3.466666666666667e-05,
+      "loss": 0.0069,
+      "step": 330
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 330
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 331
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 332
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 333
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 334
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 335
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 336
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 337
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 338
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 339
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.017544511705636978,
+      "learning_rate": 3.437037037037037e-05,
+      "loss": 0.0845,
+      "step": 340
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 340
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 341
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 342
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 343
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 344
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 345
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 346
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 347
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 348
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 349
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.04922616481781,
+      "learning_rate": 3.4074074074074077e-05,
+      "loss": 0.0402,
+      "step": 350
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 350
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 351
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 352
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.07,
+      "step": 353
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 354
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 355
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 356
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 357
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 358
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 359
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.008364695124328136,
+      "learning_rate": 3.377777777777778e-05,
+      "loss": 0.4109,
+      "step": 360
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 360
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 361
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 362
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.08,
+      "step": 363
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 364
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 365
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 366
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 367
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 368
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 369
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.010020878165960312,
+      "learning_rate": 3.348148148148148e-05,
+      "loss": 0.295,
+      "step": 370
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 370
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 371
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 373
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 374
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 375
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 376
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 377
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 378
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 379
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.015522874891757965,
+      "learning_rate": 3.318518518518519e-05,
+      "loss": 0.0262,
+      "step": 380
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 380
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 381
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 382
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 383
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 384
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 385
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 386
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 387
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 388
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 389
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.005305934697389603,
+      "learning_rate": 3.288888888888889e-05,
+      "loss": 0.0014,
+      "step": 390
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 390
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 391
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 392
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 393
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 394
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.1,
+      "step": 395
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 396
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 397
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 398
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 399
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.009330866858363152,
+      "learning_rate": 3.259259259259259e-05,
+      "loss": 0.6288,
+      "step": 400
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 400
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.1,
+      "step": 401
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 402
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 403
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 404
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 405
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 406
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 407
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 408
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 409
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.7033286094665527,
+      "learning_rate": 3.22962962962963e-05,
+      "loss": 0.976,
+      "step": 410
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 410
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 411
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 412
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 413
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.11,
+      "step": 414
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 415
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 416
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 417
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 418
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 419
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.7694259285926819,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.1421,
+      "step": 420
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 420
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 421
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 422
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 423
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 424
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 425
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 426
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 427
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 428
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 429
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.005554061848670244,
+      "learning_rate": 3.1703703703703705e-05,
+      "loss": 0.0059,
+      "step": 430
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 430
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 431
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 432
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 433
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 434
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 435
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.12,
+      "step": 436
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 437
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 438
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 439
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.03329182416200638,
+      "learning_rate": 3.140740740740741e-05,
+      "loss": 0.321,
+      "step": 440
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 440
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 441
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 442
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 443
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 444
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 445
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 446
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 447
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 448
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 449
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.13921838998794556,
+      "learning_rate": 3.111111111111112e-05,
+      "loss": 0.0028,
+      "step": 450
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 450
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 451
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 452
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 453
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.14,
+      "step": 454
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 455
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 456
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 457
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 458
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 459
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.4193977415561676,
+      "learning_rate": 3.0814814814814816e-05,
+      "loss": 0.2971,
+      "step": 460
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 460
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 461
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 462
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 463
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 464
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 465
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 466
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 467
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 468
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 469
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.012862984091043472,
+      "learning_rate": 3.051851851851852e-05,
+      "loss": 0.0012,
+      "step": 470
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 470
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 471
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 472
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 473
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 474
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 475
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 476
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 477
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 478
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 479
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.11006419360637665,
+      "learning_rate": 3.0222222222222225e-05,
+      "loss": 0.0062,
+      "step": 480
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 480
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 481
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 482
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 483
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 484
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 485
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 486
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 487
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 488
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 489
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.2075457125902176,
+      "learning_rate": 2.9925925925925927e-05,
+      "loss": 0.0013,
+      "step": 490
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.16,
+      "step": 490
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 491
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 492
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 493
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 494
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 495
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 496
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 497
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 498
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 499
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.008261927403509617,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 0.7118,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "epoch": 1.17,
+      "eval_accuracy": 0.9,
+      "eval_loss": 0.5019777417182922,
+      "eval_runtime": 120.2375,
+      "eval_samples_per_second": 0.832,
+      "eval_steps_per_second": 0.832,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.0,
+      "step": 501
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.0,
+      "step": 502
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.0,
+      "step": 503
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.0,
+      "step": 504
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.0,
+      "step": 505
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.0,
+      "step": 506
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.0,
+      "step": 507
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 508
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 509
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.053543075919151306,
+      "learning_rate": 2.9333333333333333e-05,
+      "loss": 0.3723,
+      "step": 510
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 510
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 511
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 512
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 513
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 514
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 515
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 516
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 517
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 518
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 519
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.008527850732207298,
+      "learning_rate": 2.9037037037037042e-05,
+      "loss": 0.0013,
+      "step": 520
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 520
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 521
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 522
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 523
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 524
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 525
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 526
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 527
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 528
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 529
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.030531510710716248,
+      "learning_rate": 2.8740740740740745e-05,
+      "loss": 0.0139,
+      "step": 530
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 530
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 531
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 532
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 533
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 534
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 535
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.02,
+      "step": 536
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 537
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 538
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 539
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.009555542841553688,
+      "learning_rate": 2.8444444444444447e-05,
+      "loss": 0.1067,
+      "step": 540
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 540
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 541
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 542
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 543
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 544
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 545
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 546
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 547
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 548
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 549
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.010882402770221233,
+      "learning_rate": 2.814814814814815e-05,
+      "loss": 0.0267,
+      "step": 550
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 550
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 551
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 552
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 553
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 554
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 555
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 556
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 557
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 558
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 559
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.01478774193674326,
+      "learning_rate": 2.7851851851851856e-05,
+      "loss": 0.0579,
+      "step": 560
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 560
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 561
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 562
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 563
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 564
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 565
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.04,
+      "step": 566
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 567
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 568
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 569
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.0239090658724308,
+      "learning_rate": 2.755555555555556e-05,
+      "loss": 0.0922,
+      "step": 570
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 570
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 571
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 572
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 573
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 574
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 575
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 576
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 577
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 578
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 579
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.008603918366134167,
+      "learning_rate": 2.725925925925926e-05,
+      "loss": 0.0013,
+      "step": 580
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 580
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 581
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 582
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 583
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 584
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 585
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 586
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 587
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 588
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 589
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.009832022711634636,
+      "learning_rate": 2.6962962962962964e-05,
+      "loss": 0.001,
+      "step": 590
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 590
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 591
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 592
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 593
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 594
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 595
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 596
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 597
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 598
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 599
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.013950807973742485,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 0.0007,
+      "step": 600
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 600
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 601
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 602
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 603
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 604
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 605
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 606
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 607
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 608
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 609
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.0075510661117732525,
+      "learning_rate": 2.637037037037037e-05,
+      "loss": 0.0029,
+      "step": 610
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 610
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 611
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 612
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 613
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 614
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 615
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 616
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 617
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 618
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 619
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.015888426452875137,
+      "learning_rate": 2.607407407407408e-05,
+      "loss": 0.0002,
+      "step": 620
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 620
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 621
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 622
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 623
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 624
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 625
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 626
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 627
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 628
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 629
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.006054020952433348,
+      "learning_rate": 2.577777777777778e-05,
+      "loss": 0.0129,
+      "step": 630
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 630
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 631
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 632
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 633
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 634
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 635
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 636
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 637
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 638
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 639
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.004561170935630798,
+      "learning_rate": 2.5481481481481484e-05,
+      "loss": 0.0187,
+      "step": 640
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 640
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 641
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 642
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 643
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 644
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 645
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 646
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 647
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 648
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 649
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.004491969011723995,
+      "learning_rate": 2.5185185185185187e-05,
+      "loss": 0.0006,
+      "step": 650
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 650
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 651
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 652
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 653
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 654
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 655
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 656
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 657
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 658
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 659
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.00564774964004755,
+      "learning_rate": 2.4888888888888893e-05,
+      "loss": 0.008,
+      "step": 660
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 660
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 661
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 662
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 663
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 664
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 665
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 666
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 667
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 668
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 669
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.002186964266002178,
+      "learning_rate": 2.4592592592592595e-05,
+      "loss": 0.0262,
+      "step": 670
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 670
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.11,
+      "step": 671
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 672
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 673
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 674
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 675
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 676
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 677
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 678
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 679
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.005805268418043852,
+      "learning_rate": 2.4296296296296298e-05,
+      "loss": 0.1838,
+      "step": 680
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 680
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 681
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 682
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 683
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 684
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 685
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 686
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 687
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 688
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 689
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.0018611220875754952,
+      "learning_rate": 2.4e-05,
+      "loss": 0.0007,
+      "step": 690
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 690
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 691
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 692
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 693
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.13,
+      "step": 694
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 695
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 696
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 697
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 698
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 699
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.0077749197371304035,
+      "learning_rate": 2.3703703703703703e-05,
+      "loss": 0.1827,
+      "step": 700
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 700
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 701
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 702
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 703
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 704
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 705
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 706
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 707
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 708
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 709
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.007749865297228098,
+      "learning_rate": 2.3407407407407406e-05,
+      "loss": 0.0003,
+      "step": 710
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 710
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 711
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 712
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 713
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 714
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 715
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 716
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 717
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 718
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 719
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.0038833627477288246,
+      "learning_rate": 2.3111111111111112e-05,
+      "loss": 0.0015,
+      "step": 720
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 720
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 721
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 722
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 723
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 724
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 725
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 726
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 727
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 728
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 729
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 7.431872844696045,
+      "learning_rate": 2.2814814814814818e-05,
+      "loss": 0.09,
+      "step": 730
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 730
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 731
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 732
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 733
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 734
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 735
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 736
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 737
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 738
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 739
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.002486105542629957,
+      "learning_rate": 2.251851851851852e-05,
+      "loss": 0.0109,
+      "step": 740
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 740
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 741
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 742
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 743
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 744
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 745
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 746
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 747
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 748
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 749
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.0030093807727098465,
+      "learning_rate": 2.2222222222222227e-05,
+      "loss": 0.004,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "epoch": 2.17,
+      "eval_accuracy": 0.95,
+      "eval_loss": 0.17908386886119843,
+      "eval_runtime": 117.0774,
+      "eval_samples_per_second": 0.854,
+      "eval_steps_per_second": 0.854,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 750
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.0,
+      "step": 751
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.0,
+      "step": 752
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.0,
+      "step": 753
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.0,
+      "step": 754
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.0,
+      "step": 755
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.0,
+      "step": 756
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.0,
+      "step": 757
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 758
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 759
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.0023822493385523558,
+      "learning_rate": 2.192592592592593e-05,
+      "loss": 0.0003,
+      "step": 760
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 760
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 761
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 762
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 763
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 764
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 765
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 766
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 767
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 768
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 769
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.07780895382165909,
+      "learning_rate": 2.1629629629629632e-05,
+      "loss": 0.043,
+      "step": 770
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 770
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 771
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 772
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.02,
+      "step": 773
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 774
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 775
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 776
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 777
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 778
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 779
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.0048459903337061405,
+      "learning_rate": 2.1333333333333335e-05,
+      "loss": 0.1758,
+      "step": 780
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 780
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 781
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 782
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 783
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 784
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 785
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 786
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 787
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 788
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 789
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.003220546757802367,
+      "learning_rate": 2.1037037037037037e-05,
+      "loss": 0.0003,
+      "step": 790
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 790
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 791
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 792
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 793
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 794
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 795
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 796
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 797
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 798
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 799
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.013224848546087742,
+      "learning_rate": 2.074074074074074e-05,
+      "loss": 0.0004,
+      "step": 800
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 800
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 801
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 802
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 803
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 804
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 805
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 806
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 807
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 808
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 809
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.021253325045108795,
+      "learning_rate": 2.0444444444444446e-05,
+      "loss": 0.0002,
+      "step": 810
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 810
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 811
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 812
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 813
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 814
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 815
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 816
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 817
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 818
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 819
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 0.003216712735593319,
+      "learning_rate": 2.014814814814815e-05,
+      "loss": 0.0259,
+      "step": 820
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 820
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 821
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 822
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 823
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 824
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 825
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 826
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 827
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.05,
+      "step": 828
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 829
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 44.6733512878418,
+      "learning_rate": 1.9851851851851855e-05,
+      "loss": 0.1754,
+      "step": 830
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 830
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 831
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 832
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 833
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 834
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 835
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 836
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 837
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 838
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 839
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.020853500813245773,
+      "learning_rate": 1.9555555555555557e-05,
+      "loss": 0.0022,
+      "step": 840
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 840
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 841
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 842
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 843
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 844
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 845
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 846
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 847
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 848
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 849
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.05716252699494362,
+      "learning_rate": 1.925925925925926e-05,
+      "loss": 0.0006,
+      "step": 850
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 850
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 851
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 852
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 853
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 854
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 855
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 856
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 857
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 858
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 859
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.004543145187199116,
+      "learning_rate": 1.8962962962962966e-05,
+      "loss": 0.0021,
+      "step": 860
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 860
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 861
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 862
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 863
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 864
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 865
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 866
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 867
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 868
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 869
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.0018642413197085261,
+      "learning_rate": 1.866666666666667e-05,
+      "loss": 0.0002,
+      "step": 870
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 870
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 871
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 872
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 873
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 874
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.08,
+      "step": 875
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 876
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 877
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 878
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 879
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.00903029553592205,
+      "learning_rate": 1.837037037037037e-05,
+      "loss": 0.2642,
+      "step": 880
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 880
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 881
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 882
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 883
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 884
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 885
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 886
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 887
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 888
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 889
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.00872717797756195,
+      "learning_rate": 1.8074074074074074e-05,
+      "loss": 0.0002,
+      "step": 890
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 890
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 891
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 892
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 893
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 894
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 895
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 896
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 897
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 898
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 899
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.007039394695311785,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.0314,
+      "step": 900
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 900
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 901
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 902
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 903
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 904
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 905
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 906
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 907
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 908
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 909
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.014733116142451763,
+      "learning_rate": 1.7481481481481483e-05,
+      "loss": 0.0002,
+      "step": 910
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 910
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 911
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 912
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 913
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 914
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 915
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 916
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 917
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 918
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 919
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.003551606787368655,
+      "learning_rate": 1.7185185185185185e-05,
+      "loss": 0.0003,
+      "step": 920
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 920
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 921
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 922
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 923
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 924
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 925
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 926
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 927
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 928
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 929
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.0028077142778784037,
+      "learning_rate": 1.688888888888889e-05,
+      "loss": 0.0049,
+      "step": 930
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 930
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 931
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 932
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 933
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 934
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 935
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 936
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 937
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 938
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 939
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.0034145561512559652,
+      "learning_rate": 1.6592592592592594e-05,
+      "loss": 0.0011,
+      "step": 940
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 940
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 941
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 942
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 943
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 944
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 945
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 946
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 947
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 948
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 949
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 2.191793203353882,
+      "learning_rate": 1.6296296296296297e-05,
+      "loss": 0.0029,
+      "step": 950
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 950
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 951
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 952
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 953
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 954
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 955
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 956
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 957
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 958
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 959
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.014660724438726902,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.0005,
+      "step": 960
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 960
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 961
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 962
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 963
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 964
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 965
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 966
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 967
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 968
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 969
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.0022546437103301287,
+      "learning_rate": 1.5703703703703705e-05,
+      "loss": 0.0002,
+      "step": 970
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 970
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 971
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 972
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 973
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 974
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 975
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 976
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 977
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 978
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 979
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.01155735645443201,
+      "learning_rate": 1.5407407407407408e-05,
+      "loss": 0.0005,
+      "step": 980
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 980
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 981
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 982
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 983
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 984
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 985
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 986
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 987
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 988
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 989
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.011351328343153,
+      "learning_rate": 1.5111111111111112e-05,
+      "loss": 0.0004,
+      "step": 990
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 990
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 991
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 992
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 993
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 994
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 995
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 996
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 997
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 998
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 999
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.008086876943707466,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.0001,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "epoch": 3.17,
+      "eval_accuracy": 0.94,
+      "eval_loss": 0.28937509655952454,
+      "eval_runtime": 119.0522,
+      "eval_samples_per_second": 0.84,
+      "eval_steps_per_second": 0.84,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 1000
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.0,
+      "step": 1001
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.0,
+      "step": 1002
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.0,
+      "step": 1003
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.0,
+      "step": 1004
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.0,
+      "step": 1005
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.0,
+      "step": 1006
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.0,
+      "step": 1007
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1008
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1009
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.018452847376465797,
+      "learning_rate": 1.4518518518518521e-05,
+      "loss": 0.0019,
+      "step": 1010
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1010
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1011
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1012
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1013
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1014
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1015
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1016
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1017
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1018
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1019
+    },
+    {
+      "epoch": 4.01,
+      "grad_norm": 0.002972155110910535,
+      "learning_rate": 1.4222222222222224e-05,
+      "loss": 0.0016,
+      "step": 1020
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1020
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1021
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 1022
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1023
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1024
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1025
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1026
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1027
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1028
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1029
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.004539242014288902,
+      "learning_rate": 1.3925925925925928e-05,
+      "loss": 0.0001,
+      "step": 1030
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1030
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1031
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1032
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1033
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1034
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1035
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1036
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 1037
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1038
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1039
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 0.0034846500493586063,
+      "learning_rate": 1.362962962962963e-05,
+      "loss": 0.0005,
+      "step": 1040
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1040
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1041
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1042
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1043
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1044
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1045
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1046
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1047
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1048
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1049
+    },
+    {
+      "epoch": 4.03,
+      "grad_norm": 12.088410377502441,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.008,
+      "step": 1050
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1050
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1051
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 1052
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1053
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1054
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1055
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1056
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1057
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1058
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1059
+    },
+    {
+      "epoch": 4.04,
+      "grad_norm": 0.049480143934488297,
+      "learning_rate": 1.303703703703704e-05,
+      "loss": 0.0003,
+      "step": 1060
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1060
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1061
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1062
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1063
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1064
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1065
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1066
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 1067
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1068
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1069
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.13532185554504395,
+      "learning_rate": 1.2740740740740742e-05,
+      "loss": 0.0006,
+      "step": 1070
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1070
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1071
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1072
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1073
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1074
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1075
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1076
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1077
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1078
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1079
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.005844641476869583,
+      "learning_rate": 1.2444444444444446e-05,
+      "loss": 0.0002,
+      "step": 1080
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1080
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1081
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 1082
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1083
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1084
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1085
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1086
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1087
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1088
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1089
+    },
+    {
+      "epoch": 4.06,
+      "grad_norm": 0.013352198526263237,
+      "learning_rate": 1.2148148148148149e-05,
+      "loss": 0.0004,
+      "step": 1090
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1090
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1091
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1092
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1093
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1094
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1095
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1096
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 1097
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1098
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1099
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.028877543285489082,
+      "learning_rate": 1.1851851851851852e-05,
+      "loss": 0.0003,
+      "step": 1100
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1100
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1101
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1102
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1103
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1104
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1105
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1106
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1107
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1108
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1109
+    },
+    {
+      "epoch": 4.07,
+      "grad_norm": 0.003447836497798562,
+      "learning_rate": 1.1555555555555556e-05,
+      "loss": 0.0004,
+      "step": 1110
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1110
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1111
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 1112
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1113
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1114
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1115
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1116
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1117
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1118
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1119
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.0029898551292717457,
+      "learning_rate": 1.125925925925926e-05,
+      "loss": 0.0001,
+      "step": 1120
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1120
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1121
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1122
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1123
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1124
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1125
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 1127
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1128
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1129
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.002433036919683218,
+      "learning_rate": 1.0962962962962965e-05,
+      "loss": 0.0001,
+      "step": 1130
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1130
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1131
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1132
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1133
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1134
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1135
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1136
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1137
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1138
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1139
+    },
+    {
+      "epoch": 4.09,
+      "grad_norm": 0.0014641088200733066,
+      "learning_rate": 1.0666666666666667e-05,
+      "loss": 0.0001,
+      "step": 1140
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1140
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 4.09,
+      "step": 1141
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 1142
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1143
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1144
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1145
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1146
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1147
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1148
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1149
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.0023359342012554407,
+      "learning_rate": 1.037037037037037e-05,
+      "loss": 0.1819,
+      "step": 1150
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1150
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1151
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1152
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1153
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1154
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1155
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1156
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 1157
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1158
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1159
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 0.0064132967963814735,
+      "learning_rate": 1.0074074074074074e-05,
+      "loss": 0.001,
+      "step": 1160
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1160
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1161
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1162
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1163
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1164
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1165
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1166
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1167
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1168
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1169
+    },
+    {
+      "epoch": 4.11,
+      "grad_norm": 4.7327446937561035,
+      "learning_rate": 9.777777777777779e-06,
+      "loss": 0.0053,
+      "step": 1170
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1170
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1171
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 1172
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1173
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1174
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1175
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1176
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1177
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1178
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1179
+    },
+    {
+      "epoch": 4.12,
+      "grad_norm": 0.011785696260631084,
+      "learning_rate": 9.481481481481483e-06,
+      "loss": 0.0002,
+      "step": 1180
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1180
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1181
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1182
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1183
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1184
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1185
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1186
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 1187
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1188
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1189
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.002919859951362014,
+      "learning_rate": 9.185185185185186e-06,
+      "loss": 0.0002,
+      "step": 1190
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1190
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1191
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1192
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1193
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1194
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1195
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1196
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1197
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1198
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1199
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.0023338980972766876,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.0004,
+      "step": 1200
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1200
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1201
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 1202
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1203
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1204
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1205
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1206
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1207
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1208
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1209
+    },
+    {
+      "epoch": 4.14,
+      "grad_norm": 0.001431985991075635,
+      "learning_rate": 8.592592592592593e-06,
+      "loss": 0.0003,
+      "step": 1210
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1210
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1211
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1212
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1213
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1214
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1215
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1216
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 1217
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1218
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1219
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.009923696517944336,
+      "learning_rate": 8.296296296296297e-06,
+      "loss": 0.0002,
+      "step": 1220
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1220
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1221
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1222
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1223
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1224
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1225
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1226
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1227
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1228
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1229
+    },
+    {
+      "epoch": 4.15,
+      "grad_norm": 0.0034356003161519766,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.0001,
+      "step": 1230
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1230
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1231
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 1232
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1233
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1234
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1235
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1236
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1237
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1238
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1239
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.004799115937203169,
+      "learning_rate": 7.703703703703704e-06,
+      "loss": 0.0002,
+      "step": 1240
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1240
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1241
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1242
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1243
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1244
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1245
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1246
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 1247
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1248
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1249
+    },
+    {
+      "epoch": 4.17,
+      "grad_norm": 0.0013780761510133743,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.0001,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "epoch": 4.17,
+      "eval_accuracy": 0.94,
+      "eval_loss": 0.29219138622283936,
+      "eval_runtime": 117.146,
+      "eval_samples_per_second": 0.854,
+      "eval_steps_per_second": 0.854,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 1250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.0,
+      "step": 1251
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.0,
+      "step": 1252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.0,
+      "step": 1253
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.0,
+      "step": 1254
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.0,
+      "step": 1255
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.0,
+      "step": 1256
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.0,
+      "step": 1257
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1258
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1259
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.0013477198081091046,
+      "learning_rate": 7.111111111111112e-06,
+      "loss": 0.0001,
+      "step": 1260
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1260
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1261
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1262
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1263
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1264
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1265
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1266
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1267
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1268
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1269
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.002066924935206771,
+      "learning_rate": 6.814814814814815e-06,
+      "loss": 0.0002,
+      "step": 1270
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1270
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1271
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 1272
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1273
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1274
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1275
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1276
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1277
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1278
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1279
+    },
+    {
+      "epoch": 5.02,
+      "grad_norm": 0.07563885301351547,
+      "learning_rate": 6.51851851851852e-06,
+      "loss": 0.0003,
+      "step": 1280
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1280
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1281
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1282
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1283
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1284
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1285
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1286
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 1287
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1288
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1289
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.00192207470536232,
+      "learning_rate": 6.222222222222223e-06,
+      "loss": 0.0002,
+      "step": 1290
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1290
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1291
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1292
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1293
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1294
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1295
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1296
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1297
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1298
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1299
+    },
+    {
+      "epoch": 5.03,
+      "grad_norm": 0.0015962225152179599,
+      "learning_rate": 5.925925925925926e-06,
+      "loss": 0.0001,
+      "step": 1300
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1300
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1301
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 1302
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1303
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1304
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1305
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1306
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1307
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1308
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1309
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.006703606341034174,
+      "learning_rate": 5.62962962962963e-06,
+      "loss": 0.0026,
+      "step": 1310
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1310
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1311
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1312
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1313
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1314
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1316
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 1317
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1318
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1319
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 0.0024196472950279713,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 0.0002,
+      "step": 1320
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1320
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1321
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1322
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1323
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1324
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1325
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1326
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1327
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1328
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1329
+    },
+    {
+      "epoch": 5.05,
+      "grad_norm": 0.002841345267370343,
+      "learning_rate": 5.037037037037037e-06,
+      "loss": 0.0001,
+      "step": 1330
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1330
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1331
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 1332
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1333
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1334
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1335
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1336
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1337
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1338
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1339
+    },
+    {
+      "epoch": 5.06,
+      "grad_norm": 0.03741031140089035,
+      "learning_rate": 4.7407407407407415e-06,
+      "loss": 0.0011,
+      "step": 1340
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1340
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1341
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1342
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1343
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1344
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1345
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1346
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 1347
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1348
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1349
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 0.003749188268557191,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.0003,
+      "step": 1350
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1350
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1351
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1352
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1353
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1354
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1355
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1356
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1357
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1358
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1359
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 0.05092022195458412,
+      "learning_rate": 4.1481481481481485e-06,
+      "loss": 0.0002,
+      "step": 1360
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1360
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1361
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 1362
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1363
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1364
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1365
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1366
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1367
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1368
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1369
+    },
+    {
+      "epoch": 5.08,
+      "grad_norm": 0.001492610084824264,
+      "learning_rate": 3.851851851851852e-06,
+      "loss": 0.0001,
+      "step": 1370
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1370
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1371
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1373
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1374
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1375
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1376
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 1377
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1378
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1379
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.006755433510988951,
+      "learning_rate": 3.555555555555556e-06,
+      "loss": 0.0001,
+      "step": 1380
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1380
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1381
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1382
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1383
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1384
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1385
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1386
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1387
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1388
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1389
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.008360594511032104,
+      "learning_rate": 3.25925925925926e-06,
+      "loss": 0.0001,
+      "step": 1390
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1390
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1391
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 1392
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1393
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1394
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1395
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1396
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1397
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1398
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1399
+    },
+    {
+      "epoch": 5.1,
+      "grad_norm": 0.002672309521585703,
+      "learning_rate": 2.962962962962963e-06,
+      "loss": 0.0005,
+      "step": 1400
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1400
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1401
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1402
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1403
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1404
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1405
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1406
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 1407
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1408
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1409
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 0.002087739761918783,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 0.0002,
+      "step": 1410
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1410
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1411
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1412
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1413
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1414
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1415
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1416
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1417
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1418
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1419
+    },
+    {
+      "epoch": 5.11,
+      "grad_norm": 0.016560420393943787,
+      "learning_rate": 2.3703703703703707e-06,
+      "loss": 0.0003,
+      "step": 1420
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1420
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1421
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 1422
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1423
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1424
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1425
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1426
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1427
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1428
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1429
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.002120854565873742,
+      "learning_rate": 2.0740740740740742e-06,
+      "loss": 0.0001,
+      "step": 1430
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1430
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1431
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1432
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1433
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1434
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1435
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1436
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 1437
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1438
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1439
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.03213360160589218,
+      "learning_rate": 1.777777777777778e-06,
+      "loss": 0.0003,
+      "step": 1440
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1440
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1441
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1442
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1443
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1444
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1445
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1446
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1447
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1448
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1449
+    },
+    {
+      "epoch": 5.13,
+      "grad_norm": 0.003606037702411413,
+      "learning_rate": 1.4814814814814815e-06,
+      "loss": 0.0027,
+      "step": 1450
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1450
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1451
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 1452
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1453
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1454
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1455
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1456
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1457
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1458
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1459
+    },
+    {
+      "epoch": 5.14,
+      "grad_norm": 15.364974021911621,
+      "learning_rate": 1.1851851851851854e-06,
+      "loss": 0.0164,
+      "step": 1460
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1460
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1461
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1462
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1463
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1464
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1465
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1466
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 1467
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1468
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1469
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.003104919334873557,
+      "learning_rate": 8.88888888888889e-07,
+      "loss": 0.0003,
+      "step": 1470
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1470
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1471
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1472
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1473
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1474
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1475
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1476
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1477
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1478
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1479
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.001645658747293055,
+      "learning_rate": 5.925925925925927e-07,
+      "loss": 0.0009,
+      "step": 1480
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1480
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1481
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 1482
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1483
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1484
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1485
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1486
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1487
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1488
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1489
+    },
+    {
+      "epoch": 5.16,
+      "grad_norm": 0.004309391137212515,
+      "learning_rate": 2.9629629629629634e-07,
+      "loss": 0.0001,
+      "step": 1490
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1490
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1491
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1492
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1493
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1494
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1495
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1496
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.16,
+      "step": 1497
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1498
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1499
+    },
+    {
+      "epoch": 5.17,
+      "grad_norm": 0.00799388438463211,
+      "learning_rate": 0.0,
+      "loss": 0.002,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "epoch": 5.17,
+      "eval_accuracy": 0.93,
+      "eval_loss": 0.31403449177742004,
+      "eval_runtime": 118.3792,
+      "eval_samples_per_second": 0.845,
+      "eval_steps_per_second": 0.845,
+      "step": 1500
+    },
+    {
+      "epoch": 5.17,
+      "step": 1500,
+      "total_flos": 1.16251514330112e+18,
+      "train_loss": 0.18969192462581366,
+      "train_runtime": 2692.2347,
+      "train_samples_per_second": 0.557,
+      "train_steps_per_second": 0.557
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "epoch": 5.17,
+      "eval_accuracy": 0.94,
+      "eval_loss": 0.18728949129581451,
+      "eval_runtime": 123.1389,
+      "eval_samples_per_second": 0.812,
+      "eval_steps_per_second": 0.812,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.17,
+      "step": 1500
+    },
+    {
+      "epoch": 5.17,
+      "eval_accuracy": 0.94,
+      "eval_loss": 0.1872895061969757,
+      "eval_runtime": 125.4207,
+      "eval_samples_per_second": 0.797,
+      "eval_steps_per_second": 0.797,
+      "step": 1500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 1500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "total_flos": 1.16251514330112e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}