End of training

all_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 5.15,
+    "eval_accuracy": 0.95,
+    "eval_loss": 0.14155207574367523,
+    "eval_runtime": 83.8267,
+    "eval_samples_per_second": 1.193,
+    "eval_steps_per_second": 0.298
+}

runs/Apr21_05-00-58_27f25dae16c5/events.out.tfevents.1713677885.27f25dae16c5.2738.1

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:82a620a7edd3f0eb9a31534ae189f31150ea46edd9969233bdd849e3fa0355f1
-size 2986
+oid sha256:0944b46f84ec992280936cf68178db79cc471e5f5a78f40b89690cb90b608915
+size 5884

test_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 5.15,
+    "eval_accuracy": 0.95,
+    "eval_loss": 0.14155207574367523,
+    "eval_runtime": 83.8267,
+    "eval_samples_per_second": 1.193,
+    "eval_steps_per_second": 0.298
+}

trainer_state.json

@@ -0,0 +1,3221 @@
+{
+  "best_metric": 0.98,
+  "best_model_checkpoint": "timesformer-base-finetuned-k400-kinetic400-subset-epoch6-num_frame_10_myViT2window_more_data_b4/checkpoint-63",
+  "epoch": 5.153225806451613,
+  "eval_steps": 500,
+  "global_step": 372,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "accuracy": 0.5,
+      "epoch": 0,
+      "step": 0
+    },
+    {
+      "accuracy": 0.25,
+      "epoch": 0.0,
+      "step": 1
+    },
+    {
+      "accuracy": 0.25,
+      "epoch": 0.01,
+      "step": 2
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 3
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.01,
+      "step": 4
+    },
+    {
+      "accuracy": 0.25,
+      "epoch": 0.01,
+      "step": 5
+    },
+    {
+      "accuracy": 0.5,
+      "epoch": 0.02,
+      "step": 6
+    },
+    {
+      "accuracy": 0.25,
+      "epoch": 0.02,
+      "step": 7
+    },
+    {
+      "accuracy": 0.5,
+      "epoch": 0.02,
+      "step": 8
+    },
+    {
+      "accuracy": 0.0,
+      "epoch": 0.02,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 20.43199920654297,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 1.6522,
+      "step": 10
+    },
+    {
+      "accuracy": 0.25,
+      "epoch": 0.03,
+      "step": 10
+    },
+    {
+      "accuracy": 0.5,
+      "epoch": 0.03,
+      "step": 11
+    },
+    {
+      "accuracy": 0.25,
+      "epoch": 0.03,
+      "step": 12
+    },
+    {
+      "accuracy": 0.5,
+      "epoch": 0.03,
+      "step": 13
+    },
+    {
+      "accuracy": 0.5,
+      "epoch": 0.04,
+      "step": 14
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.04,
+      "step": 15
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.04,
+      "step": 16
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.05,
+      "step": 17
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.05,
+      "step": 18
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 9.460710525512695,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 1.1214,
+      "step": 20
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.05,
+      "step": 20
+    },
+    {
+      "accuracy": 0.5,
+      "epoch": 0.06,
+      "step": 21
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 22
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.06,
+      "step": 23
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.06,
+      "step": 24
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 25
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.07,
+      "step": 26
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.07,
+      "step": 27
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 28
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.08,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 9.38272762298584,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.506,
+      "step": 30
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.08,
+      "step": 30
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.08,
+      "step": 31
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 32
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.09,
+      "step": 33
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.09,
+      "step": 34
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.09,
+      "step": 35
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.1,
+      "step": 36
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.1,
+      "step": 37
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.1,
+      "step": 38
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.1,
+      "step": 39
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 13.422501564025879,
+      "learning_rate": 4.970059880239521e-05,
+      "loss": 0.4909,
+      "step": 40
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 40
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.11,
+      "step": 41
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.11,
+      "step": 42
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 43
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 44
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 45
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.12,
+      "step": 46
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 47
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.13,
+      "step": 48
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.13,
+      "step": 49
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 12.859742164611816,
+      "learning_rate": 4.820359281437126e-05,
+      "loss": 0.175,
+      "step": 50
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.13,
+      "step": 50
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 51
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 52
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.14,
+      "step": 53
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 54
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.15,
+      "step": 55
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 56
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.15,
+      "step": 57
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.16,
+      "step": 58
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 59
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 3.605234384536743,
+      "learning_rate": 4.670658682634731e-05,
+      "loss": 0.2183,
+      "step": 60
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.16,
+      "step": 60
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.16,
+      "step": 61
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 62
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "epoch": 0.17,
+      "eval_accuracy": 0.98,
+      "eval_loss": 0.10602467507123947,
+      "eval_runtime": 82.1101,
+      "eval_samples_per_second": 1.218,
+      "eval_steps_per_second": 0.304,
+      "step": 63
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 0.17,
+      "step": 63
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.0,
+      "step": 64
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 65
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.01,
+      "step": 66
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 67
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.01,
+      "step": 68
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 69
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.08548454195261002,
+      "learning_rate": 4.5209580838323355e-05,
+      "loss": 0.2839,
+      "step": 70
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 70
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 71
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.02,
+      "step": 72
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 73
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 74
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 75
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.03,
+      "step": 76
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 77
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 78
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.04,
+      "step": 79
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.060699224472046,
+      "learning_rate": 4.3712574850299406e-05,
+      "loss": 0.0087,
+      "step": 80
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 80
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 81
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 82
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.05,
+      "step": 83
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.06,
+      "step": 84
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 85
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 86
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.06,
+      "step": 87
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 88
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 89
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.10145212709903717,
+      "learning_rate": 4.221556886227545e-05,
+      "loss": 0.0395,
+      "step": 90
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.07,
+      "step": 90
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 91
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 92
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 93
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.08,
+      "step": 94
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 95
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 96
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 97
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.09,
+      "step": 98
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.1,
+      "step": 99
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 15.697983741760254,
+      "learning_rate": 4.07185628742515e-05,
+      "loss": 0.0766,
+      "step": 100
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.1,
+      "step": 100
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 101
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.1,
+      "step": 102
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 103
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 104
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.11,
+      "step": 105
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 106
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 107
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 108
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.12,
+      "step": 109
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.3895547688007355,
+      "learning_rate": 3.9221556886227544e-05,
+      "loss": 0.0845,
+      "step": 110
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 110
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 111
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 112
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.13,
+      "step": 113
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.14,
+      "step": 114
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 115
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.14,
+      "step": 116
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 117
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 118
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 119
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 3.2577812671661377,
+      "learning_rate": 3.7724550898203595e-05,
+      "loss": 0.0407,
+      "step": 120
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.15,
+      "step": 120
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 121
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 122
+    },
+    {
+      "accuracy": 0.5,
+      "epoch": 1.16,
+      "step": 123
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.16,
+      "step": 124
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 125
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "epoch": 1.17,
+      "eval_accuracy": 0.96,
+      "eval_loss": 0.14332203567028046,
+      "eval_runtime": 78.0385,
+      "eval_samples_per_second": 1.281,
+      "eval_steps_per_second": 0.32,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 1.17,
+      "step": 126
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.0,
+      "step": 127
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 128
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 129
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.02145831100642681,
+      "learning_rate": 3.6227544910179645e-05,
+      "loss": 0.0417,
+      "step": 130
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 130
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.01,
+      "step": 131
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 132
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 133
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 134
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.02,
+      "step": 135
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 136
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 137
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.03,
+      "step": 138
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 2.03,
+      "step": 139
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 18.207517623901367,
+      "learning_rate": 3.473053892215569e-05,
+      "loss": 0.0238,
+      "step": 140
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 140
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 141
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.04,
+      "step": 142
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 143
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 144
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 145
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.05,
+      "step": 146
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 147
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 148
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 149
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.02513033337891102,
+      "learning_rate": 3.323353293413174e-05,
+      "loss": 0.0035,
+      "step": 150
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.06,
+      "step": 150
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 151
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 152
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.07,
+      "step": 153
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 154
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 155
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 156
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.08,
+      "step": 157
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 158
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 159
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 5.617037773132324,
+      "learning_rate": 3.1736526946107784e-05,
+      "loss": 0.0078,
+      "step": 160
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 160
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.09,
+      "step": 161
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 162
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 163
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 164
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.1,
+      "step": 165
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 166
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 167
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.11,
+      "step": 168
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 169
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.052510615438222885,
+      "learning_rate": 3.0239520958083834e-05,
+      "loss": 0.0015,
+      "step": 170
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 170
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 171
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.12,
+      "step": 172
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 173
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 174
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 175
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.13,
+      "step": 176
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 177
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 178
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.14,
+      "step": 179
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.025016725063323975,
+      "learning_rate": 2.874251497005988e-05,
+      "loss": 0.0015,
+      "step": 180
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 180
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 181
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 182
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.15,
+      "step": 183
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 184
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 185
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 186
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.16,
+      "step": 187
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 188
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "epoch": 2.17,
+      "eval_accuracy": 0.97,
+      "eval_loss": 0.11863651126623154,
+      "eval_runtime": 78.3799,
+      "eval_samples_per_second": 1.276,
+      "eval_steps_per_second": 0.319,
+      "step": 189
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 2.17,
+      "step": 189
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.019851716235280037,
+      "learning_rate": 2.724550898203593e-05,
+      "loss": 0.0039,
+      "step": 190
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.0,
+      "step": 190
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 191
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 192
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 193
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.01,
+      "step": 194
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 195
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 196
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 197
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.02,
+      "step": 198
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 199
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 0.10042139142751694,
+      "learning_rate": 2.5748502994011976e-05,
+      "loss": 0.0039,
+      "step": 200
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 200
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 201
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.03,
+      "step": 202
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 203
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 3.04,
+      "step": 204
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.04,
+      "step": 205
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 206
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 207
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 208
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.05,
+      "step": 209
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.016782386228442192,
+      "learning_rate": 2.4251497005988023e-05,
+      "loss": 0.1046,
+      "step": 210
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 210
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 211
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 212
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.06,
+      "step": 213
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 214
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 215
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.07,
+      "step": 216
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 217
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 218
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.08,
+      "step": 219
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.00896859634667635,
+      "learning_rate": 2.275449101796407e-05,
+      "loss": 0.0063,
+      "step": 220
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 3.08,
+      "step": 220
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 221
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 222
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 223
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.09,
+      "step": 224
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 225
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 226
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 227
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.1,
+      "step": 228
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 229
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.4414406418800354,
+      "learning_rate": 2.125748502994012e-05,
+      "loss": 0.0333,
+      "step": 230
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 230
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.11,
+      "step": 231
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 232
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 233
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 234
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.12,
+      "step": 235
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 236
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 237
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 238
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.13,
+      "step": 239
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.010348650626838207,
+      "learning_rate": 1.9760479041916168e-05,
+      "loss": 0.0041,
+      "step": 240
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 240
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 241
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.14,
+      "step": 242
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 243
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 244
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 245
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.15,
+      "step": 246
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 3.16,
+      "step": 247
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 248
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 249
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.011760951951146126,
+      "learning_rate": 1.8263473053892215e-05,
+      "loss": 0.0257,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.16,
+      "step": 250
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 251
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "epoch": 3.17,
+      "eval_accuracy": 0.97,
+      "eval_loss": 0.1484651118516922,
+      "eval_runtime": 78.9052,
+      "eval_samples_per_second": 1.267,
+      "eval_steps_per_second": 0.317,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 3.17,
+      "step": 252
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.0,
+      "step": 253
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 254
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 255
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 256
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.01,
+      "step": 257
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 258
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 259
+    },
+    {
+      "epoch": 4.02,
+      "grad_norm": 0.011730861850082874,
+      "learning_rate": 1.6766467065868263e-05,
+      "loss": 0.0008,
+      "step": 260
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 260
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.02,
+      "step": 261
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 262
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 263
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 264
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.03,
+      "step": 265
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 266
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 267
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.04,
+      "step": 268
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 269
+    },
+    {
+      "epoch": 4.05,
+      "grad_norm": 0.05271562933921814,
+      "learning_rate": 1.5269461077844313e-05,
+      "loss": 0.001,
+      "step": 270
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 270
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 271
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.05,
+      "step": 272
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 273
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 274
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 275
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.06,
+      "step": 276
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 277
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 278
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.07,
+      "step": 279
+    },
+    {
+      "epoch": 4.08,
+      "grad_norm": 0.008417515084147453,
+      "learning_rate": 1.377245508982036e-05,
+      "loss": 0.0012,
+      "step": 280
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 280
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 281
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 282
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.08,
+      "step": 283
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 284
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 285
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 286
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.09,
+      "step": 287
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 288
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 289
+    },
+    {
+      "epoch": 4.1,
+      "grad_norm": 0.012302696704864502,
+      "learning_rate": 1.2275449101796408e-05,
+      "loss": 0.001,
+      "step": 290
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 290
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.1,
+      "step": 291
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 292
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 293
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.11,
+      "step": 294
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 295
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 296
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 297
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.12,
+      "step": 298
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 299
+    },
+    {
+      "epoch": 4.13,
+      "grad_norm": 0.012150724418461323,
+      "learning_rate": 1.0778443113772455e-05,
+      "loss": 0.0006,
+      "step": 300
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 300
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 301
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.13,
+      "step": 302
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 303
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 304
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.14,
+      "step": 305
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 306
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 307
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 308
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.15,
+      "step": 309
+    },
+    {
+      "epoch": 4.16,
+      "grad_norm": 0.007512472104281187,
+      "learning_rate": 9.281437125748502e-06,
+      "loss": 0.0007,
+      "step": 310
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 310
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 311
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 312
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.16,
+      "step": 313
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 314
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "epoch": 4.17,
+      "eval_accuracy": 0.96,
+      "eval_loss": 0.11020946502685547,
+      "eval_runtime": 78.3612,
+      "eval_samples_per_second": 1.276,
+      "eval_steps_per_second": 0.319,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 4.17,
+      "step": 315
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.0,
+      "step": 316
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 317
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 318
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 319
+    },
+    {
+      "epoch": 5.01,
+      "grad_norm": 0.008234703913331032,
+      "learning_rate": 7.784431137724551e-06,
+      "loss": 0.0006,
+      "step": 320
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.01,
+      "step": 320
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 321
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 322
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 323
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.02,
+      "step": 324
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 325
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 326
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 327
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.03,
+      "step": 328
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 329
+    },
+    {
+      "epoch": 5.04,
+      "grad_norm": 0.030246170237660408,
+      "learning_rate": 6.2874251497005985e-06,
+      "loss": 0.0043,
+      "step": 330
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 330
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.04,
+      "step": 331
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 332
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 333
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 334
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.05,
+      "step": 335
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 336
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 337
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 338
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.06,
+      "step": 339
+    },
+    {
+      "epoch": 5.07,
+      "grad_norm": 0.2442082017660141,
+      "learning_rate": 4.7904191616766475e-06,
+      "loss": 0.0016,
+      "step": 340
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 340
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 341
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.07,
+      "step": 342
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 343
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 344
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 345
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.08,
+      "step": 346
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 347
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 348
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 349
+    },
+    {
+      "epoch": 5.09,
+      "grad_norm": 0.045361634343862534,
+      "learning_rate": 3.293413173652695e-06,
+      "loss": 0.0006,
+      "step": 350
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.09,
+      "step": 350
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 351
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 352
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 353
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.1,
+      "step": 354
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 355
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 356
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.11,
+      "step": 357
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 358
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 359
+    },
+    {
+      "epoch": 5.12,
+      "grad_norm": 0.005745638627558947,
+      "learning_rate": 1.7964071856287426e-06,
+      "loss": 0.0005,
+      "step": 360
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 360
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.12,
+      "step": 361
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 362
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 363
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 364
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.13,
+      "step": 365
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 366
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 367
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.14,
+      "step": 368
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 369
+    },
+    {
+      "epoch": 5.15,
+      "grad_norm": 0.008824752643704414,
+      "learning_rate": 2.9940119760479047e-07,
+      "loss": 0.0008,
+      "step": 370
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 370
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 371
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "epoch": 5.15,
+      "eval_accuracy": 0.97,
+      "eval_loss": 0.10981950163841248,
+      "eval_runtime": 78.1073,
+      "eval_samples_per_second": 1.28,
+      "eval_steps_per_second": 0.32,
+      "step": 372
+    },
+    {
+      "epoch": 5.15,
+      "step": 372,
+      "total_flos": 1.6187334525160243e+18,
+      "train_loss": 0.13372177068865107,
+      "train_runtime": 2192.2823,
+      "train_samples_per_second": 0.679,
+      "train_steps_per_second": 0.17
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "epoch": 5.15,
+      "eval_accuracy": 0.95,
+      "eval_loss": 0.14155207574367523,
+      "eval_runtime": 81.9579,
+      "eval_samples_per_second": 1.22,
+      "eval_steps_per_second": 0.305,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 0.75,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "accuracy": 1.0,
+      "epoch": 5.15,
+      "step": 372
+    },
+    {
+      "epoch": 5.15,
+      "eval_accuracy": 0.95,
+      "eval_loss": 0.14155207574367523,
+      "eval_runtime": 83.8267,
+      "eval_samples_per_second": 1.193,
+      "eval_steps_per_second": 0.298,
+      "step": 372
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 372,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "total_flos": 1.6187334525160243e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}