diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26812 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.885304659498208,
+  "global_step": 4400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.997759856630825e-05,
+      "loss": 1.9458,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.995519713261649e-05,
+      "loss": 2.0693,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.993279569892474e-05,
+      "loss": 2.0732,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.991039426523298e-05,
+      "loss": 2.2314,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988799283154122e-05,
+      "loss": 2.3916,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.986559139784947e-05,
+      "loss": 2.02,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.984318996415771e-05,
+      "loss": 1.9331,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.982078853046596e-05,
+      "loss": 1.9219,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.97983870967742e-05,
+      "loss": 2.3125,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.977598566308244e-05,
+      "loss": 2.1377,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.975358422939069e-05,
+      "loss": 2.1064,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.973118279569893e-05,
+      "loss": 2.0078,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.970878136200718e-05,
+      "loss": 1.9814,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.968637992831542e-05,
+      "loss": 1.9502,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.966397849462365e-05,
+      "loss": 1.9561,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.96415770609319e-05,
+      "loss": 1.9385,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.961917562724015e-05,
+      "loss": 1.9619,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.95967741935484e-05,
+      "loss": 1.8096,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.957437275985664e-05,
+      "loss": 1.8887,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.955197132616488e-05,
+      "loss": 2.0596,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.952956989247311e-05,
+      "loss": 2.0977,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.950716845878137e-05,
+      "loss": 1.875,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.948476702508961e-05,
+      "loss": 2.1445,
+      "step": 23
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.946236559139786e-05,
+      "loss": 2.3311,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.94399641577061e-05,
+      "loss": 2.166,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.941756272401435e-05,
+      "loss": 1.9375,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.939516129032258e-05,
+      "loss": 2.3115,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.937275985663082e-05,
+      "loss": 1.9727,
+      "step": 28
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.935035842293908e-05,
+      "loss": 1.769,
+      "step": 29
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.932795698924732e-05,
+      "loss": 1.7832,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930555555555556e-05,
+      "loss": 2.2773,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928315412186381e-05,
+      "loss": 1.7368,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.926075268817204e-05,
+      "loss": 2.0239,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.923835125448028e-05,
+      "loss": 2.3203,
+      "step": 34
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.921594982078854e-05,
+      "loss": 2.2314,
+      "step": 35
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.919354838709678e-05,
+      "loss": 2.3105,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.917114695340503e-05,
+      "loss": 1.8604,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.914874551971327e-05,
+      "loss": 1.8682,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.91263440860215e-05,
+      "loss": 1.8701,
+      "step": 39
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.910394265232975e-05,
+      "loss": 1.9448,
+      "step": 40
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9081541218638e-05,
+      "loss": 2.1826,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.905913978494625e-05,
+      "loss": 1.9546,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.903673835125449e-05,
+      "loss": 1.8872,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.901433691756273e-05,
+      "loss": 2.1553,
+      "step": 44
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.899193548387097e-05,
+      "loss": 1.9175,
+      "step": 45
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.896953405017921e-05,
+      "loss": 2.0146,
+      "step": 46
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.894713261648745e-05,
+      "loss": 1.8613,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892473118279571e-05,
+      "loss": 2.0918,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890232974910395e-05,
+      "loss": 1.8633,
+      "step": 49
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.88799283154122e-05,
+      "loss": 1.8799,
+      "step": 50
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.885752688172043e-05,
+      "loss": 1.7466,
+      "step": 51
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.883512544802867e-05,
+      "loss": 1.7817,
+      "step": 52
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.881272401433692e-05,
+      "loss": 1.709,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879032258064517e-05,
+      "loss": 2.334,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876792114695342e-05,
+      "loss": 1.9536,
+      "step": 55
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874551971326166e-05,
+      "loss": 1.9429,
+      "step": 56
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872311827956989e-05,
+      "loss": 1.7954,
+      "step": 57
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870071684587814e-05,
+      "loss": 2.2544,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.867831541218638e-05,
+      "loss": 1.9858,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.865591397849462e-05,
+      "loss": 1.7744,
+      "step": 60
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.863351254480288e-05,
+      "loss": 1.8828,
+      "step": 61
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.861111111111112e-05,
+      "loss": 1.8745,
+      "step": 62
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.858870967741935e-05,
+      "loss": 2.1504,
+      "step": 63
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85663082437276e-05,
+      "loss": 2.1758,
+      "step": 64
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.854390681003584e-05,
+      "loss": 1.7305,
+      "step": 65
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.852150537634409e-05,
+      "loss": 2.187,
+      "step": 66
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.849910394265234e-05,
+      "loss": 2.3789,
+      "step": 67
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.847670250896059e-05,
+      "loss": 1.9897,
+      "step": 68
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.845430107526882e-05,
+      "loss": 2.1738,
+      "step": 69
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843189964157706e-05,
+      "loss": 1.9429,
+      "step": 70
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84094982078853e-05,
+      "loss": 1.9395,
+      "step": 71
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.838709677419355e-05,
+      "loss": 1.9746,
+      "step": 72
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.836469534050181e-05,
+      "loss": 1.9277,
+      "step": 73
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.834229390681005e-05,
+      "loss": 2.0693,
+      "step": 74
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.831989247311828e-05,
+      "loss": 1.9346,
+      "step": 75
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.829749103942652e-05,
+      "loss": 1.8271,
+      "step": 76
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827508960573477e-05,
+      "loss": 1.6675,
+      "step": 77
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825268817204301e-05,
+      "loss": 1.895,
+      "step": 78
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823028673835126e-05,
+      "loss": 1.9624,
+      "step": 79
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.820788530465951e-05,
+      "loss": 2.0698,
+      "step": 80
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.818548387096774e-05,
+      "loss": 2.0029,
+      "step": 81
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.816308243727599e-05,
+      "loss": 1.8086,
+      "step": 82
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.814068100358423e-05,
+      "loss": 1.7808,
+      "step": 83
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.811827956989248e-05,
+      "loss": 1.8701,
+      "step": 84
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.809587813620072e-05,
+      "loss": 1.9805,
+      "step": 85
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.807347670250898e-05,
+      "loss": 2.0532,
+      "step": 86
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.805107526881721e-05,
+      "loss": 1.7168,
+      "step": 87
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.802867383512545e-05,
+      "loss": 1.8696,
+      "step": 88
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.80062724014337e-05,
+      "loss": 2.0068,
+      "step": 89
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.798387096774194e-05,
+      "loss": 1.9917,
+      "step": 90
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.796146953405018e-05,
+      "loss": 2.252,
+      "step": 91
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.793906810035843e-05,
+      "loss": 1.583,
+      "step": 92
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.791666666666667e-05,
+      "loss": 1.7402,
+      "step": 93
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.789426523297491e-05,
+      "loss": 2.2891,
+      "step": 94
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.787186379928316e-05,
+      "loss": 2.0659,
+      "step": 95
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.78494623655914e-05,
+      "loss": 1.646,
+      "step": 96
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.782706093189965e-05,
+      "loss": 1.5259,
+      "step": 97
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.780465949820789e-05,
+      "loss": 2.04,
+      "step": 98
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.778225806451613e-05,
+      "loss": 2.064,
+      "step": 99
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.775985663082438e-05,
+      "loss": 1.408,
+      "step": 100
+    },
+    {
+      "epoch": 0.18,
+      "eval_accuracy": 0.15714286267757416,
+      "eval_loss": 2.0277483463287354,
+      "eval_runtime": 67.7973,
+      "eval_samples_per_second": 4.13,
+      "eval_steps_per_second": 4.13,
+      "step": 100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.773745519713262e-05,
+      "loss": 2.3169,
+      "step": 101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.771505376344087e-05,
+      "loss": 1.5103,
+      "step": 102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.769265232974911e-05,
+      "loss": 2.377,
+      "step": 103
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.767025089605735e-05,
+      "loss": 2.1582,
+      "step": 104
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.76478494623656e-05,
+      "loss": 1.9458,
+      "step": 105
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.762544802867384e-05,
+      "loss": 1.7324,
+      "step": 106
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.760304659498208e-05,
+      "loss": 2.3589,
+      "step": 107
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.758064516129033e-05,
+      "loss": 2.5938,
+      "step": 108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.755824372759857e-05,
+      "loss": 1.8462,
+      "step": 109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.753584229390682e-05,
+      "loss": 1.5845,
+      "step": 110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.751344086021506e-05,
+      "loss": 2.543,
+      "step": 111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74910394265233e-05,
+      "loss": 2.0352,
+      "step": 112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.746863799283155e-05,
+      "loss": 1.9814,
+      "step": 113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.744623655913979e-05,
+      "loss": 2.2949,
+      "step": 114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.742383512544804e-05,
+      "loss": 1.7202,
+      "step": 115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.740143369175627e-05,
+      "loss": 2.1416,
+      "step": 116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.737903225806452e-05,
+      "loss": 2.0293,
+      "step": 117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.735663082437277e-05,
+      "loss": 2.0518,
+      "step": 118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.733422939068101e-05,
+      "loss": 2.2393,
+      "step": 119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.731182795698925e-05,
+      "loss": 1.6167,
+      "step": 120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.72894265232975e-05,
+      "loss": 2.0151,
+      "step": 121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.726702508960573e-05,
+      "loss": 1.6899,
+      "step": 122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.724462365591399e-05,
+      "loss": 2.5811,
+      "step": 123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.722222222222223e-05,
+      "loss": 1.7607,
+      "step": 124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.719982078853047e-05,
+      "loss": 2.3062,
+      "step": 125
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.717741935483872e-05,
+      "loss": 1.9917,
+      "step": 126
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.715501792114696e-05,
+      "loss": 2.1055,
+      "step": 127
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.713261648745519e-05,
+      "loss": 1.8516,
+      "step": 128
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.711021505376344e-05,
+      "loss": 1.7441,
+      "step": 129
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.70878136200717e-05,
+      "loss": 1.8398,
+      "step": 130
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.706541218637994e-05,
+      "loss": 2.2021,
+      "step": 131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.704301075268818e-05,
+      "loss": 1.9531,
+      "step": 132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.702060931899643e-05,
+      "loss": 2.0576,
+      "step": 133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.699820788530466e-05,
+      "loss": 1.979,
+      "step": 134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.69758064516129e-05,
+      "loss": 1.8149,
+      "step": 135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.695340501792116e-05,
+      "loss": 2.1973,
+      "step": 136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.69310035842294e-05,
+      "loss": 1.8628,
+      "step": 137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.690860215053764e-05,
+      "loss": 1.7559,
+      "step": 138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.688620071684589e-05,
+      "loss": 1.8296,
+      "step": 139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.686379928315412e-05,
+      "loss": 1.7002,
+      "step": 140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.684139784946236e-05,
+      "loss": 1.8628,
+      "step": 141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.681899641577062e-05,
+      "loss": 1.8823,
+      "step": 142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.679659498207886e-05,
+      "loss": 2.0132,
+      "step": 143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.677419354838711e-05,
+      "loss": 1.7051,
+      "step": 144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.675179211469535e-05,
+      "loss": 1.8511,
+      "step": 145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.672939068100358e-05,
+      "loss": 1.6753,
+      "step": 146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.670698924731183e-05,
+      "loss": 2.002,
+      "step": 147
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.668458781362007e-05,
+      "loss": 2.4492,
+      "step": 148
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.666218637992833e-05,
+      "loss": 1.939,
+      "step": 149
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.663978494623657e-05,
+      "loss": 1.9834,
+      "step": 150
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.661738351254481e-05,
+      "loss": 2.3945,
+      "step": 151
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.659498207885304e-05,
+      "loss": 1.7734,
+      "step": 152
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.657258064516129e-05,
+      "loss": 2.2705,
+      "step": 153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.655017921146953e-05,
+      "loss": 2.3125,
+      "step": 154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.652777777777779e-05,
+      "loss": 1.8203,
+      "step": 155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.650537634408603e-05,
+      "loss": 1.7852,
+      "step": 156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.648297491039428e-05,
+      "loss": 1.8438,
+      "step": 157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.646057347670251e-05,
+      "loss": 1.7598,
+      "step": 158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.643817204301075e-05,
+      "loss": 2.2593,
+      "step": 159
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.6415770609319e-05,
+      "loss": 1.9888,
+      "step": 160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.639336917562724e-05,
+      "loss": 1.9673,
+      "step": 161
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.63709677419355e-05,
+      "loss": 2.2012,
+      "step": 162
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.634856630824374e-05,
+      "loss": 2.0352,
+      "step": 163
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.632616487455197e-05,
+      "loss": 1.8564,
+      "step": 164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.630376344086022e-05,
+      "loss": 1.8345,
+      "step": 165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.628136200716846e-05,
+      "loss": 1.7202,
+      "step": 166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.62589605734767e-05,
+      "loss": 1.8677,
+      "step": 167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.623655913978496e-05,
+      "loss": 1.814,
+      "step": 168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.62141577060932e-05,
+      "loss": 1.9692,
+      "step": 169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.619175627240143e-05,
+      "loss": 2.0137,
+      "step": 170
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.616935483870968e-05,
+      "loss": 2.0977,
+      "step": 171
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.614695340501792e-05,
+      "loss": 2.1611,
+      "step": 172
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.612455197132617e-05,
+      "loss": 1.8267,
+      "step": 173
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.610215053763442e-05,
+      "loss": 1.7158,
+      "step": 174
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.607974910394267e-05,
+      "loss": 2.1685,
+      "step": 175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.60573476702509e-05,
+      "loss": 1.8545,
+      "step": 176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.603494623655914e-05,
+      "loss": 1.5874,
+      "step": 177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.601254480286739e-05,
+      "loss": 1.6948,
+      "step": 178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.599014336917563e-05,
+      "loss": 2.0156,
+      "step": 179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.596774193548387e-05,
+      "loss": 1.0918,
+      "step": 180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.594534050179213e-05,
+      "loss": 1.6011,
+      "step": 181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.592293906810036e-05,
+      "loss": 2.5806,
+      "step": 182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.59005376344086e-05,
+      "loss": 2.4238,
+      "step": 183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.587813620071685e-05,
+      "loss": 2.3491,
+      "step": 184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.585573476702509e-05,
+      "loss": 1.9224,
+      "step": 185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.583333333333334e-05,
+      "loss": 1.1726,
+      "step": 186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.581093189964158e-05,
+      "loss": 2.2979,
+      "step": 187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.578853046594982e-05,
+      "loss": 1.6709,
+      "step": 188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.576612903225807e-05,
+      "loss": 1.7559,
+      "step": 189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.574372759856631e-05,
+      "loss": 2.3813,
+      "step": 190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.572132616487456e-05,
+      "loss": 1.9302,
+      "step": 191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.56989247311828e-05,
+      "loss": 1.333,
+      "step": 192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.567652329749104e-05,
+      "loss": 1.5432,
+      "step": 193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.565412186379929e-05,
+      "loss": 1.6885,
+      "step": 194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.563172043010753e-05,
+      "loss": 1.9277,
+      "step": 195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.560931899641577e-05,
+      "loss": 1.6655,
+      "step": 196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.558691756272402e-05,
+      "loss": 2.293,
+      "step": 197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.556451612903226e-05,
+      "loss": 1.4375,
+      "step": 198
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.55421146953405e-05,
+      "loss": 1.604,
+      "step": 199
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.551971326164875e-05,
+      "loss": 3.1152,
+      "step": 200
+    },
+    {
+      "epoch": 0.36,
+      "eval_accuracy": 0.26428571343421936,
+      "eval_loss": 1.903194785118103,
+      "eval_runtime": 71.7605,
+      "eval_samples_per_second": 3.902,
+      "eval_steps_per_second": 3.902,
+      "step": 200
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.5497311827957e-05,
+      "loss": 1.5874,
+      "step": 201
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.547491039426524e-05,
+      "loss": 1.6455,
+      "step": 202
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.545250896057348e-05,
+      "loss": 1.752,
+      "step": 203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.543010752688173e-05,
+      "loss": 3.418,
+      "step": 204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.540770609318997e-05,
+      "loss": 2.5723,
+      "step": 205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.538530465949821e-05,
+      "loss": 1.3198,
+      "step": 206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.536290322580646e-05,
+      "loss": 2.6504,
+      "step": 207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.53405017921147e-05,
+      "loss": 2.2188,
+      "step": 208
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.531810035842295e-05,
+      "loss": 0.7317,
+      "step": 209
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.529569892473119e-05,
+      "loss": 1.3713,
+      "step": 210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.527329749103943e-05,
+      "loss": 1.1133,
+      "step": 211
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.525089605734768e-05,
+      "loss": 1.6272,
+      "step": 212
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.522849462365592e-05,
+      "loss": 1.9185,
+      "step": 213
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.520609318996416e-05,
+      "loss": 1.0336,
+      "step": 214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.518369175627241e-05,
+      "loss": 1.4431,
+      "step": 215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.516129032258065e-05,
+      "loss": 0.9397,
+      "step": 216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.513888888888888e-05,
+      "loss": 3.375,
+      "step": 217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.511648745519714e-05,
+      "loss": 2.541,
+      "step": 218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.509408602150538e-05,
+      "loss": 3.1904,
+      "step": 219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.507168458781363e-05,
+      "loss": 2.1523,
+      "step": 220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.504928315412187e-05,
+      "loss": 1.3022,
+      "step": 221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.502688172043012e-05,
+      "loss": 2.4785,
+      "step": 222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.500448028673835e-05,
+      "loss": 1.7271,
+      "step": 223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.49820788530466e-05,
+      "loss": 1.7183,
+      "step": 224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.495967741935485e-05,
+      "loss": 1.5359,
+      "step": 225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.493727598566309e-05,
+      "loss": 2.3042,
+      "step": 226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.491487455197133e-05,
+      "loss": 1.8901,
+      "step": 227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.489247311827958e-05,
+      "loss": 1.9575,
+      "step": 228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.487007168458781e-05,
+      "loss": 2.4829,
+      "step": 229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.484767025089605e-05,
+      "loss": 2.3301,
+      "step": 230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.482526881720431e-05,
+      "loss": 1.6475,
+      "step": 231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.480286738351255e-05,
+      "loss": 1.9849,
+      "step": 232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.47804659498208e-05,
+      "loss": 1.7402,
+      "step": 233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.475806451612904e-05,
+      "loss": 2.0972,
+      "step": 234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.473566308243727e-05,
+      "loss": 1.9307,
+      "step": 235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.471326164874552e-05,
+      "loss": 1.6899,
+      "step": 236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.469086021505377e-05,
+      "loss": 2.1123,
+      "step": 237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.466845878136202e-05,
+      "loss": 1.7539,
+      "step": 238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.464605734767026e-05,
+      "loss": 2.1416,
+      "step": 239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.46236559139785e-05,
+      "loss": 1.7764,
+      "step": 240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.460125448028674e-05,
+      "loss": 1.1521,
+      "step": 241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.457885304659498e-05,
+      "loss": 0.8562,
+      "step": 242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.455645161290322e-05,
+      "loss": 1.7446,
+      "step": 243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.453405017921148e-05,
+      "loss": 1.7251,
+      "step": 244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.451164874551972e-05,
+      "loss": 1.729,
+      "step": 245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.448924731182797e-05,
+      "loss": 1.6221,
+      "step": 246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.44668458781362e-05,
+      "loss": 2.2266,
+      "step": 247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.444444444444444e-05,
+      "loss": 1.8247,
+      "step": 248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.442204301075269e-05,
+      "loss": 1.4399,
+      "step": 249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.439964157706094e-05,
+      "loss": 2.0449,
+      "step": 250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.437724014336919e-05,
+      "loss": 1.8638,
+      "step": 251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.435483870967743e-05,
+      "loss": 1.8652,
+      "step": 252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.433243727598566e-05,
+      "loss": 1.7354,
+      "step": 253
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.43100358422939e-05,
+      "loss": 1.2703,
+      "step": 254
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.428763440860215e-05,
+      "loss": 1.7529,
+      "step": 255
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.42652329749104e-05,
+      "loss": 2.063,
+      "step": 256
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.424283154121865e-05,
+      "loss": 1.5049,
+      "step": 257
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.42204301075269e-05,
+      "loss": 2.3071,
+      "step": 258
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.419802867383512e-05,
+      "loss": 2.6289,
+      "step": 259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.417562724014337e-05,
+      "loss": 1.5601,
+      "step": 260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.415322580645161e-05,
+      "loss": 1.9238,
+      "step": 261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.413082437275986e-05,
+      "loss": 2.1479,
+      "step": 262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.410842293906811e-05,
+      "loss": 1.7578,
+      "step": 263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.408602150537636e-05,
+      "loss": 2.5537,
+      "step": 264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.406362007168459e-05,
+      "loss": 0.6189,
+      "step": 265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.404121863799283e-05,
+      "loss": 1.5542,
+      "step": 266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.401881720430108e-05,
+      "loss": 2.1865,
+      "step": 267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.399641577060932e-05,
+      "loss": 2.2139,
+      "step": 268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.397401433691758e-05,
+      "loss": 1.7031,
+      "step": 269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.395161290322582e-05,
+      "loss": 1.7866,
+      "step": 270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.392921146953405e-05,
+      "loss": 2.126,
+      "step": 271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.39068100358423e-05,
+      "loss": 1.4075,
+      "step": 272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.388440860215054e-05,
+      "loss": 2.0996,
+      "step": 273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.386200716845878e-05,
+      "loss": 1.7822,
+      "step": 274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.383960573476703e-05,
+      "loss": 2.0142,
+      "step": 275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.381720430107528e-05,
+      "loss": 2.0195,
+      "step": 276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.379480286738351e-05,
+      "loss": 1.8711,
+      "step": 277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.377240143369176e-05,
+      "loss": 1.7266,
+      "step": 278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.375e-05,
+      "loss": 1.6421,
+      "step": 279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.372759856630825e-05,
+      "loss": 2.0674,
+      "step": 280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.370519713261649e-05,
+      "loss": 1.7114,
+      "step": 281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.368279569892473e-05,
+      "loss": 2.5137,
+      "step": 282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.366039426523298e-05,
+      "loss": 1.8247,
+      "step": 283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.363799283154122e-05,
+      "loss": 1.7983,
+      "step": 284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.361559139784947e-05,
+      "loss": 1.9551,
+      "step": 285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.359318996415771e-05,
+      "loss": 1.8818,
+      "step": 286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.357078853046595e-05,
+      "loss": 2.1816,
+      "step": 287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.35483870967742e-05,
+      "loss": 2.1577,
+      "step": 288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.352598566308244e-05,
+      "loss": 2.8096,
+      "step": 289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.350358422939068e-05,
+      "loss": 1.7227,
+      "step": 290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.348118279569893e-05,
+      "loss": 2.0254,
+      "step": 291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.345878136200717e-05,
+      "loss": 2.0508,
+      "step": 292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.343637992831542e-05,
+      "loss": 1.9819,
+      "step": 293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.341397849462366e-05,
+      "loss": 1.501,
+      "step": 294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.33915770609319e-05,
+      "loss": 1.3389,
+      "step": 295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.336917562724015e-05,
+      "loss": 2.2305,
+      "step": 296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.334677419354839e-05,
+      "loss": 1.2847,
+      "step": 297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.332437275985664e-05,
+      "loss": 2.2793,
+      "step": 298
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330197132616488e-05,
+      "loss": 2.4346,
+      "step": 299
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327956989247312e-05,
+      "loss": 1.3911,
+      "step": 300
+    },
+    {
+      "epoch": 0.54,
+      "eval_accuracy": 0.2928571403026581,
+      "eval_loss": 1.7479692697525024,
+      "eval_runtime": 71.6578,
+      "eval_samples_per_second": 3.907,
+      "eval_steps_per_second": 3.907,
+      "step": 300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325716845878137e-05,
+      "loss": 1.791,
+      "step": 301
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323476702508961e-05,
+      "loss": 1.9092,
+      "step": 302
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321236559139785e-05,
+      "loss": 2.2676,
+      "step": 303
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31899641577061e-05,
+      "loss": 2.0947,
+      "step": 304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.316756272401434e-05,
+      "loss": 1.4858,
+      "step": 305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.314516129032259e-05,
+      "loss": 1.9712,
+      "step": 306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.312275985663083e-05,
+      "loss": 1.374,
+      "step": 307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.310035842293907e-05,
+      "loss": 2.2729,
+      "step": 308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.307795698924732e-05,
+      "loss": 2.6484,
+      "step": 309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.305555555555556e-05,
+      "loss": 1.2771,
+      "step": 310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.30331541218638e-05,
+      "loss": 2.3662,
+      "step": 311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.301075268817204e-05,
+      "loss": 1.897,
+      "step": 312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.29883512544803e-05,
+      "loss": 1.4868,
+      "step": 313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.296594982078854e-05,
+      "loss": 1.8608,
+      "step": 314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.294354838709678e-05,
+      "loss": 1.1875,
+      "step": 315
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.292114695340502e-05,
+      "loss": 3.0498,
+      "step": 316
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.289874551971327e-05,
+      "loss": 0.8433,
+      "step": 317
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.28763440860215e-05,
+      "loss": 1.8291,
+      "step": 318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.285394265232976e-05,
+      "loss": 2.1348,
+      "step": 319
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.2831541218638e-05,
+      "loss": 1.0796,
+      "step": 320
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.280913978494624e-05,
+      "loss": 2.1655,
+      "step": 321
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.278673835125449e-05,
+      "loss": 2.5464,
+      "step": 322
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.276433691756273e-05,
+      "loss": 2.0605,
+      "step": 323
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.274193548387096e-05,
+      "loss": 2.0498,
+      "step": 324
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.271953405017922e-05,
+      "loss": 1.7983,
+      "step": 325
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.269713261648746e-05,
+      "loss": 2.6162,
+      "step": 326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.267473118279571e-05,
+      "loss": 2.1865,
+      "step": 327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.265232974910395e-05,
+      "loss": 2.0786,
+      "step": 328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.26299283154122e-05,
+      "loss": 2.1006,
+      "step": 329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.260752688172043e-05,
+      "loss": 1.4985,
+      "step": 330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.258512544802867e-05,
+      "loss": 1.5972,
+      "step": 331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.256272401433693e-05,
+      "loss": 1.7339,
+      "step": 332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.254032258064517e-05,
+      "loss": 2.0078,
+      "step": 333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.251792114695341e-05,
+      "loss": 1.9092,
+      "step": 334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.249551971326166e-05,
+      "loss": 1.981,
+      "step": 335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.247311827956989e-05,
+      "loss": 1.7412,
+      "step": 336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.245071684587813e-05,
+      "loss": 2.3896,
+      "step": 337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.242831541218639e-05,
+      "loss": 2.2383,
+      "step": 338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.240591397849463e-05,
+      "loss": 1.7515,
+      "step": 339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.238351254480288e-05,
+      "loss": 1.8711,
+      "step": 340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.236111111111112e-05,
+      "loss": 1.4512,
+      "step": 341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.233870967741935e-05,
+      "loss": 1.8291,
+      "step": 342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.23163082437276e-05,
+      "loss": 1.8071,
+      "step": 343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.229390681003584e-05,
+      "loss": 2.0073,
+      "step": 344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.22715053763441e-05,
+      "loss": 1.9229,
+      "step": 345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.224910394265234e-05,
+      "loss": 1.3093,
+      "step": 346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.222670250896058e-05,
+      "loss": 1.3926,
+      "step": 347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.220430107526881e-05,
+      "loss": 2.0093,
+      "step": 348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.218189964157706e-05,
+      "loss": 1.9609,
+      "step": 349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.21594982078853e-05,
+      "loss": 1.8403,
+      "step": 350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.213709677419356e-05,
+      "loss": 1.9556,
+      "step": 351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.21146953405018e-05,
+      "loss": 1.074,
+      "step": 352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.209229390681005e-05,
+      "loss": 1.7666,
+      "step": 353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.206989247311828e-05,
+      "loss": 1.3896,
+      "step": 354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.204749103942652e-05,
+      "loss": 1.2461,
+      "step": 355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.202508960573477e-05,
+      "loss": 1.127,
+      "step": 356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.200268817204302e-05,
+      "loss": 0.2274,
+      "step": 357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.198028673835127e-05,
+      "loss": 0.8892,
+      "step": 358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.195788530465951e-05,
+      "loss": 2.5693,
+      "step": 359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.193548387096774e-05,
+      "loss": 1.0413,
+      "step": 360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.191308243727599e-05,
+      "loss": 0.842,
+      "step": 361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.189068100358423e-05,
+      "loss": 1.8721,
+      "step": 362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.186827956989247e-05,
+      "loss": 1.374,
+      "step": 363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.184587813620073e-05,
+      "loss": 1.1377,
+      "step": 364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.182347670250897e-05,
+      "loss": 1.6479,
+      "step": 365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.18010752688172e-05,
+      "loss": 2.2778,
+      "step": 366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.177867383512545e-05,
+      "loss": 2.4688,
+      "step": 367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.175627240143369e-05,
+      "loss": 1.9976,
+      "step": 368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.173387096774194e-05,
+      "loss": 2.4951,
+      "step": 369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.17114695340502e-05,
+      "loss": 1.2477,
+      "step": 370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.168906810035844e-05,
+      "loss": 0.8584,
+      "step": 371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.166666666666667e-05,
+      "loss": 0.9356,
+      "step": 372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.164426523297491e-05,
+      "loss": 0.2082,
+      "step": 373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.162186379928316e-05,
+      "loss": 2.1602,
+      "step": 374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.15994623655914e-05,
+      "loss": 1.6208,
+      "step": 375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.157706093189964e-05,
+      "loss": 2.0015,
+      "step": 376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.155465949820789e-05,
+      "loss": 1.1007,
+      "step": 377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.153225806451613e-05,
+      "loss": 0.9801,
+      "step": 378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.150985663082437e-05,
+      "loss": 2.3232,
+      "step": 379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.148745519713262e-05,
+      "loss": 1.021,
+      "step": 380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.146505376344086e-05,
+      "loss": 3.1465,
+      "step": 381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.14426523297491e-05,
+      "loss": 0.5134,
+      "step": 382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.142025089605735e-05,
+      "loss": 1.5024,
+      "step": 383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.13978494623656e-05,
+      "loss": 1.6792,
+      "step": 384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.137544802867384e-05,
+      "loss": 1.8535,
+      "step": 385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.135304659498208e-05,
+      "loss": 1.2944,
+      "step": 386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.133064516129033e-05,
+      "loss": 1.7695,
+      "step": 387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.130824372759857e-05,
+      "loss": 2.2422,
+      "step": 388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.128584229390681e-05,
+      "loss": 2.1948,
+      "step": 389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.126344086021506e-05,
+      "loss": 1.9922,
+      "step": 390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.12410394265233e-05,
+      "loss": 1.5488,
+      "step": 391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.121863799283154e-05,
+      "loss": 2.9473,
+      "step": 392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.119623655913979e-05,
+      "loss": 1.8184,
+      "step": 393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.117383512544803e-05,
+      "loss": 1.4939,
+      "step": 394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.115143369175628e-05,
+      "loss": 2.1553,
+      "step": 395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.112903225806452e-05,
+      "loss": 2.3203,
+      "step": 396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.110663082437276e-05,
+      "loss": 2.6646,
+      "step": 397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.108422939068101e-05,
+      "loss": 1.8284,
+      "step": 398
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.106182795698925e-05,
+      "loss": 1.3679,
+      "step": 399
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.10394265232975e-05,
+      "loss": 1.9399,
+      "step": 400
+    },
+    {
+      "epoch": 0.72,
+      "eval_accuracy": 0.2607142925262451,
+      "eval_loss": 1.8801453113555908,
+      "eval_runtime": 71.6091,
+      "eval_samples_per_second": 3.91,
+      "eval_steps_per_second": 3.91,
+      "step": 400
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.101702508960574e-05,
+      "loss": 1.916,
+      "step": 401
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.099462365591398e-05,
+      "loss": 2.1592,
+      "step": 402
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.097222222222223e-05,
+      "loss": 1.8662,
+      "step": 403
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.094982078853047e-05,
+      "loss": 2.0229,
+      "step": 404
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.092741935483872e-05,
+      "loss": 1.5234,
+      "step": 405
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.090501792114696e-05,
+      "loss": 1.48,
+      "step": 406
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.08826164874552e-05,
+      "loss": 2.2002,
+      "step": 407
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.086021505376345e-05,
+      "loss": 2.0815,
+      "step": 408
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.083781362007169e-05,
+      "loss": 1.1716,
+      "step": 409
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.081541218637993e-05,
+      "loss": 1.623,
+      "step": 410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.079301075268818e-05,
+      "loss": 2.1104,
+      "step": 411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.077060931899642e-05,
+      "loss": 2.0332,
+      "step": 412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.074820788530465e-05,
+      "loss": 2.2471,
+      "step": 413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.072580645161291e-05,
+      "loss": 1.3816,
+      "step": 414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.070340501792115e-05,
+      "loss": 1.8525,
+      "step": 415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.06810035842294e-05,
+      "loss": 1.7935,
+      "step": 416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.065860215053764e-05,
+      "loss": 1.668,
+      "step": 417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.063620071684589e-05,
+      "loss": 1.8501,
+      "step": 418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.061379928315412e-05,
+      "loss": 1.7402,
+      "step": 419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.059139784946237e-05,
+      "loss": 1.8403,
+      "step": 420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.056899641577062e-05,
+      "loss": 1.8994,
+      "step": 421
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.054659498207886e-05,
+      "loss": 1.7544,
+      "step": 422
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.05241935483871e-05,
+      "loss": 1.5854,
+      "step": 423
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.050179211469535e-05,
+      "loss": 1.6934,
+      "step": 424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.047939068100358e-05,
+      "loss": 1.2817,
+      "step": 425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.045698924731184e-05,
+      "loss": 1.8232,
+      "step": 426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.043458781362008e-05,
+      "loss": 1.71,
+      "step": 427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.041218637992832e-05,
+      "loss": 1.6772,
+      "step": 428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.038978494623657e-05,
+      "loss": 1.3875,
+      "step": 429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.036738351254481e-05,
+      "loss": 1.8511,
+      "step": 430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.034498207885304e-05,
+      "loss": 2.1738,
+      "step": 431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.032258064516129e-05,
+      "loss": 1.749,
+      "step": 432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.030017921146954e-05,
+      "loss": 1.3457,
+      "step": 433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.027777777777779e-05,
+      "loss": 1.4021,
+      "step": 434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.025537634408603e-05,
+      "loss": 1.1914,
+      "step": 435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.023297491039427e-05,
+      "loss": 0.7045,
+      "step": 436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.02105734767025e-05,
+      "loss": 2.2197,
+      "step": 437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.018817204301075e-05,
+      "loss": 1.4697,
+      "step": 438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.0165770609319e-05,
+      "loss": 1.6069,
+      "step": 439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.014336917562725e-05,
+      "loss": 1.219,
+      "step": 440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.01209677419355e-05,
+      "loss": 1.7324,
+      "step": 441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.009856630824374e-05,
+      "loss": 1.0017,
+      "step": 442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.007616487455197e-05,
+      "loss": 3.9473,
+      "step": 443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.005376344086021e-05,
+      "loss": 4.2852,
+      "step": 444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.003136200716846e-05,
+      "loss": 2.0669,
+      "step": 445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.000896057347671e-05,
+      "loss": 2.2927,
+      "step": 446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.998655913978496e-05,
+      "loss": 1.287,
+      "step": 447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.99641577060932e-05,
+      "loss": 1.6377,
+      "step": 448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.994175627240143e-05,
+      "loss": 3.1631,
+      "step": 449
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.991935483870968e-05,
+      "loss": 1.7661,
+      "step": 450
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.989695340501792e-05,
+      "loss": 1.1614,
+      "step": 451
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.987455197132618e-05,
+      "loss": 2.1816,
+      "step": 452
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.985215053763442e-05,
+      "loss": 2.6953,
+      "step": 453
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.982974910394266e-05,
+      "loss": 2.5391,
+      "step": 454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.98073476702509e-05,
+      "loss": 2.2876,
+      "step": 455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.978494623655914e-05,
+      "loss": 2.2529,
+      "step": 456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.976254480286738e-05,
+      "loss": 1.8457,
+      "step": 457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.974014336917564e-05,
+      "loss": 2.5249,
+      "step": 458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.971774193548388e-05,
+      "loss": 1.9814,
+      "step": 459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.969534050179213e-05,
+      "loss": 1.8091,
+      "step": 460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.967293906810036e-05,
+      "loss": 1.896,
+      "step": 461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.96505376344086e-05,
+      "loss": 1.2417,
+      "step": 462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.962813620071685e-05,
+      "loss": 2.0239,
+      "step": 463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.960573476702509e-05,
+      "loss": 1.7192,
+      "step": 464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.958333333333335e-05,
+      "loss": 1.9629,
+      "step": 465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.956093189964158e-05,
+      "loss": 2.002,
+      "step": 466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.953853046594982e-05,
+      "loss": 1.5371,
+      "step": 467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.951612903225806e-05,
+      "loss": 1.6499,
+      "step": 468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.949372759856631e-05,
+      "loss": 1.5063,
+      "step": 469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.947132616487455e-05,
+      "loss": 1.5608,
+      "step": 470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.944892473118281e-05,
+      "loss": 2.0591,
+      "step": 471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.942652329749104e-05,
+      "loss": 0.9446,
+      "step": 472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.940412186379928e-05,
+      "loss": 1.2068,
+      "step": 473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.938172043010753e-05,
+      "loss": 1.6704,
+      "step": 474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.935931899641577e-05,
+      "loss": 1.8096,
+      "step": 475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.933691756272402e-05,
+      "loss": 0.9185,
+      "step": 476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.931451612903226e-05,
+      "loss": 2.582,
+      "step": 477
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.92921146953405e-05,
+      "loss": 1.6436,
+      "step": 478
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.926971326164875e-05,
+      "loss": 2.1694,
+      "step": 479
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.924731182795699e-05,
+      "loss": 1.2703,
+      "step": 480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.922491039426524e-05,
+      "loss": 2.0156,
+      "step": 481
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.920250896057348e-05,
+      "loss": 3.0352,
+      "step": 482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.918010752688172e-05,
+      "loss": 2.2617,
+      "step": 483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.915770609318997e-05,
+      "loss": 2.2812,
+      "step": 484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.913530465949821e-05,
+      "loss": 2.061,
+      "step": 485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.911290322580645e-05,
+      "loss": 1.5635,
+      "step": 486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.90905017921147e-05,
+      "loss": 1.8516,
+      "step": 487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.906810035842294e-05,
+      "loss": 1.8984,
+      "step": 488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.904569892473119e-05,
+      "loss": 1.4932,
+      "step": 489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.902329749103943e-05,
+      "loss": 2.4019,
+      "step": 490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.900089605734767e-05,
+      "loss": 0.4492,
+      "step": 491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.897849462365592e-05,
+      "loss": 1.6909,
+      "step": 492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.895609318996416e-05,
+      "loss": 2.1616,
+      "step": 493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.89336917562724e-05,
+      "loss": 2.251,
+      "step": 494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.891129032258065e-05,
+      "loss": 2.0,
+      "step": 495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 1.478,
+      "step": 496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.886648745519714e-05,
+      "loss": 1.8101,
+      "step": 497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.884408602150538e-05,
+      "loss": 1.5688,
+      "step": 498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.882168458781362e-05,
+      "loss": 1.6016,
+      "step": 499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.879928315412187e-05,
+      "loss": 1.4307,
+      "step": 500
+    },
+    {
+      "epoch": 0.9,
+      "eval_accuracy": 0.3857142925262451,
+      "eval_loss": 1.6276655197143555,
+      "eval_runtime": 73.5912,
+      "eval_samples_per_second": 3.805,
+      "eval_steps_per_second": 3.805,
+      "step": 500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.877688172043011e-05,
+      "loss": 1.4912,
+      "step": 501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.875448028673836e-05,
+      "loss": 1.3667,
+      "step": 502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.87320788530466e-05,
+      "loss": 1.2695,
+      "step": 503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.870967741935484e-05,
+      "loss": 1.7344,
+      "step": 504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.868727598566309e-05,
+      "loss": 1.4678,
+      "step": 505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.866487455197133e-05,
+      "loss": 2.271,
+      "step": 506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.864247311827958e-05,
+      "loss": 0.8962,
+      "step": 507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.862007168458782e-05,
+      "loss": 2.0215,
+      "step": 508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.859767025089606e-05,
+      "loss": 1.9561,
+      "step": 509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.857526881720431e-05,
+      "loss": 2.3936,
+      "step": 510
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.855286738351255e-05,
+      "loss": 1.4082,
+      "step": 511
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.85304659498208e-05,
+      "loss": 2.2334,
+      "step": 512
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.850806451612904e-05,
+      "loss": 1.1714,
+      "step": 513
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.848566308243727e-05,
+      "loss": 1.7935,
+      "step": 514
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.846326164874553e-05,
+      "loss": 1.6655,
+      "step": 515
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.844086021505377e-05,
+      "loss": 2.1167,
+      "step": 516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.841845878136201e-05,
+      "loss": 0.5005,
+      "step": 517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.839605734767026e-05,
+      "loss": 0.947,
+      "step": 518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.83736559139785e-05,
+      "loss": 1.8589,
+      "step": 519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.835125448028673e-05,
+      "loss": 1.3933,
+      "step": 520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.832885304659499e-05,
+      "loss": 2.541,
+      "step": 521
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.830645161290323e-05,
+      "loss": 1.3613,
+      "step": 522
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.828405017921148e-05,
+      "loss": 2.0708,
+      "step": 523
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.826164874551972e-05,
+      "loss": 2.4546,
+      "step": 524
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.823924731182797e-05,
+      "loss": 1.9714,
+      "step": 525
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.82168458781362e-05,
+      "loss": 2.2002,
+      "step": 526
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.819444444444445e-05,
+      "loss": 1.4688,
+      "step": 527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.81720430107527e-05,
+      "loss": 1.8623,
+      "step": 528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.814964157706094e-05,
+      "loss": 1.7202,
+      "step": 529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.812724014336918e-05,
+      "loss": 0.5217,
+      "step": 530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.810483870967743e-05,
+      "loss": 2.4199,
+      "step": 531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.808243727598566e-05,
+      "loss": 1.0787,
+      "step": 532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.80600358422939e-05,
+      "loss": 1.3494,
+      "step": 533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.803763440860216e-05,
+      "loss": 1.4258,
+      "step": 534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.80152329749104e-05,
+      "loss": 1.2393,
+      "step": 535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.799283154121865e-05,
+      "loss": 1.6782,
+      "step": 536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.797043010752689e-05,
+      "loss": 1.2218,
+      "step": 537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.794802867383512e-05,
+      "loss": 0.3288,
+      "step": 538
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.792562724014337e-05,
+      "loss": 1.7256,
+      "step": 539
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.790322580645162e-05,
+      "loss": 1.7393,
+      "step": 540
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.788082437275987e-05,
+      "loss": 2.0449,
+      "step": 541
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.785842293906811e-05,
+      "loss": 0.7985,
+      "step": 542
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.783602150537635e-05,
+      "loss": 1.3423,
+      "step": 543
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.781362007168459e-05,
+      "loss": 1.9751,
+      "step": 544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.779121863799283e-05,
+      "loss": 1.7485,
+      "step": 545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.776881720430107e-05,
+      "loss": 2.3511,
+      "step": 546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.774641577060933e-05,
+      "loss": 1.3589,
+      "step": 547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.772401433691757e-05,
+      "loss": 0.6724,
+      "step": 548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.770161290322582e-05,
+      "loss": 1.138,
+      "step": 549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.767921146953405e-05,
+      "loss": 3.124,
+      "step": 550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.765681003584229e-05,
+      "loss": 1.9717,
+      "step": 551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.763440860215054e-05,
+      "loss": 2.2417,
+      "step": 552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.76120071684588e-05,
+      "loss": 1.563,
+      "step": 553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.758960573476704e-05,
+      "loss": 0.8355,
+      "step": 554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.756720430107528e-05,
+      "loss": 2.1084,
+      "step": 555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.754480286738351e-05,
+      "loss": 2.4531,
+      "step": 556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.752240143369176e-05,
+      "loss": 1.9697,
+      "step": 557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.75e-05,
+      "loss": 2.084,
+      "step": 558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.747759856630826e-05,
+      "loss": 1.677,
+      "step": 559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.74551971326165e-05,
+      "loss": 1.6162,
+      "step": 560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.743279569892473e-05,
+      "loss": 0.7837,
+      "step": 561
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.741039426523297e-05,
+      "loss": 1.9287,
+      "step": 562
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.738799283154122e-05,
+      "loss": 1.564,
+      "step": 563
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.736559139784946e-05,
+      "loss": 1.0551,
+      "step": 564
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.73431899641577e-05,
+      "loss": 2.5762,
+      "step": 565
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.732078853046596e-05,
+      "loss": 1.751,
+      "step": 566
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.72983870967742e-05,
+      "loss": 1.073,
+      "step": 567
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.727598566308244e-05,
+      "loss": 1.1838,
+      "step": 568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.725358422939068e-05,
+      "loss": 2.4639,
+      "step": 569
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.723118279569893e-05,
+      "loss": 1.5996,
+      "step": 570
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.720878136200717e-05,
+      "loss": 1.1963,
+      "step": 571
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.718637992831543e-05,
+      "loss": 2.4766,
+      "step": 572
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.716397849462366e-05,
+      "loss": 1.3667,
+      "step": 573
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.71415770609319e-05,
+      "loss": 1.3931,
+      "step": 574
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.711917562724014e-05,
+      "loss": 2.0127,
+      "step": 575
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.709677419354839e-05,
+      "loss": 1.1589,
+      "step": 576
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.707437275985663e-05,
+      "loss": 1.5146,
+      "step": 577
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.705197132616488e-05,
+      "loss": 1.502,
+      "step": 578
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.702956989247312e-05,
+      "loss": 1.6062,
+      "step": 579
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.700716845878136e-05,
+      "loss": 1.6909,
+      "step": 580
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.698476702508961e-05,
+      "loss": 1.7241,
+      "step": 581
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.696236559139785e-05,
+      "loss": 1.5938,
+      "step": 582
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.69399641577061e-05,
+      "loss": 1.7642,
+      "step": 583
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.691756272401434e-05,
+      "loss": 1.9829,
+      "step": 584
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.689516129032258e-05,
+      "loss": 1.2065,
+      "step": 585
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.687275985663083e-05,
+      "loss": 1.7197,
+      "step": 586
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.685035842293907e-05,
+      "loss": 0.9531,
+      "step": 587
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.682795698924732e-05,
+      "loss": 0.7196,
+      "step": 588
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.680555555555556e-05,
+      "loss": 2.396,
+      "step": 589
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.67831541218638e-05,
+      "loss": 1.6641,
+      "step": 590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.676075268817205e-05,
+      "loss": 1.7705,
+      "step": 591
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.673835125448029e-05,
+      "loss": 1.3604,
+      "step": 592
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.671594982078853e-05,
+      "loss": 0.2497,
+      "step": 593
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.669354838709678e-05,
+      "loss": 1.4526,
+      "step": 594
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.667114695340502e-05,
+      "loss": 1.1375,
+      "step": 595
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.664874551971327e-05,
+      "loss": 1.1075,
+      "step": 596
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.662634408602151e-05,
+      "loss": 2.2949,
+      "step": 597
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.660394265232975e-05,
+      "loss": 2.4438,
+      "step": 598
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.6581541218638e-05,
+      "loss": 1.3958,
+      "step": 599
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.655913978494624e-05,
+      "loss": 2.0425,
+      "step": 600
+    },
+    {
+      "epoch": 1.08,
+      "eval_accuracy": 0.37857142090797424,
+      "eval_loss": 1.6400291919708252,
+      "eval_runtime": 68.0816,
+      "eval_samples_per_second": 4.113,
+      "eval_steps_per_second": 4.113,
+      "step": 600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.653673835125449e-05,
+      "loss": 0.1631,
+      "step": 601
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.651433691756273e-05,
+      "loss": 1.5117,
+      "step": 602
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.649193548387097e-05,
+      "loss": 2.5332,
+      "step": 603
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.646953405017922e-05,
+      "loss": 0.9135,
+      "step": 604
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.646953405017922e-05,
+      "loss": 1.5693,
+      "step": 605
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.644713261648746e-05,
+      "loss": 1.6929,
+      "step": 606
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.64247311827957e-05,
+      "loss": 1.834,
+      "step": 607
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.640232974910395e-05,
+      "loss": 2.3213,
+      "step": 608
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.637992831541219e-05,
+      "loss": 0.7445,
+      "step": 609
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.635752688172044e-05,
+      "loss": 0.7114,
+      "step": 610
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.633512544802868e-05,
+      "loss": 0.7075,
+      "step": 611
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.631272401433692e-05,
+      "loss": 1.8706,
+      "step": 612
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.629032258064517e-05,
+      "loss": 1.3271,
+      "step": 613
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.626792114695341e-05,
+      "loss": 1.3462,
+      "step": 614
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.624551971326166e-05,
+      "loss": 1.108,
+      "step": 615
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.622311827956989e-05,
+      "loss": 1.9233,
+      "step": 616
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.620071684587814e-05,
+      "loss": 2.2646,
+      "step": 617
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.617831541218639e-05,
+      "loss": 0.7799,
+      "step": 618
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.615591397849463e-05,
+      "loss": 0.9399,
+      "step": 619
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.613351254480287e-05,
+      "loss": 1.3552,
+      "step": 620
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.611111111111112e-05,
+      "loss": 1.4927,
+      "step": 621
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.608870967741935e-05,
+      "loss": 2.3369,
+      "step": 622
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.60663082437276e-05,
+      "loss": 0.8125,
+      "step": 623
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.604390681003585e-05,
+      "loss": 1.4315,
+      "step": 624
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.60215053763441e-05,
+      "loss": 1.96,
+      "step": 625
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.599910394265234e-05,
+      "loss": 2.3286,
+      "step": 626
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.597670250896058e-05,
+      "loss": 1.7153,
+      "step": 627
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.595430107526881e-05,
+      "loss": 2.2896,
+      "step": 628
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.593189964157706e-05,
+      "loss": 1.6265,
+      "step": 629
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.590949820788531e-05,
+      "loss": 1.7207,
+      "step": 630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.588709677419356e-05,
+      "loss": 1.8999,
+      "step": 631
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.58646953405018e-05,
+      "loss": 1.6899,
+      "step": 632
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.584229390681004e-05,
+      "loss": 2.0322,
+      "step": 633
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.581989247311828e-05,
+      "loss": 0.8866,
+      "step": 634
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.579749103942652e-05,
+      "loss": 1.874,
+      "step": 635
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.577508960573478e-05,
+      "loss": 0.7322,
+      "step": 636
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.575268817204302e-05,
+      "loss": 1.981,
+      "step": 637
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.573028673835126e-05,
+      "loss": 1.2945,
+      "step": 638
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.570788530465951e-05,
+      "loss": 1.8857,
+      "step": 639
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.568548387096774e-05,
+      "loss": 2.0093,
+      "step": 640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.566308243727598e-05,
+      "loss": 3.9951,
+      "step": 641
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.564068100358424e-05,
+      "loss": 1.7852,
+      "step": 642
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.561827956989248e-05,
+      "loss": 0.8218,
+      "step": 643
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.559587813620073e-05,
+      "loss": 1.5381,
+      "step": 644
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.557347670250897e-05,
+      "loss": 0.5192,
+      "step": 645
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.55510752688172e-05,
+      "loss": 1.1389,
+      "step": 646
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.552867383512545e-05,
+      "loss": 0.7243,
+      "step": 647
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.550627240143369e-05,
+      "loss": 0.6862,
+      "step": 648
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.548387096774195e-05,
+      "loss": 0.6665,
+      "step": 649
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.546146953405019e-05,
+      "loss": 2.0005,
+      "step": 650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.543906810035843e-05,
+      "loss": 2.293,
+      "step": 651
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.541666666666666e-05,
+      "loss": 0.6967,
+      "step": 652
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.539426523297491e-05,
+      "loss": 0.5178,
+      "step": 653
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.537186379928315e-05,
+      "loss": 1.3145,
+      "step": 654
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.534946236559141e-05,
+      "loss": 1.3098,
+      "step": 655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.532706093189965e-05,
+      "loss": 0.5188,
+      "step": 656
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.530465949820788e-05,
+      "loss": 0.4375,
+      "step": 657
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.528225806451613e-05,
+      "loss": 1.8096,
+      "step": 658
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.525985663082437e-05,
+      "loss": 1.645,
+      "step": 659
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.523745519713262e-05,
+      "loss": 1.1252,
+      "step": 660
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.521505376344086e-05,
+      "loss": 2.041,
+      "step": 661
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.519265232974912e-05,
+      "loss": 2.1512,
+      "step": 662
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.517025089605735e-05,
+      "loss": 2.3179,
+      "step": 663
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.514784946236559e-05,
+      "loss": 0.385,
+      "step": 664
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.512544802867384e-05,
+      "loss": 3.0366,
+      "step": 665
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.510304659498208e-05,
+      "loss": 1.9922,
+      "step": 666
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.508064516129032e-05,
+      "loss": 0.7699,
+      "step": 667
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.505824372759858e-05,
+      "loss": 2.2764,
+      "step": 668
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.503584229390681e-05,
+      "loss": 1.5791,
+      "step": 669
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.501344086021505e-05,
+      "loss": 1.8477,
+      "step": 670
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.49910394265233e-05,
+      "loss": 1.7646,
+      "step": 671
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.496863799283154e-05,
+      "loss": 2.1377,
+      "step": 672
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.494623655913979e-05,
+      "loss": 0.1815,
+      "step": 673
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.492383512544804e-05,
+      "loss": 1.24,
+      "step": 674
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.490143369175627e-05,
+      "loss": 0.0883,
+      "step": 675
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.487903225806452e-05,
+      "loss": 0.8481,
+      "step": 676
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.485663082437276e-05,
+      "loss": 1.6841,
+      "step": 677
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.4834229390681e-05,
+      "loss": 1.4165,
+      "step": 678
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.481182795698925e-05,
+      "loss": 1.1509,
+      "step": 679
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.478942652329749e-05,
+      "loss": 2.457,
+      "step": 680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.476702508960574e-05,
+      "loss": 2.1748,
+      "step": 681
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.474462365591398e-05,
+      "loss": 1.6724,
+      "step": 682
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.472222222222222e-05,
+      "loss": 1.1281,
+      "step": 683
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.469982078853047e-05,
+      "loss": 0.434,
+      "step": 684
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.467741935483871e-05,
+      "loss": 1.5586,
+      "step": 685
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.465501792114696e-05,
+      "loss": 1.2427,
+      "step": 686
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.46326164874552e-05,
+      "loss": 1.261,
+      "step": 687
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.461021505376344e-05,
+      "loss": 2.5706,
+      "step": 688
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.458781362007169e-05,
+      "loss": 1.5532,
+      "step": 689
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.456541218637993e-05,
+      "loss": 1.2065,
+      "step": 690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.454301075268818e-05,
+      "loss": 1.3765,
+      "step": 691
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.452060931899642e-05,
+      "loss": 1.4312,
+      "step": 692
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.449820788530466e-05,
+      "loss": 1.9006,
+      "step": 693
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.447580645161291e-05,
+      "loss": 2.355,
+      "step": 694
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.445340501792115e-05,
+      "loss": 0.708,
+      "step": 695
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.44310035842294e-05,
+      "loss": 2.6206,
+      "step": 696
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.440860215053764e-05,
+      "loss": 1.7021,
+      "step": 697
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.438620071684588e-05,
+      "loss": 1.5596,
+      "step": 698
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.436379928315413e-05,
+      "loss": 1.1562,
+      "step": 699
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.434139784946237e-05,
+      "loss": 2.2441,
+      "step": 700
+    },
+    {
+      "epoch": 1.25,
+      "eval_accuracy": 0.41428571939468384,
+      "eval_loss": 1.6656556129455566,
+      "eval_runtime": 72.3292,
+      "eval_samples_per_second": 3.871,
+      "eval_steps_per_second": 3.871,
+      "step": 700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.431899641577061e-05,
+      "loss": 1.6909,
+      "step": 701
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.429659498207886e-05,
+      "loss": 1.9697,
+      "step": 702
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.42741935483871e-05,
+      "loss": 1.8889,
+      "step": 703
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.425179211469535e-05,
+      "loss": 1.0661,
+      "step": 704
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.422939068100359e-05,
+      "loss": 2.0146,
+      "step": 705
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.420698924731183e-05,
+      "loss": 0.7656,
+      "step": 706
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.418458781362008e-05,
+      "loss": 1.4375,
+      "step": 707
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.416218637992832e-05,
+      "loss": 0.9893,
+      "step": 708
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.413978494623657e-05,
+      "loss": 0.8765,
+      "step": 709
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.411738351254481e-05,
+      "loss": 1.2578,
+      "step": 710
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.409498207885305e-05,
+      "loss": 1.1987,
+      "step": 711
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.40725806451613e-05,
+      "loss": 1.8105,
+      "step": 712
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.405017921146954e-05,
+      "loss": 1.7139,
+      "step": 713
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.402777777777778e-05,
+      "loss": 2.0947,
+      "step": 714
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.400537634408603e-05,
+      "loss": 1.0601,
+      "step": 715
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.398297491039427e-05,
+      "loss": 2.9722,
+      "step": 716
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.39605734767025e-05,
+      "loss": 2.1509,
+      "step": 717
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.393817204301076e-05,
+      "loss": 1.5361,
+      "step": 718
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.3915770609319e-05,
+      "loss": 0.9551,
+      "step": 719
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.389336917562725e-05,
+      "loss": 2.3413,
+      "step": 720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.387096774193549e-05,
+      "loss": 2.0957,
+      "step": 721
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.384856630824374e-05,
+      "loss": 1.375,
+      "step": 722
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.382616487455197e-05,
+      "loss": 1.1423,
+      "step": 723
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.380376344086022e-05,
+      "loss": 1.3345,
+      "step": 724
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.378136200716847e-05,
+      "loss": 1.4033,
+      "step": 725
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.375896057347671e-05,
+      "loss": 1.3496,
+      "step": 726
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.373655913978495e-05,
+      "loss": 1.031,
+      "step": 727
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.37141577060932e-05,
+      "loss": 1.1985,
+      "step": 728
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.369175627240143e-05,
+      "loss": 2.7236,
+      "step": 729
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.366935483870967e-05,
+      "loss": 1.5041,
+      "step": 730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.364695340501793e-05,
+      "loss": 1.9395,
+      "step": 731
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.362455197132617e-05,
+      "loss": 0.6042,
+      "step": 732
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.360215053763442e-05,
+      "loss": 1.708,
+      "step": 733
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.357974910394266e-05,
+      "loss": 2.2485,
+      "step": 734
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.355734767025089e-05,
+      "loss": 1.7017,
+      "step": 735
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.353494623655914e-05,
+      "loss": 1.6519,
+      "step": 736
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.351254480286739e-05,
+      "loss": 0.406,
+      "step": 737
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.349014336917564e-05,
+      "loss": 1.8145,
+      "step": 738
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.346774193548388e-05,
+      "loss": 1.2134,
+      "step": 739
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.344534050179212e-05,
+      "loss": 1.7598,
+      "step": 740
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.342293906810036e-05,
+      "loss": 1.3921,
+      "step": 741
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.34005376344086e-05,
+      "loss": 1.7202,
+      "step": 742
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.337813620071686e-05,
+      "loss": 0.5819,
+      "step": 743
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.33557347670251e-05,
+      "loss": 1.4595,
+      "step": 744
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.333333333333334e-05,
+      "loss": 1.418,
+      "step": 745
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.331093189964157e-05,
+      "loss": 1.134,
+      "step": 746
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.328853046594982e-05,
+      "loss": 1.9927,
+      "step": 747
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.326612903225806e-05,
+      "loss": 0.4854,
+      "step": 748
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.32437275985663e-05,
+      "loss": 0.993,
+      "step": 749
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.322132616487456e-05,
+      "loss": 1.7871,
+      "step": 750
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.319892473118281e-05,
+      "loss": 1.0767,
+      "step": 751
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.317652329749104e-05,
+      "loss": 1.6538,
+      "step": 752
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.315412186379928e-05,
+      "loss": 1.7847,
+      "step": 753
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.313172043010753e-05,
+      "loss": 0.4056,
+      "step": 754
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.310931899641577e-05,
+      "loss": 0.7401,
+      "step": 755
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.308691756272403e-05,
+      "loss": 1.8746,
+      "step": 756
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.306451612903227e-05,
+      "loss": 0.5989,
+      "step": 757
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.30421146953405e-05,
+      "loss": 0.8958,
+      "step": 758
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.301971326164874e-05,
+      "loss": 1.0583,
+      "step": 759
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.299731182795699e-05,
+      "loss": 1.7059,
+      "step": 760
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.297491039426523e-05,
+      "loss": 1.5839,
+      "step": 761
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.295250896057348e-05,
+      "loss": 0.9373,
+      "step": 762
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.293010752688173e-05,
+      "loss": 2.1914,
+      "step": 763
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.290770609318996e-05,
+      "loss": 1.3962,
+      "step": 764
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.288530465949821e-05,
+      "loss": 0.8971,
+      "step": 765
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.286290322580645e-05,
+      "loss": 1.6825,
+      "step": 766
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.28405017921147e-05,
+      "loss": 2.5186,
+      "step": 767
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.281810035842294e-05,
+      "loss": 1.3931,
+      "step": 768
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.27956989247312e-05,
+      "loss": 1.3789,
+      "step": 769
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.277329749103943e-05,
+      "loss": 0.5911,
+      "step": 770
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.275089605734767e-05,
+      "loss": 1.4648,
+      "step": 771
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.272849462365591e-05,
+      "loss": 0.9783,
+      "step": 772
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.270609318996416e-05,
+      "loss": 1.9331,
+      "step": 773
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.26836917562724e-05,
+      "loss": 1.2107,
+      "step": 774
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.266129032258066e-05,
+      "loss": 2.0386,
+      "step": 775
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.263888888888889e-05,
+      "loss": 1.1836,
+      "step": 776
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.261648745519713e-05,
+      "loss": 0.6439,
+      "step": 777
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.259408602150538e-05,
+      "loss": 1.1245,
+      "step": 778
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.257168458781362e-05,
+      "loss": 1.3555,
+      "step": 779
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.254928315412187e-05,
+      "loss": 0.8435,
+      "step": 780
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.252688172043011e-05,
+      "loss": 0.8406,
+      "step": 781
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.250448028673835e-05,
+      "loss": 1.874,
+      "step": 782
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.24820788530466e-05,
+      "loss": 1.675,
+      "step": 783
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.245967741935484e-05,
+      "loss": 0.9182,
+      "step": 784
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.243727598566309e-05,
+      "loss": 1.894,
+      "step": 785
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.241487455197133e-05,
+      "loss": 0.8936,
+      "step": 786
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.239247311827957e-05,
+      "loss": 1.0576,
+      "step": 787
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.237007168458782e-05,
+      "loss": 1.8984,
+      "step": 788
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.234767025089606e-05,
+      "loss": 1.6123,
+      "step": 789
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.23252688172043e-05,
+      "loss": 2.1042,
+      "step": 790
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.230286738351255e-05,
+      "loss": 0.2871,
+      "step": 791
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.228046594982079e-05,
+      "loss": 0.2499,
+      "step": 792
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.225806451612904e-05,
+      "loss": 0.352,
+      "step": 793
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.223566308243728e-05,
+      "loss": 0.7305,
+      "step": 794
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.221326164874552e-05,
+      "loss": 1.6785,
+      "step": 795
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.219086021505377e-05,
+      "loss": 0.7473,
+      "step": 796
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.216845878136201e-05,
+      "loss": 1.9878,
+      "step": 797
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.214605734767026e-05,
+      "loss": 1.3333,
+      "step": 798
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.21236559139785e-05,
+      "loss": 1.3811,
+      "step": 799
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.210125448028674e-05,
+      "loss": 1.3828,
+      "step": 800
+    },
+    {
+      "epoch": 1.43,
+      "eval_accuracy": 0.44285714626312256,
+      "eval_loss": 1.3938608169555664,
+      "eval_runtime": 72.0646,
+      "eval_samples_per_second": 3.885,
+      "eval_steps_per_second": 3.885,
+      "step": 800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.207885304659499e-05,
+      "loss": 1.0645,
+      "step": 801
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.205645161290323e-05,
+      "loss": 1.2969,
+      "step": 802
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.203405017921147e-05,
+      "loss": 1.0817,
+      "step": 803
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.201164874551972e-05,
+      "loss": 2.6523,
+      "step": 804
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.198924731182796e-05,
+      "loss": 1.9668,
+      "step": 805
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.19668458781362e-05,
+      "loss": 0.8972,
+      "step": 806
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.194444444444445e-05,
+      "loss": 1.1556,
+      "step": 807
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.19220430107527e-05,
+      "loss": 1.2393,
+      "step": 808
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.189964157706094e-05,
+      "loss": 0.9473,
+      "step": 809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.187724014336918e-05,
+      "loss": 1.5625,
+      "step": 810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.185483870967743e-05,
+      "loss": 1.3135,
+      "step": 811
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.183243727598567e-05,
+      "loss": 0.823,
+      "step": 812
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.181003584229391e-05,
+      "loss": 1.8521,
+      "step": 813
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.178763440860216e-05,
+      "loss": 0.8483,
+      "step": 814
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.17652329749104e-05,
+      "loss": 0.7674,
+      "step": 815
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.174283154121864e-05,
+      "loss": 0.4813,
+      "step": 816
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.172043010752689e-05,
+      "loss": 0.9792,
+      "step": 817
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.169802867383512e-05,
+      "loss": 1.0889,
+      "step": 818
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.167562724014338e-05,
+      "loss": 0.3524,
+      "step": 819
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.165322580645162e-05,
+      "loss": 0.0221,
+      "step": 820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.163082437275986e-05,
+      "loss": 0.5209,
+      "step": 821
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.160842293906811e-05,
+      "loss": 1.2841,
+      "step": 822
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.158602150537635e-05,
+      "loss": 2.2139,
+      "step": 823
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.156362007168458e-05,
+      "loss": 2.9517,
+      "step": 824
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.154121863799284e-05,
+      "loss": 1.8062,
+      "step": 825
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.151881720430108e-05,
+      "loss": 0.8574,
+      "step": 826
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.149641577060933e-05,
+      "loss": 1.394,
+      "step": 827
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.147401433691757e-05,
+      "loss": 1.8721,
+      "step": 828
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.145161290322582e-05,
+      "loss": 0.6349,
+      "step": 829
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.142921146953405e-05,
+      "loss": 1.2915,
+      "step": 830
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.140681003584229e-05,
+      "loss": 2.6484,
+      "step": 831
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.138440860215055e-05,
+      "loss": 0.8102,
+      "step": 832
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.136200716845879e-05,
+      "loss": 2.5332,
+      "step": 833
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.133960573476703e-05,
+      "loss": 2.7959,
+      "step": 834
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.131720430107528e-05,
+      "loss": 1.2441,
+      "step": 835
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.129480286738351e-05,
+      "loss": 0.7471,
+      "step": 836
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.127240143369175e-05,
+      "loss": 1.2144,
+      "step": 837
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 1.5024,
+      "step": 838
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.122759856630825e-05,
+      "loss": 1.3842,
+      "step": 839
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.12051971326165e-05,
+      "loss": 0.9773,
+      "step": 840
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.118279569892473e-05,
+      "loss": 0.4269,
+      "step": 841
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.116039426523297e-05,
+      "loss": 0.0926,
+      "step": 842
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.113799283154122e-05,
+      "loss": 1.5227,
+      "step": 843
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.111559139784947e-05,
+      "loss": 1.8347,
+      "step": 844
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.109318996415772e-05,
+      "loss": 0.1099,
+      "step": 845
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.107078853046596e-05,
+      "loss": 0.091,
+      "step": 846
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.104838709677419e-05,
+      "loss": 1.8896,
+      "step": 847
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.102598566308243e-05,
+      "loss": 1.9394,
+      "step": 848
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.100358422939068e-05,
+      "loss": 2.8784,
+      "step": 849
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.098118279569892e-05,
+      "loss": 0.0854,
+      "step": 850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.095878136200718e-05,
+      "loss": 3.0371,
+      "step": 851
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.093637992831542e-05,
+      "loss": 2.4032,
+      "step": 852
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.091397849462365e-05,
+      "loss": 0.9441,
+      "step": 853
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.08915770609319e-05,
+      "loss": 1.5615,
+      "step": 854
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.086917562724014e-05,
+      "loss": 2.6577,
+      "step": 855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.084677419354839e-05,
+      "loss": 1.5986,
+      "step": 856
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.082437275985664e-05,
+      "loss": 0.9976,
+      "step": 857
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.080197132616489e-05,
+      "loss": 0.0674,
+      "step": 858
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.077956989247312e-05,
+      "loss": 0.0194,
+      "step": 859
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.075716845878136e-05,
+      "loss": 0.2408,
+      "step": 860
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.07347670250896e-05,
+      "loss": 2.5933,
+      "step": 861
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.071236559139785e-05,
+      "loss": 2.1333,
+      "step": 862
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.068996415770609e-05,
+      "loss": 3.7295,
+      "step": 863
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.066756272401435e-05,
+      "loss": 1.6323,
+      "step": 864
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.064516129032258e-05,
+      "loss": 2.3125,
+      "step": 865
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.062275985663082e-05,
+      "loss": 0.6023,
+      "step": 866
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.060035842293907e-05,
+      "loss": 2.46,
+      "step": 867
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.057795698924731e-05,
+      "loss": 0.8926,
+      "step": 868
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.055555555555556e-05,
+      "loss": 1.0042,
+      "step": 869
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.053315412186381e-05,
+      "loss": 2.7021,
+      "step": 870
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.051075268817204e-05,
+      "loss": 0.9603,
+      "step": 871
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.048835125448029e-05,
+      "loss": 1.998,
+      "step": 872
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.046594982078853e-05,
+      "loss": 0.4167,
+      "step": 873
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.044354838709678e-05,
+      "loss": 0.7897,
+      "step": 874
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.042114695340502e-05,
+      "loss": 1.2462,
+      "step": 875
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.039874551971328e-05,
+      "loss": 1.7317,
+      "step": 876
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.037634408602151e-05,
+      "loss": 0.7854,
+      "step": 877
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.035394265232975e-05,
+      "loss": 1.7472,
+      "step": 878
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.035394265232975e-05,
+      "loss": 1.0493,
+      "step": 879
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.0331541218638e-05,
+      "loss": 1.3927,
+      "step": 880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.030913978494624e-05,
+      "loss": 0.2318,
+      "step": 881
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.028673835125448e-05,
+      "loss": 1.6774,
+      "step": 882
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.026433691756273e-05,
+      "loss": 0.7754,
+      "step": 883
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.024193548387097e-05,
+      "loss": 1.052,
+      "step": 884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.021953405017921e-05,
+      "loss": 2.8124,
+      "step": 885
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.019713261648746e-05,
+      "loss": 0.886,
+      "step": 886
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.01747311827957e-05,
+      "loss": 1.9741,
+      "step": 887
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.015232974910395e-05,
+      "loss": 1.4648,
+      "step": 888
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.012992831541219e-05,
+      "loss": 1.2797,
+      "step": 889
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.010752688172043e-05,
+      "loss": 0.7076,
+      "step": 890
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.008512544802868e-05,
+      "loss": 1.4478,
+      "step": 891
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.006272401433692e-05,
+      "loss": 0.3945,
+      "step": 892
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.004032258064516e-05,
+      "loss": 1.5894,
+      "step": 893
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.001792114695341e-05,
+      "loss": 1.1672,
+      "step": 894
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.999551971326165e-05,
+      "loss": 2.106,
+      "step": 895
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.99731182795699e-05,
+      "loss": 0.8855,
+      "step": 896
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.995071684587814e-05,
+      "loss": 1.2304,
+      "step": 897
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.992831541218638e-05,
+      "loss": 0.8303,
+      "step": 898
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.990591397849463e-05,
+      "loss": 0.1693,
+      "step": 899
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.988351254480287e-05,
+      "loss": 0.0412,
+      "step": 900
+    },
+    {
+      "epoch": 1.61,
+      "eval_accuracy": 0.3928571343421936,
+      "eval_loss": 1.5016026496887207,
+      "eval_runtime": 72.0453,
+      "eval_samples_per_second": 3.886,
+      "eval_steps_per_second": 3.886,
+      "step": 900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.986111111111112e-05,
+      "loss": 0.4989,
+      "step": 901
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.983870967741936e-05,
+      "loss": 1.9785,
+      "step": 902
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.98163082437276e-05,
+      "loss": 1.4447,
+      "step": 903
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.979390681003585e-05,
+      "loss": 3.3613,
+      "step": 904
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.977150537634409e-05,
+      "loss": 1.6592,
+      "step": 905
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.974910394265234e-05,
+      "loss": 1.7651,
+      "step": 906
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.972670250896058e-05,
+      "loss": 1.7915,
+      "step": 907
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.970430107526882e-05,
+      "loss": 1.0236,
+      "step": 908
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.968189964157707e-05,
+      "loss": 1.0505,
+      "step": 909
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.965949820788531e-05,
+      "loss": 2.6653,
+      "step": 910
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.963709677419355e-05,
+      "loss": 2.582,
+      "step": 911
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.96146953405018e-05,
+      "loss": 2.2676,
+      "step": 912
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.959229390681004e-05,
+      "loss": 0.7974,
+      "step": 913
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.956989247311829e-05,
+      "loss": 0.849,
+      "step": 914
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.954749103942653e-05,
+      "loss": 1.9766,
+      "step": 915
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.952508960573477e-05,
+      "loss": 0.2508,
+      "step": 916
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.950268817204302e-05,
+      "loss": 1.8765,
+      "step": 917
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.948028673835126e-05,
+      "loss": 2.5605,
+      "step": 918
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.94578853046595e-05,
+      "loss": 1.0904,
+      "step": 919
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.943548387096774e-05,
+      "loss": 1.7644,
+      "step": 920
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.941308243727599e-05,
+      "loss": 1.3379,
+      "step": 921
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.939068100358424e-05,
+      "loss": 1.0667,
+      "step": 922
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.936827956989248e-05,
+      "loss": 0.0131,
+      "step": 923
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.934587813620072e-05,
+      "loss": 1.0342,
+      "step": 924
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.932347670250897e-05,
+      "loss": 1.3677,
+      "step": 925
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.93010752688172e-05,
+      "loss": 1.9189,
+      "step": 926
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.927867383512546e-05,
+      "loss": 0.9065,
+      "step": 927
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.92562724014337e-05,
+      "loss": 0.8306,
+      "step": 928
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.923387096774194e-05,
+      "loss": 1.7051,
+      "step": 929
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.921146953405019e-05,
+      "loss": 0.6057,
+      "step": 930
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.918906810035843e-05,
+      "loss": 2.4355,
+      "step": 931
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.916666666666666e-05,
+      "loss": 1.8196,
+      "step": 932
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.91442652329749e-05,
+      "loss": 1.6187,
+      "step": 933
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.912186379928316e-05,
+      "loss": 1.8667,
+      "step": 934
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.909946236559141e-05,
+      "loss": 0.4504,
+      "step": 935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.907706093189965e-05,
+      "loss": 0.6923,
+      "step": 936
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.905465949820788e-05,
+      "loss": 1.1389,
+      "step": 937
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.903225806451613e-05,
+      "loss": 1.4614,
+      "step": 938
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.900985663082437e-05,
+      "loss": 0.7169,
+      "step": 939
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.898745519713263e-05,
+      "loss": 1.0205,
+      "step": 940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.896505376344087e-05,
+      "loss": 0.4212,
+      "step": 941
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.894265232974911e-05,
+      "loss": 0.3751,
+      "step": 942
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.892025089605734e-05,
+      "loss": 0.6633,
+      "step": 943
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.889784946236559e-05,
+      "loss": 1.0117,
+      "step": 944
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.887544802867383e-05,
+      "loss": 0.9495,
+      "step": 945
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.885304659498209e-05,
+      "loss": 3.1533,
+      "step": 946
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.883064516129033e-05,
+      "loss": 1.2847,
+      "step": 947
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.880824372759858e-05,
+      "loss": 2.2734,
+      "step": 948
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.878584229390681e-05,
+      "loss": 0.0974,
+      "step": 949
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.876344086021505e-05,
+      "loss": 0.9856,
+      "step": 950
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.87410394265233e-05,
+      "loss": 1.6772,
+      "step": 951
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.871863799283154e-05,
+      "loss": 0.2524,
+      "step": 952
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.86962365591398e-05,
+      "loss": 5.0742,
+      "step": 953
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.867383512544804e-05,
+      "loss": 0.9263,
+      "step": 954
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.865143369175627e-05,
+      "loss": 1.4941,
+      "step": 955
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.862903225806451e-05,
+      "loss": 2.2373,
+      "step": 956
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.860663082437276e-05,
+      "loss": 0.96,
+      "step": 957
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.8584229390681e-05,
+      "loss": 2.48,
+      "step": 958
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.856182795698926e-05,
+      "loss": 1.5889,
+      "step": 959
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.85394265232975e-05,
+      "loss": 0.8099,
+      "step": 960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.851702508960573e-05,
+      "loss": 1.0735,
+      "step": 961
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.849462365591398e-05,
+      "loss": 0.4583,
+      "step": 962
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.847222222222222e-05,
+      "loss": 3.0645,
+      "step": 963
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.844982078853047e-05,
+      "loss": 4.2046,
+      "step": 964
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.842741935483871e-05,
+      "loss": 2.0555,
+      "step": 965
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.840501792114697e-05,
+      "loss": 0.9392,
+      "step": 966
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.83826164874552e-05,
+      "loss": 2.9594,
+      "step": 967
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.836021505376344e-05,
+      "loss": 1.4448,
+      "step": 968
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.833781362007168e-05,
+      "loss": 0.814,
+      "step": 969
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.831541218637993e-05,
+      "loss": 0.4761,
+      "step": 970
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.829301075268817e-05,
+      "loss": 0.5323,
+      "step": 971
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.827060931899643e-05,
+      "loss": 0.4292,
+      "step": 972
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.824820788530466e-05,
+      "loss": 0.8596,
+      "step": 973
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82258064516129e-05,
+      "loss": 1.1561,
+      "step": 974
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.820340501792115e-05,
+      "loss": 1.189,
+      "step": 975
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818100358422939e-05,
+      "loss": 0.7239,
+      "step": 976
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.815860215053764e-05,
+      "loss": 0.1214,
+      "step": 977
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81362007168459e-05,
+      "loss": 2.521,
+      "step": 978
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.811379928315412e-05,
+      "loss": 2.1324,
+      "step": 979
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.809139784946237e-05,
+      "loss": 1.147,
+      "step": 980
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.806899641577061e-05,
+      "loss": 2.0291,
+      "step": 981
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.804659498207886e-05,
+      "loss": 0.1995,
+      "step": 982
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.80241935483871e-05,
+      "loss": 1.7343,
+      "step": 983
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.800179211469534e-05,
+      "loss": 2.0667,
+      "step": 984
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.797939068100359e-05,
+      "loss": 1.7056,
+      "step": 985
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.795698924731183e-05,
+      "loss": 1.1552,
+      "step": 986
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.793458781362007e-05,
+      "loss": 0.1544,
+      "step": 987
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.791218637992832e-05,
+      "loss": 1.6089,
+      "step": 988
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.788978494623656e-05,
+      "loss": 0.9895,
+      "step": 989
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.78673835125448e-05,
+      "loss": 1.6501,
+      "step": 990
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.784498207885305e-05,
+      "loss": 1.9869,
+      "step": 991
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.78225806451613e-05,
+      "loss": 0.7802,
+      "step": 992
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.780017921146954e-05,
+      "loss": 0.5745,
+      "step": 993
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.777777777777778e-05,
+      "loss": 0.0585,
+      "step": 994
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.775537634408603e-05,
+      "loss": 0.7927,
+      "step": 995
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.773297491039427e-05,
+      "loss": 2.6152,
+      "step": 996
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.771057347670251e-05,
+      "loss": 1.249,
+      "step": 997
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.768817204301076e-05,
+      "loss": 1.6444,
+      "step": 998
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.7665770609319e-05,
+      "loss": 1.9186,
+      "step": 999
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.764336917562724e-05,
+      "loss": 0.9976,
+      "step": 1000
+    },
+    {
+      "epoch": 1.79,
+      "eval_accuracy": 0.4714285731315613,
+      "eval_loss": 1.5475257635116577,
+      "eval_runtime": 71.8489,
+      "eval_samples_per_second": 3.897,
+      "eval_steps_per_second": 3.897,
+      "step": 1000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.762096774193549e-05,
+      "loss": 2.8672,
+      "step": 1001
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.759856630824373e-05,
+      "loss": 2.0803,
+      "step": 1002
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.757616487455198e-05,
+      "loss": 0.202,
+      "step": 1003
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.755376344086022e-05,
+      "loss": 1.7153,
+      "step": 1004
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.753136200716846e-05,
+      "loss": 0.8655,
+      "step": 1005
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.750896057347671e-05,
+      "loss": 1.3639,
+      "step": 1006
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.748655913978495e-05,
+      "loss": 2.701,
+      "step": 1007
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.74641577060932e-05,
+      "loss": 2.2705,
+      "step": 1008
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.744175627240144e-05,
+      "loss": 1.6078,
+      "step": 1009
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.741935483870968e-05,
+      "loss": 0.874,
+      "step": 1010
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.739695340501793e-05,
+      "loss": 0.1712,
+      "step": 1011
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.737455197132617e-05,
+      "loss": 1.502,
+      "step": 1012
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.735215053763441e-05,
+      "loss": 0.07,
+      "step": 1013
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.732974910394266e-05,
+      "loss": 0.5346,
+      "step": 1014
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.73073476702509e-05,
+      "loss": 1.9307,
+      "step": 1015
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.728494623655915e-05,
+      "loss": 2.0566,
+      "step": 1016
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.726254480286739e-05,
+      "loss": 1.4526,
+      "step": 1017
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.724014336917563e-05,
+      "loss": 0.6086,
+      "step": 1018
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.721774193548388e-05,
+      "loss": 3.209,
+      "step": 1019
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.719534050179212e-05,
+      "loss": 0.2267,
+      "step": 1020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.717293906810035e-05,
+      "loss": 0.9615,
+      "step": 1021
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.715053763440861e-05,
+      "loss": 1.4376,
+      "step": 1022
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.712813620071685e-05,
+      "loss": 0.8532,
+      "step": 1023
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.71057347670251e-05,
+      "loss": 0.597,
+      "step": 1024
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.708333333333334e-05,
+      "loss": 2.1692,
+      "step": 1025
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.706093189964157e-05,
+      "loss": 1.793,
+      "step": 1026
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.703853046594982e-05,
+      "loss": 1.0454,
+      "step": 1027
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.701612903225807e-05,
+      "loss": 1.1287,
+      "step": 1028
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.699372759856632e-05,
+      "loss": 1.4194,
+      "step": 1029
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.697132616487456e-05,
+      "loss": 0.856,
+      "step": 1030
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.69489247311828e-05,
+      "loss": 2.3477,
+      "step": 1031
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.692652329749103e-05,
+      "loss": 1.6172,
+      "step": 1032
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.690412186379928e-05,
+      "loss": 1.0181,
+      "step": 1033
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.688172043010752e-05,
+      "loss": 1.0886,
+      "step": 1034
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.685931899641578e-05,
+      "loss": 0.3696,
+      "step": 1035
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.683691756272402e-05,
+      "loss": 3.6318,
+      "step": 1036
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.681451612903227e-05,
+      "loss": 2.7825,
+      "step": 1037
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.67921146953405e-05,
+      "loss": 2.0396,
+      "step": 1038
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.676971326164874e-05,
+      "loss": 1.5801,
+      "step": 1039
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.674731182795699e-05,
+      "loss": 0.7253,
+      "step": 1040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.672491039426524e-05,
+      "loss": 1.9897,
+      "step": 1041
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.670250896057349e-05,
+      "loss": 0.9763,
+      "step": 1042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.668010752688173e-05,
+      "loss": 1.1396,
+      "step": 1043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.665770609318996e-05,
+      "loss": 0.6596,
+      "step": 1044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.66353046594982e-05,
+      "loss": 0.9097,
+      "step": 1045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.661290322580645e-05,
+      "loss": 1.3569,
+      "step": 1046
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.659050179211469e-05,
+      "loss": 1.2688,
+      "step": 1047
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.656810035842295e-05,
+      "loss": 1.1831,
+      "step": 1048
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.65456989247312e-05,
+      "loss": 1.3794,
+      "step": 1049
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.652329749103942e-05,
+      "loss": 1.3579,
+      "step": 1050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.650089605734767e-05,
+      "loss": 0.5106,
+      "step": 1051
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.647849462365591e-05,
+      "loss": 1.2717,
+      "step": 1052
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.645609318996416e-05,
+      "loss": 1.1086,
+      "step": 1053
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.643369175627241e-05,
+      "loss": 1.0176,
+      "step": 1054
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.641129032258066e-05,
+      "loss": 0.7025,
+      "step": 1055
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.638888888888889e-05,
+      "loss": 1.4233,
+      "step": 1056
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.636648745519713e-05,
+      "loss": 2.4231,
+      "step": 1057
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.634408602150538e-05,
+      "loss": 1.7656,
+      "step": 1058
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.632168458781362e-05,
+      "loss": 1.6426,
+      "step": 1059
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.629928315412188e-05,
+      "loss": 0.3948,
+      "step": 1060
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.627688172043012e-05,
+      "loss": 1.4663,
+      "step": 1061
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.625448028673835e-05,
+      "loss": 0.4755,
+      "step": 1062
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.62320788530466e-05,
+      "loss": 1.1604,
+      "step": 1063
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.620967741935484e-05,
+      "loss": 0.0273,
+      "step": 1064
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.618727598566308e-05,
+      "loss": 1.2039,
+      "step": 1065
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.616487455197133e-05,
+      "loss": 1.2018,
+      "step": 1066
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.614247311827958e-05,
+      "loss": 0.2492,
+      "step": 1067
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.612007168458781e-05,
+      "loss": 2.2378,
+      "step": 1068
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.609767025089606e-05,
+      "loss": 1.0945,
+      "step": 1069
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.60752688172043e-05,
+      "loss": 1.8391,
+      "step": 1070
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.605286738351255e-05,
+      "loss": 0.1317,
+      "step": 1071
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.603046594982079e-05,
+      "loss": 0.7649,
+      "step": 1072
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.600806451612905e-05,
+      "loss": 0.3607,
+      "step": 1073
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.598566308243728e-05,
+      "loss": 1.9378,
+      "step": 1074
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.596326164874552e-05,
+      "loss": 1.0668,
+      "step": 1075
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.594086021505376e-05,
+      "loss": 1.1428,
+      "step": 1076
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.591845878136201e-05,
+      "loss": 0.1534,
+      "step": 1077
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.589605734767025e-05,
+      "loss": 0.3712,
+      "step": 1078
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.58736559139785e-05,
+      "loss": 1.2196,
+      "step": 1079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.585125448028674e-05,
+      "loss": 0.3439,
+      "step": 1080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.582885304659498e-05,
+      "loss": 2.1572,
+      "step": 1081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.580645161290323e-05,
+      "loss": 0.1022,
+      "step": 1082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.578405017921147e-05,
+      "loss": 1.4248,
+      "step": 1083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.576164874551972e-05,
+      "loss": 1.1132,
+      "step": 1084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.573924731182796e-05,
+      "loss": 1.3383,
+      "step": 1085
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.57168458781362e-05,
+      "loss": 1.915,
+      "step": 1086
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.569444444444445e-05,
+      "loss": 0.1524,
+      "step": 1087
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.567204301075269e-05,
+      "loss": 1.079,
+      "step": 1088
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.564964157706093e-05,
+      "loss": 0.0541,
+      "step": 1089
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.562724014336918e-05,
+      "loss": 0.4227,
+      "step": 1090
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.560483870967742e-05,
+      "loss": 2.5977,
+      "step": 1091
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.558243727598567e-05,
+      "loss": 1.3701,
+      "step": 1092
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.556003584229391e-05,
+      "loss": 1.8921,
+      "step": 1093
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.553763440860215e-05,
+      "loss": 0.363,
+      "step": 1094
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.55152329749104e-05,
+      "loss": 1.3594,
+      "step": 1095
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.549283154121864e-05,
+      "loss": 2.5776,
+      "step": 1096
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.547043010752689e-05,
+      "loss": 0.456,
+      "step": 1097
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.544802867383513e-05,
+      "loss": 1.4789,
+      "step": 1098
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.542562724014337e-05,
+      "loss": 2.9741,
+      "step": 1099
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.540322580645162e-05,
+      "loss": 0.8512,
+      "step": 1100
+    },
+    {
+      "epoch": 1.97,
+      "eval_accuracy": 0.5107142925262451,
+      "eval_loss": 1.427456259727478,
+      "eval_runtime": 68.4208,
+      "eval_samples_per_second": 4.092,
+      "eval_steps_per_second": 4.092,
+      "step": 1100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.538082437275986e-05,
+      "loss": 0.177,
+      "step": 1101
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.53584229390681e-05,
+      "loss": 0.1509,
+      "step": 1102
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.533602150537635e-05,
+      "loss": 2.4365,
+      "step": 1103
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.531362007168459e-05,
+      "loss": 2.5918,
+      "step": 1104
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.529121863799284e-05,
+      "loss": 0.9267,
+      "step": 1105
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.526881720430108e-05,
+      "loss": 3.0869,
+      "step": 1106
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 7.524641577060932e-05,
+      "loss": 3.8027,
+      "step": 1107
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.522401433691757e-05,
+      "loss": 1.626,
+      "step": 1108
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.520161290322581e-05,
+      "loss": 1.0806,
+      "step": 1109
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.517921146953406e-05,
+      "loss": 1.6025,
+      "step": 1110
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.51568100358423e-05,
+      "loss": 0.5961,
+      "step": 1111
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.513440860215054e-05,
+      "loss": 0.211,
+      "step": 1112
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 7.511200716845879e-05,
+      "loss": 0.1217,
+      "step": 1113
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.508960573476703e-05,
+      "loss": 1.0835,
+      "step": 1114
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.506720430107528e-05,
+      "loss": 2.7246,
+      "step": 1115
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.50448028673835e-05,
+      "loss": 1.0815,
+      "step": 1116
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.502240143369176e-05,
+      "loss": 1.6309,
+      "step": 1117
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.065,
+      "step": 1118
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.497759856630825e-05,
+      "loss": 1.4546,
+      "step": 1119
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.49551971326165e-05,
+      "loss": 0.639,
+      "step": 1120
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.493279569892473e-05,
+      "loss": 1.6753,
+      "step": 1121
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.491039426523297e-05,
+      "loss": 0.1378,
+      "step": 1122
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.488799283154123e-05,
+      "loss": 0.2998,
+      "step": 1123
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 7.486559139784947e-05,
+      "loss": 2.4194,
+      "step": 1124
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.484318996415771e-05,
+      "loss": 1.9736,
+      "step": 1125
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.482078853046596e-05,
+      "loss": 0.6331,
+      "step": 1126
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.479838709677419e-05,
+      "loss": 0.5656,
+      "step": 1127
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.477598566308243e-05,
+      "loss": 0.174,
+      "step": 1128
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 7.475358422939069e-05,
+      "loss": 0.2468,
+      "step": 1129
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.473118279569893e-05,
+      "loss": 0.0838,
+      "step": 1130
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.470878136200718e-05,
+      "loss": 0.6094,
+      "step": 1131
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.468637992831542e-05,
+      "loss": 1.7049,
+      "step": 1132
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.466397849462365e-05,
+      "loss": 1.4821,
+      "step": 1133
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.46415770609319e-05,
+      "loss": 1.4834,
+      "step": 1134
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 7.461917562724014e-05,
+      "loss": 0.0044,
+      "step": 1135
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.45967741935484e-05,
+      "loss": 0.0336,
+      "step": 1136
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.457437275985664e-05,
+      "loss": 2.9839,
+      "step": 1137
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.455197132616488e-05,
+      "loss": 1.7681,
+      "step": 1138
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.452956989247311e-05,
+      "loss": 0.0167,
+      "step": 1139
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.450716845878136e-05,
+      "loss": 0.3078,
+      "step": 1140
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 7.44847670250896e-05,
+      "loss": 0.4535,
+      "step": 1141
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.446236559139786e-05,
+      "loss": 1.0175,
+      "step": 1142
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.44399641577061e-05,
+      "loss": 1.2373,
+      "step": 1143
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.441756272401435e-05,
+      "loss": 0.2878,
+      "step": 1144
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.439516129032258e-05,
+      "loss": 0.0072,
+      "step": 1145
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 7.437275985663082e-05,
+      "loss": 0.7262,
+      "step": 1146
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.435035842293907e-05,
+      "loss": 1.2964,
+      "step": 1147
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.432795698924731e-05,
+      "loss": 0.4963,
+      "step": 1148
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.430555555555557e-05,
+      "loss": 0.8378,
+      "step": 1149
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.428315412186381e-05,
+      "loss": 0.2937,
+      "step": 1150
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.426075268817204e-05,
+      "loss": 0.9787,
+      "step": 1151
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 7.423835125448028e-05,
+      "loss": 1.5408,
+      "step": 1152
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.421594982078853e-05,
+      "loss": 1.0588,
+      "step": 1153
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.419354838709677e-05,
+      "loss": 0.3164,
+      "step": 1154
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.417114695340503e-05,
+      "loss": 0.4333,
+      "step": 1155
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.414874551971327e-05,
+      "loss": 1.3704,
+      "step": 1156
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 7.41263440860215e-05,
+      "loss": 3.1211,
+      "step": 1157
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.410394265232975e-05,
+      "loss": 1.054,
+      "step": 1158
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.408154121863799e-05,
+      "loss": 1.4521,
+      "step": 1159
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.405913978494624e-05,
+      "loss": 0.7332,
+      "step": 1160
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.403673835125449e-05,
+      "loss": 2.9287,
+      "step": 1161
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.401433691756274e-05,
+      "loss": 1.1777,
+      "step": 1162
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 7.399193548387097e-05,
+      "loss": 0.9929,
+      "step": 1163
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.396953405017921e-05,
+      "loss": 0.4992,
+      "step": 1164
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.394713261648745e-05,
+      "loss": 1.1924,
+      "step": 1165
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.39247311827957e-05,
+      "loss": 0.5521,
+      "step": 1166
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.390232974910394e-05,
+      "loss": 0.3939,
+      "step": 1167
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.38799283154122e-05,
+      "loss": 0.1292,
+      "step": 1168
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 7.385752688172043e-05,
+      "loss": 0.4818,
+      "step": 1169
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.383512544802867e-05,
+      "loss": 0.3264,
+      "step": 1170
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.381272401433692e-05,
+      "loss": 0.4059,
+      "step": 1171
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.379032258064516e-05,
+      "loss": 0.1539,
+      "step": 1172
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.37679211469534e-05,
+      "loss": 0.9587,
+      "step": 1173
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 7.374551971326166e-05,
+      "loss": 0.3022,
+      "step": 1174
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.37231182795699e-05,
+      "loss": 1.8271,
+      "step": 1175
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.370071684587814e-05,
+      "loss": 1.2603,
+      "step": 1176
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.367831541218638e-05,
+      "loss": 3.2998,
+      "step": 1177
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.365591397849463e-05,
+      "loss": 0.2936,
+      "step": 1178
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.363351254480287e-05,
+      "loss": 2.522,
+      "step": 1179
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 7.361111111111111e-05,
+      "loss": 1.7012,
+      "step": 1180
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.358870967741936e-05,
+      "loss": 0.5223,
+      "step": 1181
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.35663082437276e-05,
+      "loss": 0.0509,
+      "step": 1182
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.354390681003584e-05,
+      "loss": 1.8867,
+      "step": 1183
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.352150537634409e-05,
+      "loss": 0.7935,
+      "step": 1184
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 7.349910394265233e-05,
+      "loss": 0.463,
+      "step": 1185
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.347670250896058e-05,
+      "loss": 0.574,
+      "step": 1186
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.345430107526882e-05,
+      "loss": 0.4169,
+      "step": 1187
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.343189964157706e-05,
+      "loss": 0.3526,
+      "step": 1188
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.340949820788531e-05,
+      "loss": 0.5065,
+      "step": 1189
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.338709677419355e-05,
+      "loss": 1.3012,
+      "step": 1190
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 7.33646953405018e-05,
+      "loss": 0.0649,
+      "step": 1191
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.334229390681004e-05,
+      "loss": 0.4318,
+      "step": 1192
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.331989247311828e-05,
+      "loss": 0.3875,
+      "step": 1193
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.329749103942653e-05,
+      "loss": 0.7372,
+      "step": 1194
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.327508960573477e-05,
+      "loss": 1.3831,
+      "step": 1195
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 7.325268817204301e-05,
+      "loss": 1.0366,
+      "step": 1196
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.323028673835126e-05,
+      "loss": 0.7322,
+      "step": 1197
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.32078853046595e-05,
+      "loss": 0.5963,
+      "step": 1198
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.318548387096775e-05,
+      "loss": 0.2065,
+      "step": 1199
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.316308243727599e-05,
+      "loss": 0.0304,
+      "step": 1200
+    },
+    {
+      "epoch": 2.15,
+      "eval_accuracy": 0.5464285612106323,
+      "eval_loss": 1.3712011575698853,
+      "eval_runtime": 72.6437,
+      "eval_samples_per_second": 3.854,
+      "eval_steps_per_second": 3.854,
+      "step": 1200
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.314068100358423e-05,
+      "loss": 1.2666,
+      "step": 1201
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 7.311827956989248e-05,
+      "loss": 1.5315,
+      "step": 1202
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.309587813620072e-05,
+      "loss": 2.5492,
+      "step": 1203
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.307347670250897e-05,
+      "loss": 0.3435,
+      "step": 1204
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.305107526881721e-05,
+      "loss": 0.5912,
+      "step": 1205
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.302867383512545e-05,
+      "loss": 0.3219,
+      "step": 1206
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.30062724014337e-05,
+      "loss": 0.0279,
+      "step": 1207
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 7.298387096774194e-05,
+      "loss": 0.4967,
+      "step": 1208
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.296146953405018e-05,
+      "loss": 0.9169,
+      "step": 1209
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.293906810035843e-05,
+      "loss": 0.9258,
+      "step": 1210
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.291666666666667e-05,
+      "loss": 1.8357,
+      "step": 1211
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.289426523297492e-05,
+      "loss": 0.1183,
+      "step": 1212
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 7.287186379928316e-05,
+      "loss": 0.4578,
+      "step": 1213
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.28494623655914e-05,
+      "loss": 0.4351,
+      "step": 1214
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.282706093189965e-05,
+      "loss": 0.626,
+      "step": 1215
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.280465949820788e-05,
+      "loss": 2.1978,
+      "step": 1216
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.278225806451612e-05,
+      "loss": 0.0547,
+      "step": 1217
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.275985663082438e-05,
+      "loss": 0.9761,
+      "step": 1218
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 7.273745519713262e-05,
+      "loss": 2.0412,
+      "step": 1219
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.271505376344087e-05,
+      "loss": 0.0705,
+      "step": 1220
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.269265232974911e-05,
+      "loss": 0.1428,
+      "step": 1221
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.267025089605734e-05,
+      "loss": 0.335,
+      "step": 1222
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.264784946236559e-05,
+      "loss": 0.3334,
+      "step": 1223
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 7.262544802867384e-05,
+      "loss": 0.318,
+      "step": 1224
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.260304659498209e-05,
+      "loss": 3.166,
+      "step": 1225
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.258064516129033e-05,
+      "loss": 0.4047,
+      "step": 1226
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.255824372759857e-05,
+      "loss": 1.3428,
+      "step": 1227
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.25358422939068e-05,
+      "loss": 2.1782,
+      "step": 1228
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.251344086021505e-05,
+      "loss": 0.6939,
+      "step": 1229
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.24910394265233e-05,
+      "loss": 0.5728,
+      "step": 1230
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.246863799283155e-05,
+      "loss": 0.6873,
+      "step": 1231
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.24462365591398e-05,
+      "loss": 0.7395,
+      "step": 1232
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.242383512544804e-05,
+      "loss": 0.0398,
+      "step": 1233
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.240143369175627e-05,
+      "loss": 0.6733,
+      "step": 1234
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.237903225806451e-05,
+      "loss": 0.767,
+      "step": 1235
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.235663082437276e-05,
+      "loss": 1.1877,
+      "step": 1236
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.233422939068101e-05,
+      "loss": 3.0905,
+      "step": 1237
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.231182795698926e-05,
+      "loss": 3.6279,
+      "step": 1238
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.22894265232975e-05,
+      "loss": 0.2471,
+      "step": 1239
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.226702508960573e-05,
+      "loss": 2.5187,
+      "step": 1240
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.224462365591398e-05,
+      "loss": 0.5319,
+      "step": 1241
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.222222222222222e-05,
+      "loss": 1.0938,
+      "step": 1242
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.219982078853048e-05,
+      "loss": 1.0259,
+      "step": 1243
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.217741935483872e-05,
+      "loss": 0.425,
+      "step": 1244
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.215501792114696e-05,
+      "loss": 0.5832,
+      "step": 1245
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.21326164874552e-05,
+      "loss": 1.4336,
+      "step": 1246
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.211021505376344e-05,
+      "loss": 1.0503,
+      "step": 1247
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.208781362007168e-05,
+      "loss": 0.3872,
+      "step": 1248
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.206541218637993e-05,
+      "loss": 1.6108,
+      "step": 1249
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.204301075268818e-05,
+      "loss": 0.2674,
+      "step": 1250
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.202060931899643e-05,
+      "loss": 3.1968,
+      "step": 1251
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.199820788530466e-05,
+      "loss": 0.7699,
+      "step": 1252
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.19758064516129e-05,
+      "loss": 0.816,
+      "step": 1253
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.195340501792115e-05,
+      "loss": 0.0578,
+      "step": 1254
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.193100358422939e-05,
+      "loss": 0.084,
+      "step": 1255
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.190860215053765e-05,
+      "loss": 0.1936,
+      "step": 1256
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.188620071684589e-05,
+      "loss": 1.5447,
+      "step": 1257
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.186379928315412e-05,
+      "loss": 0.0762,
+      "step": 1258
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.184139784946236e-05,
+      "loss": 1.587,
+      "step": 1259
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.181899641577061e-05,
+      "loss": 0.148,
+      "step": 1260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.179659498207885e-05,
+      "loss": 0.118,
+      "step": 1261
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.177419354838711e-05,
+      "loss": 0.219,
+      "step": 1262
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.175179211469535e-05,
+      "loss": 0.1507,
+      "step": 1263
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.172939068100358e-05,
+      "loss": 0.8237,
+      "step": 1264
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.170698924731183e-05,
+      "loss": 0.1371,
+      "step": 1265
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.168458781362007e-05,
+      "loss": 2.3354,
+      "step": 1266
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.166218637992832e-05,
+      "loss": 0.1139,
+      "step": 1267
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.163978494623656e-05,
+      "loss": 0.361,
+      "step": 1268
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.161738351254482e-05,
+      "loss": 1.0671,
+      "step": 1269
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.159498207885305e-05,
+      "loss": 1.917,
+      "step": 1270
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.157258064516129e-05,
+      "loss": 0.5178,
+      "step": 1271
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.155017921146953e-05,
+      "loss": 0.2438,
+      "step": 1272
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.152777777777778e-05,
+      "loss": 1.5649,
+      "step": 1273
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.150537634408602e-05,
+      "loss": 2.0386,
+      "step": 1274
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.148297491039428e-05,
+      "loss": 1.7915,
+      "step": 1275
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.146057347670251e-05,
+      "loss": 0.5459,
+      "step": 1276
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.143817204301075e-05,
+      "loss": 0.4764,
+      "step": 1277
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.1415770609319e-05,
+      "loss": 0.8708,
+      "step": 1278
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.139336917562724e-05,
+      "loss": 1.7861,
+      "step": 1279
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.137096774193549e-05,
+      "loss": 0.083,
+      "step": 1280
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.134856630824373e-05,
+      "loss": 0.5055,
+      "step": 1281
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.132616487455197e-05,
+      "loss": 2.5425,
+      "step": 1282
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.130376344086022e-05,
+      "loss": 2.8982,
+      "step": 1283
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.128136200716846e-05,
+      "loss": 0.2093,
+      "step": 1284
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.12589605734767e-05,
+      "loss": 1.2324,
+      "step": 1285
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.123655913978495e-05,
+      "loss": 1.7659,
+      "step": 1286
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.121415770609319e-05,
+      "loss": 1.3555,
+      "step": 1287
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.119175627240144e-05,
+      "loss": 0.3705,
+      "step": 1288
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.116935483870968e-05,
+      "loss": 0.0613,
+      "step": 1289
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.114695340501792e-05,
+      "loss": 1.4536,
+      "step": 1290
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.112455197132617e-05,
+      "loss": 1.7085,
+      "step": 1291
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.110215053763441e-05,
+      "loss": 4.8633,
+      "step": 1292
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.107974910394266e-05,
+      "loss": 1.6104,
+      "step": 1293
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.10573476702509e-05,
+      "loss": 1.479,
+      "step": 1294
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.103494623655914e-05,
+      "loss": 0.0378,
+      "step": 1295
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.101254480286739e-05,
+      "loss": 0.9243,
+      "step": 1296
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.099014336917563e-05,
+      "loss": 1.7075,
+      "step": 1297
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.096774193548388e-05,
+      "loss": 1.7358,
+      "step": 1298
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.094534050179212e-05,
+      "loss": 1.7489,
+      "step": 1299
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.092293906810036e-05,
+      "loss": 0.0159,
+      "step": 1300
+    },
+    {
+      "epoch": 2.33,
+      "eval_accuracy": 0.574999988079071,
+      "eval_loss": 1.2517281770706177,
+      "eval_runtime": 72.2478,
+      "eval_samples_per_second": 3.876,
+      "eval_steps_per_second": 3.876,
+      "step": 1300
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.090053763440861e-05,
+      "loss": 1.4635,
+      "step": 1301
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.087813620071685e-05,
+      "loss": 0.062,
+      "step": 1302
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.08557347670251e-05,
+      "loss": 0.0485,
+      "step": 1303
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.083333333333334e-05,
+      "loss": 0.0606,
+      "step": 1304
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.081093189964157e-05,
+      "loss": 2.4578,
+      "step": 1305
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.078853046594983e-05,
+      "loss": 0.229,
+      "step": 1306
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.076612903225807e-05,
+      "loss": 0.5273,
+      "step": 1307
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.074372759856631e-05,
+      "loss": 0.8966,
+      "step": 1308
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.072132616487456e-05,
+      "loss": 0.3737,
+      "step": 1309
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.06989247311828e-05,
+      "loss": 0.6891,
+      "step": 1310
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.067652329749103e-05,
+      "loss": 1.9072,
+      "step": 1311
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.065412186379929e-05,
+      "loss": 0.3431,
+      "step": 1312
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.063172043010753e-05,
+      "loss": 3.1299,
+      "step": 1313
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.060931899641578e-05,
+      "loss": 0.6583,
+      "step": 1314
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.058691756272402e-05,
+      "loss": 0.2449,
+      "step": 1315
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.056451612903226e-05,
+      "loss": 0.258,
+      "step": 1316
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.05421146953405e-05,
+      "loss": 2.0439,
+      "step": 1317
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.051971326164874e-05,
+      "loss": 0.0088,
+      "step": 1318
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.0497311827957e-05,
+      "loss": 0.2888,
+      "step": 1319
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.047491039426524e-05,
+      "loss": 2.5803,
+      "step": 1320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.045250896057348e-05,
+      "loss": 0.6903,
+      "step": 1321
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.043010752688173e-05,
+      "loss": 0.9184,
+      "step": 1322
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.040770609318996e-05,
+      "loss": 0.918,
+      "step": 1323
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.03853046594982e-05,
+      "loss": 0.305,
+      "step": 1324
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.036290322580646e-05,
+      "loss": 0.3347,
+      "step": 1325
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.03405017921147e-05,
+      "loss": 0.761,
+      "step": 1326
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.031810035842295e-05,
+      "loss": 0.6304,
+      "step": 1327
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.029569892473119e-05,
+      "loss": 2.0106,
+      "step": 1328
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.027329749103942e-05,
+      "loss": 0.1083,
+      "step": 1329
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.025089605734767e-05,
+      "loss": 0.0068,
+      "step": 1330
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.022849462365592e-05,
+      "loss": 1.103,
+      "step": 1331
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.020609318996417e-05,
+      "loss": 1.0818,
+      "step": 1332
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.018369175627241e-05,
+      "loss": 1.3018,
+      "step": 1333
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.016129032258065e-05,
+      "loss": 0.6389,
+      "step": 1334
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.013888888888888e-05,
+      "loss": 0.8979,
+      "step": 1335
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.011648745519713e-05,
+      "loss": 0.4564,
+      "step": 1336
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.009408602150537e-05,
+      "loss": 0.1543,
+      "step": 1337
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.007168458781363e-05,
+      "loss": 0.3697,
+      "step": 1338
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.004928315412187e-05,
+      "loss": 3.2539,
+      "step": 1339
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.002688172043012e-05,
+      "loss": 0.1488,
+      "step": 1340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.000448028673835e-05,
+      "loss": 0.0343,
+      "step": 1341
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.998207885304659e-05,
+      "loss": 0.4142,
+      "step": 1342
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.995967741935484e-05,
+      "loss": 0.137,
+      "step": 1343
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.993727598566309e-05,
+      "loss": 0.2138,
+      "step": 1344
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.991487455197134e-05,
+      "loss": 0.0132,
+      "step": 1345
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.989247311827958e-05,
+      "loss": 0.0051,
+      "step": 1346
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 6.987007168458781e-05,
+      "loss": 0.0548,
+      "step": 1347
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.984767025089605e-05,
+      "loss": 1.6956,
+      "step": 1348
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.98252688172043e-05,
+      "loss": 0.136,
+      "step": 1349
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.980286738351254e-05,
+      "loss": 0.1538,
+      "step": 1350
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.97804659498208e-05,
+      "loss": 0.4677,
+      "step": 1351
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.975806451612904e-05,
+      "loss": 4.2939,
+      "step": 1352
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 6.973566308243727e-05,
+      "loss": 0.0344,
+      "step": 1353
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.971326164874552e-05,
+      "loss": 0.1729,
+      "step": 1354
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.969086021505376e-05,
+      "loss": 1.7356,
+      "step": 1355
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.9668458781362e-05,
+      "loss": 0.5868,
+      "step": 1356
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.964605734767026e-05,
+      "loss": 0.0154,
+      "step": 1357
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 6.962365591397851e-05,
+      "loss": 2.9897,
+      "step": 1358
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.960125448028674e-05,
+      "loss": 1.6955,
+      "step": 1359
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.957885304659498e-05,
+      "loss": 0.0263,
+      "step": 1360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.955645161290323e-05,
+      "loss": 2.9023,
+      "step": 1361
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.953405017921147e-05,
+      "loss": 1.6306,
+      "step": 1362
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.951164874551973e-05,
+      "loss": 0.4573,
+      "step": 1363
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.948924731182797e-05,
+      "loss": 0.0319,
+      "step": 1364
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.94668458781362e-05,
+      "loss": 0.0075,
+      "step": 1365
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.944444444444444e-05,
+      "loss": 0.0055,
+      "step": 1366
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.942204301075269e-05,
+      "loss": 0.4745,
+      "step": 1367
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.939964157706093e-05,
+      "loss": 0.7544,
+      "step": 1368
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.937724014336918e-05,
+      "loss": 0.0431,
+      "step": 1369
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.935483870967743e-05,
+      "loss": 0.0186,
+      "step": 1370
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.933243727598566e-05,
+      "loss": 0.0401,
+      "step": 1371
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.931003584229391e-05,
+      "loss": 3.8564,
+      "step": 1372
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.928763440860215e-05,
+      "loss": 1.6675,
+      "step": 1373
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.92652329749104e-05,
+      "loss": 1.3076,
+      "step": 1374
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.924283154121864e-05,
+      "loss": 3.694,
+      "step": 1375
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.92204301075269e-05,
+      "loss": 1.9456,
+      "step": 1376
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.919802867383513e-05,
+      "loss": 1.5937,
+      "step": 1377
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.917562724014337e-05,
+      "loss": 0.3119,
+      "step": 1378
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.915322580645161e-05,
+      "loss": 3.04,
+      "step": 1379
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.913082437275986e-05,
+      "loss": 0.0446,
+      "step": 1380
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.91084229390681e-05,
+      "loss": 1.4175,
+      "step": 1381
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.908602150537635e-05,
+      "loss": 2.9658,
+      "step": 1382
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.906362007168459e-05,
+      "loss": 0.7922,
+      "step": 1383
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.904121863799283e-05,
+      "loss": 0.7257,
+      "step": 1384
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.901881720430108e-05,
+      "loss": 0.1062,
+      "step": 1385
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.899641577060932e-05,
+      "loss": 1.9233,
+      "step": 1386
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.897401433691757e-05,
+      "loss": 1.2731,
+      "step": 1387
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.895161290322581e-05,
+      "loss": 0.3265,
+      "step": 1388
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.892921146953405e-05,
+      "loss": 0.0479,
+      "step": 1389
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.89068100358423e-05,
+      "loss": 0.0285,
+      "step": 1390
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.888440860215054e-05,
+      "loss": 2.9868,
+      "step": 1391
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.886200716845878e-05,
+      "loss": 0.7749,
+      "step": 1392
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.883960573476703e-05,
+      "loss": 0.2287,
+      "step": 1393
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.881720430107527e-05,
+      "loss": 2.2798,
+      "step": 1394
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.879480286738352e-05,
+      "loss": 2.2275,
+      "step": 1395
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.877240143369176e-05,
+      "loss": 1.4648,
+      "step": 1396
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.875e-05,
+      "loss": 0.011,
+      "step": 1397
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.872759856630825e-05,
+      "loss": 0.5698,
+      "step": 1398
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.870519713261649e-05,
+      "loss": 0.709,
+      "step": 1399
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.868279569892474e-05,
+      "loss": 0.6615,
+      "step": 1400
+    },
+    {
+      "epoch": 2.51,
+      "eval_accuracy": 0.5428571701049805,
+      "eval_loss": 1.4520440101623535,
+      "eval_runtime": 72.6581,
+      "eval_samples_per_second": 3.854,
+      "eval_steps_per_second": 3.854,
+      "step": 1400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.866039426523298e-05,
+      "loss": 0.0415,
+      "step": 1401
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.863799283154122e-05,
+      "loss": 1.4161,
+      "step": 1402
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.861559139784947e-05,
+      "loss": 0.7386,
+      "step": 1403
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.859318996415771e-05,
+      "loss": 0.1957,
+      "step": 1404
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.857078853046596e-05,
+      "loss": 0.3132,
+      "step": 1405
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.854838709677419e-05,
+      "loss": 0.5039,
+      "step": 1406
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.852598566308244e-05,
+      "loss": 0.4355,
+      "step": 1407
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.850358422939069e-05,
+      "loss": 0.5018,
+      "step": 1408
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.848118279569893e-05,
+      "loss": 0.1687,
+      "step": 1409
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.845878136200717e-05,
+      "loss": 0.8709,
+      "step": 1410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.843637992831542e-05,
+      "loss": 0.0659,
+      "step": 1411
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.841397849462365e-05,
+      "loss": 2.2664,
+      "step": 1412
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.83915770609319e-05,
+      "loss": 1.8903,
+      "step": 1413
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.836917562724015e-05,
+      "loss": 1.0728,
+      "step": 1414
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.83467741935484e-05,
+      "loss": 2.4663,
+      "step": 1415
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.832437275985664e-05,
+      "loss": 0.0164,
+      "step": 1416
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.830197132616488e-05,
+      "loss": 0.2553,
+      "step": 1417
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.827956989247311e-05,
+      "loss": 1.2539,
+      "step": 1418
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.825716845878136e-05,
+      "loss": 1.237,
+      "step": 1419
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.823476702508961e-05,
+      "loss": 2.1656,
+      "step": 1420
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.821236559139786e-05,
+      "loss": 2.1243,
+      "step": 1421
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.81899641577061e-05,
+      "loss": 0.0413,
+      "step": 1422
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.816756272401434e-05,
+      "loss": 0.0523,
+      "step": 1423
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.814516129032257e-05,
+      "loss": 0.55,
+      "step": 1424
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.812275985663082e-05,
+      "loss": 0.0957,
+      "step": 1425
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.810035842293908e-05,
+      "loss": 0.0229,
+      "step": 1426
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.807795698924732e-05,
+      "loss": 0.2871,
+      "step": 1427
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.805555555555556e-05,
+      "loss": 0.0151,
+      "step": 1428
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.803315412186381e-05,
+      "loss": 0.0037,
+      "step": 1429
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.801075268817204e-05,
+      "loss": 0.4795,
+      "step": 1430
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.798835125448028e-05,
+      "loss": 0.0048,
+      "step": 1431
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.796594982078853e-05,
+      "loss": 1.0339,
+      "step": 1432
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.794354838709678e-05,
+      "loss": 1.4697,
+      "step": 1433
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.792114695340503e-05,
+      "loss": 0.0741,
+      "step": 1434
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.789874551971327e-05,
+      "loss": 0.125,
+      "step": 1435
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.78763440860215e-05,
+      "loss": 0.3217,
+      "step": 1436
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.785394265232975e-05,
+      "loss": 2.1173,
+      "step": 1437
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.783154121863799e-05,
+      "loss": 0.1514,
+      "step": 1438
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.780913978494625e-05,
+      "loss": 0.4083,
+      "step": 1439
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.778673835125449e-05,
+      "loss": 0.2237,
+      "step": 1440
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.776433691756273e-05,
+      "loss": 0.3427,
+      "step": 1441
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.774193548387096e-05,
+      "loss": 0.7031,
+      "step": 1442
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.771953405017921e-05,
+      "loss": 0.025,
+      "step": 1443
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.769713261648745e-05,
+      "loss": 0.4714,
+      "step": 1444
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.767473118279571e-05,
+      "loss": 0.0966,
+      "step": 1445
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.765232974910395e-05,
+      "loss": 0.6847,
+      "step": 1446
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.76299283154122e-05,
+      "loss": 0.0602,
+      "step": 1447
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.760752688172043e-05,
+      "loss": 0.207,
+      "step": 1448
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.758512544802867e-05,
+      "loss": 3.8429,
+      "step": 1449
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.756272401433692e-05,
+      "loss": 0.2631,
+      "step": 1450
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.754032258064516e-05,
+      "loss": 1.1927,
+      "step": 1451
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.751792114695342e-05,
+      "loss": 0.1019,
+      "step": 1452
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.749551971326166e-05,
+      "loss": 0.9542,
+      "step": 1453
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.747311827956989e-05,
+      "loss": 0.0183,
+      "step": 1454
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.745071684587813e-05,
+      "loss": 1.295,
+      "step": 1455
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.742831541218638e-05,
+      "loss": 0.01,
+      "step": 1456
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.740591397849462e-05,
+      "loss": 0.0138,
+      "step": 1457
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.738351254480288e-05,
+      "loss": 0.0116,
+      "step": 1458
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.736111111111112e-05,
+      "loss": 0.1123,
+      "step": 1459
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.733870967741935e-05,
+      "loss": 0.0113,
+      "step": 1460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.73163082437276e-05,
+      "loss": 0.1059,
+      "step": 1461
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.729390681003584e-05,
+      "loss": 0.0212,
+      "step": 1462
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.727150537634409e-05,
+      "loss": 1.7229,
+      "step": 1463
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.724910394265233e-05,
+      "loss": 0.6398,
+      "step": 1464
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.722670250896059e-05,
+      "loss": 0.8136,
+      "step": 1465
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.720430107526882e-05,
+      "loss": 0.2155,
+      "step": 1466
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.718189964157706e-05,
+      "loss": 2.8564,
+      "step": 1467
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.71594982078853e-05,
+      "loss": 0.1213,
+      "step": 1468
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.713709677419355e-05,
+      "loss": 2.1088,
+      "step": 1469
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.711469534050179e-05,
+      "loss": 0.0103,
+      "step": 1470
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.709229390681005e-05,
+      "loss": 0.907,
+      "step": 1471
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.706989247311828e-05,
+      "loss": 4.7783,
+      "step": 1472
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.704749103942652e-05,
+      "loss": 0.5856,
+      "step": 1473
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.702508960573477e-05,
+      "loss": 0.2397,
+      "step": 1474
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.700268817204301e-05,
+      "loss": 3.2148,
+      "step": 1475
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.698028673835126e-05,
+      "loss": 0.3055,
+      "step": 1476
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.695788530465951e-05,
+      "loss": 0.1264,
+      "step": 1477
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.693548387096774e-05,
+      "loss": 0.6147,
+      "step": 1478
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.691308243727599e-05,
+      "loss": 0.9165,
+      "step": 1479
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.689068100358423e-05,
+      "loss": 1.6201,
+      "step": 1480
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.686827956989248e-05,
+      "loss": 0.3284,
+      "step": 1481
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.684587813620072e-05,
+      "loss": 0.0841,
+      "step": 1482
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.682347670250896e-05,
+      "loss": 0.065,
+      "step": 1483
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.68010752688172e-05,
+      "loss": 0.8459,
+      "step": 1484
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.677867383512545e-05,
+      "loss": 2.8351,
+      "step": 1485
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.67562724014337e-05,
+      "loss": 0.0077,
+      "step": 1486
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.673387096774194e-05,
+      "loss": 1.072,
+      "step": 1487
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.671146953405018e-05,
+      "loss": 2.5537,
+      "step": 1488
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.668906810035843e-05,
+      "loss": 4.3037,
+      "step": 1489
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.1138,
+      "step": 1490
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.664426523297491e-05,
+      "loss": 0.0035,
+      "step": 1491
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.662186379928316e-05,
+      "loss": 0.0932,
+      "step": 1492
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.65994623655914e-05,
+      "loss": 0.0624,
+      "step": 1493
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.657706093189965e-05,
+      "loss": 0.2815,
+      "step": 1494
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.655465949820789e-05,
+      "loss": 1.79,
+      "step": 1495
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.653225806451613e-05,
+      "loss": 0.0017,
+      "step": 1496
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.650985663082438e-05,
+      "loss": 3.2533,
+      "step": 1497
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.648745519713262e-05,
+      "loss": 3.2997,
+      "step": 1498
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.646505376344086e-05,
+      "loss": 0.0488,
+      "step": 1499
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.644265232974911e-05,
+      "loss": 1.7532,
+      "step": 1500
+    },
+    {
+      "epoch": 2.69,
+      "eval_accuracy": 0.5535714030265808,
+      "eval_loss": 1.6732311248779297,
+      "eval_runtime": 72.5913,
+      "eval_samples_per_second": 3.857,
+      "eval_steps_per_second": 3.857,
+      "step": 1500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.642025089605734e-05,
+      "loss": 0.5277,
+      "step": 1501
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.63978494623656e-05,
+      "loss": 1.7285,
+      "step": 1502
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 6.637544802867384e-05,
+      "loss": 0.862,
+      "step": 1503
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.635304659498208e-05,
+      "loss": 1.7075,
+      "step": 1504
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.633064516129033e-05,
+      "loss": 1.3007,
+      "step": 1505
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.630824372759857e-05,
+      "loss": 0.0018,
+      "step": 1506
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.62858422939068e-05,
+      "loss": 0.043,
+      "step": 1507
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.626344086021506e-05,
+      "loss": 0.0428,
+      "step": 1508
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 6.62410394265233e-05,
+      "loss": 0.0357,
+      "step": 1509
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.621863799283155e-05,
+      "loss": 0.0289,
+      "step": 1510
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.619623655913979e-05,
+      "loss": 2.3282,
+      "step": 1511
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.617383512544803e-05,
+      "loss": 1.1104,
+      "step": 1512
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.615143369175627e-05,
+      "loss": 0.0033,
+      "step": 1513
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 6.612903225806452e-05,
+      "loss": 0.0269,
+      "step": 1514
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.610663082437277e-05,
+      "loss": 0.0355,
+      "step": 1515
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.608422939068101e-05,
+      "loss": 1.0957,
+      "step": 1516
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.606182795698925e-05,
+      "loss": 2.9209,
+      "step": 1517
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.60394265232975e-05,
+      "loss": 0.0126,
+      "step": 1518
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.601702508960573e-05,
+      "loss": 1.3969,
+      "step": 1519
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 6.599462365591397e-05,
+      "loss": 3.8626,
+      "step": 1520
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.597222222222223e-05,
+      "loss": 2.6569,
+      "step": 1521
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.594982078853047e-05,
+      "loss": 0.0569,
+      "step": 1522
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.592741935483872e-05,
+      "loss": 0.0044,
+      "step": 1523
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.590501792114696e-05,
+      "loss": 0.0777,
+      "step": 1524
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.588261648745519e-05,
+      "loss": 1.4592,
+      "step": 1525
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 6.586021505376344e-05,
+      "loss": 1.8533,
+      "step": 1526
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.583781362007169e-05,
+      "loss": 2.668,
+      "step": 1527
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.581541218637994e-05,
+      "loss": 1.4333,
+      "step": 1528
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.579301075268818e-05,
+      "loss": 0.486,
+      "step": 1529
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.577060931899642e-05,
+      "loss": 0.5677,
+      "step": 1530
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 6.574820788530465e-05,
+      "loss": 0.1495,
+      "step": 1531
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.57258064516129e-05,
+      "loss": 0.5447,
+      "step": 1532
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.570340501792114e-05,
+      "loss": 0.4986,
+      "step": 1533
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.56810035842294e-05,
+      "loss": 0.9862,
+      "step": 1534
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.565860215053764e-05,
+      "loss": 0.7301,
+      "step": 1535
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.563620071684589e-05,
+      "loss": 0.0187,
+      "step": 1536
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 6.561379928315412e-05,
+      "loss": 0.2794,
+      "step": 1537
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.559139784946236e-05,
+      "loss": 0.1978,
+      "step": 1538
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.55689964157706e-05,
+      "loss": 0.0109,
+      "step": 1539
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.554659498207886e-05,
+      "loss": 0.0131,
+      "step": 1540
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.552419354838711e-05,
+      "loss": 1.6674,
+      "step": 1541
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 6.550179211469535e-05,
+      "loss": 0.0221,
+      "step": 1542
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.547939068100358e-05,
+      "loss": 2.3879,
+      "step": 1543
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.545698924731182e-05,
+      "loss": 3.2332,
+      "step": 1544
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.543458781362007e-05,
+      "loss": 0.1444,
+      "step": 1545
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.541218637992833e-05,
+      "loss": 1.3524,
+      "step": 1546
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.538978494623657e-05,
+      "loss": 0.219,
+      "step": 1547
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 6.536738351254481e-05,
+      "loss": 0.04,
+      "step": 1548
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.534498207885304e-05,
+      "loss": 0.7835,
+      "step": 1549
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.532258064516129e-05,
+      "loss": 0.3541,
+      "step": 1550
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.530017921146953e-05,
+      "loss": 0.4721,
+      "step": 1551
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.527777777777778e-05,
+      "loss": 2.2721,
+      "step": 1552
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.525537634408603e-05,
+      "loss": 0.0724,
+      "step": 1553
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 6.523297491039428e-05,
+      "loss": 2.5258,
+      "step": 1554
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.521057347670251e-05,
+      "loss": 2.4541,
+      "step": 1555
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.518817204301075e-05,
+      "loss": 1.5164,
+      "step": 1556
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.5165770609319e-05,
+      "loss": 0.0095,
+      "step": 1557
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.514336917562724e-05,
+      "loss": 0.023,
+      "step": 1558
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.51209677419355e-05,
+      "loss": 0.2657,
+      "step": 1559
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.509856630824374e-05,
+      "loss": 0.0359,
+      "step": 1560
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.507616487455197e-05,
+      "loss": 1.0953,
+      "step": 1561
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.505376344086021e-05,
+      "loss": 0.2186,
+      "step": 1562
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.503136200716846e-05,
+      "loss": 0.0128,
+      "step": 1563
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.50089605734767e-05,
+      "loss": 1.6052,
+      "step": 1564
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.498655913978495e-05,
+      "loss": 2.0732,
+      "step": 1565
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.49641577060932e-05,
+      "loss": 0.0222,
+      "step": 1566
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.494175627240143e-05,
+      "loss": 0.0051,
+      "step": 1567
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.491935483870968e-05,
+      "loss": 2.1639,
+      "step": 1568
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.489695340501792e-05,
+      "loss": 1.8936,
+      "step": 1569
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.487455197132617e-05,
+      "loss": 0.1239,
+      "step": 1570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.485215053763441e-05,
+      "loss": 4.1094,
+      "step": 1571
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.482974910394267e-05,
+      "loss": 0.0436,
+      "step": 1572
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.48073476702509e-05,
+      "loss": 2.4928,
+      "step": 1573
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.478494623655914e-05,
+      "loss": 0.8147,
+      "step": 1574
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.476254480286738e-05,
+      "loss": 0.0233,
+      "step": 1575
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.474014336917563e-05,
+      "loss": 0.0047,
+      "step": 1576
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.471774193548387e-05,
+      "loss": 0.0126,
+      "step": 1577
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.469534050179213e-05,
+      "loss": 3.5653,
+      "step": 1578
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.467293906810036e-05,
+      "loss": 0.2278,
+      "step": 1579
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.46505376344086e-05,
+      "loss": 3.3286,
+      "step": 1580
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 6.462813620071685e-05,
+      "loss": 2.8218,
+      "step": 1581
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.460573476702509e-05,
+      "loss": 2.3619,
+      "step": 1582
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.458333333333334e-05,
+      "loss": 3.1395,
+      "step": 1583
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.456093189964158e-05,
+      "loss": 0.0181,
+      "step": 1584
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.453853046594982e-05,
+      "loss": 1.2697,
+      "step": 1585
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.451612903225807e-05,
+      "loss": 0.8174,
+      "step": 1586
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 6.449372759856631e-05,
+      "loss": 0.0589,
+      "step": 1587
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.447132616487455e-05,
+      "loss": 2.529,
+      "step": 1588
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.44489247311828e-05,
+      "loss": 2.5891,
+      "step": 1589
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.442652329749104e-05,
+      "loss": 0.1418,
+      "step": 1590
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.440412186379929e-05,
+      "loss": 3.3861,
+      "step": 1591
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.438172043010753e-05,
+      "loss": 0.6497,
+      "step": 1592
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 6.435931899641577e-05,
+      "loss": 4.9434,
+      "step": 1593
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.433691756272402e-05,
+      "loss": 1.9254,
+      "step": 1594
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.431451612903226e-05,
+      "loss": 1.228,
+      "step": 1595
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.42921146953405e-05,
+      "loss": 0.0109,
+      "step": 1596
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.426971326164875e-05,
+      "loss": 1.238,
+      "step": 1597
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 6.4247311827957e-05,
+      "loss": 0.0153,
+      "step": 1598
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.422491039426524e-05,
+      "loss": 0.0123,
+      "step": 1599
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.420250896057348e-05,
+      "loss": 0.5431,
+      "step": 1600
+    },
+    {
+      "epoch": 2.87,
+      "eval_accuracy": 0.5607143044471741,
+      "eval_loss": 1.3744310140609741,
+      "eval_runtime": 68.7081,
+      "eval_samples_per_second": 4.075,
+      "eval_steps_per_second": 4.075,
+      "step": 1600
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.418010752688173e-05,
+      "loss": 0.3064,
+      "step": 1601
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.415770609318996e-05,
+      "loss": 0.2332,
+      "step": 1602
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.413530465949821e-05,
+      "loss": 0.0474,
+      "step": 1603
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 6.411290322580646e-05,
+      "loss": 1.882,
+      "step": 1604
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.40905017921147e-05,
+      "loss": 2.9502,
+      "step": 1605
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.406810035842294e-05,
+      "loss": 1.1745,
+      "step": 1606
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.404569892473119e-05,
+      "loss": 0.5651,
+      "step": 1607
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.402329749103942e-05,
+      "loss": 0.1967,
+      "step": 1608
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 6.400089605734768e-05,
+      "loss": 0.0185,
+      "step": 1609
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.397849462365592e-05,
+      "loss": 0.5308,
+      "step": 1610
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.395609318996416e-05,
+      "loss": 1.2665,
+      "step": 1611
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.393369175627241e-05,
+      "loss": 0.0435,
+      "step": 1612
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.391129032258065e-05,
+      "loss": 0.4028,
+      "step": 1613
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.388888888888888e-05,
+      "loss": 1.1752,
+      "step": 1614
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.386648745519714e-05,
+      "loss": 0.1156,
+      "step": 1615
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.384408602150538e-05,
+      "loss": 1.367,
+      "step": 1616
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.382168458781363e-05,
+      "loss": 0.0647,
+      "step": 1617
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.379928315412187e-05,
+      "loss": 0.4652,
+      "step": 1618
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.377688172043011e-05,
+      "loss": 1.6769,
+      "step": 1619
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.375448028673834e-05,
+      "loss": 0.264,
+      "step": 1620
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.373207885304659e-05,
+      "loss": 0.4732,
+      "step": 1621
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.370967741935485e-05,
+      "loss": 1.8759,
+      "step": 1622
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.368727598566309e-05,
+      "loss": 4.8799,
+      "step": 1623
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.366487455197133e-05,
+      "loss": 0.144,
+      "step": 1624
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.364247311827958e-05,
+      "loss": 0.3807,
+      "step": 1625
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 6.362007168458781e-05,
+      "loss": 0.0943,
+      "step": 1626
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.359767025089605e-05,
+      "loss": 0.5685,
+      "step": 1627
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.357526881720431e-05,
+      "loss": 0.1398,
+      "step": 1628
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.355286738351255e-05,
+      "loss": 1.6889,
+      "step": 1629
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.35304659498208e-05,
+      "loss": 0.4762,
+      "step": 1630
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.350806451612904e-05,
+      "loss": 2.5532,
+      "step": 1631
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 6.348566308243727e-05,
+      "loss": 1.3274,
+      "step": 1632
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.346326164874552e-05,
+      "loss": 0.0273,
+      "step": 1633
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.344086021505376e-05,
+      "loss": 0.248,
+      "step": 1634
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.341845878136202e-05,
+      "loss": 0.663,
+      "step": 1635
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.339605734767026e-05,
+      "loss": 0.1754,
+      "step": 1636
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.33736559139785e-05,
+      "loss": 0.2556,
+      "step": 1637
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.335125448028673e-05,
+      "loss": 0.0521,
+      "step": 1638
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.332885304659498e-05,
+      "loss": 0.6925,
+      "step": 1639
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.330645161290322e-05,
+      "loss": 1.8528,
+      "step": 1640
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.328405017921148e-05,
+      "loss": 3.0913,
+      "step": 1641
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.326164874551972e-05,
+      "loss": 0.217,
+      "step": 1642
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.323924731182797e-05,
+      "loss": 0.0708,
+      "step": 1643
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.32168458781362e-05,
+      "loss": 0.0217,
+      "step": 1644
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.319444444444444e-05,
+      "loss": 0.9354,
+      "step": 1645
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.317204301075269e-05,
+      "loss": 0.0146,
+      "step": 1646
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.314964157706094e-05,
+      "loss": 4.6191,
+      "step": 1647
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 6.312724014336919e-05,
+      "loss": 0.0103,
+      "step": 1648
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.310483870967743e-05,
+      "loss": 2.9755,
+      "step": 1649
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.308243727598566e-05,
+      "loss": 1.4273,
+      "step": 1650
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.30600358422939e-05,
+      "loss": 0.2218,
+      "step": 1651
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.303763440860215e-05,
+      "loss": 0.0315,
+      "step": 1652
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.301523297491039e-05,
+      "loss": 1.182,
+      "step": 1653
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.299283154121865e-05,
+      "loss": 3.8418,
+      "step": 1654
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.29704301075269e-05,
+      "loss": 0.0075,
+      "step": 1655
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.294802867383512e-05,
+      "loss": 2.7354,
+      "step": 1656
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.292562724014337e-05,
+      "loss": 0.0023,
+      "step": 1657
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.290322580645161e-05,
+      "loss": 2.6918,
+      "step": 1658
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.288082437275986e-05,
+      "loss": 2.0246,
+      "step": 1659
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.285842293906811e-05,
+      "loss": 0.3217,
+      "step": 1660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.283602150537636e-05,
+      "loss": 1.4339,
+      "step": 1661
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.281362007168459e-05,
+      "loss": 2.1166,
+      "step": 1662
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.279121863799283e-05,
+      "loss": 0.0241,
+      "step": 1663
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.276881720430107e-05,
+      "loss": 2.4449,
+      "step": 1664
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.274641577060932e-05,
+      "loss": 0.0581,
+      "step": 1665
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.272401433691756e-05,
+      "loss": 0.6913,
+      "step": 1666
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.270161290322582e-05,
+      "loss": 1.8219,
+      "step": 1667
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.267921146953405e-05,
+      "loss": 0.0197,
+      "step": 1668
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.26568100358423e-05,
+      "loss": 2.2056,
+      "step": 1669
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.263440860215054e-05,
+      "loss": 3.2378,
+      "step": 1670
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.261200716845878e-05,
+      "loss": 1.0054,
+      "step": 1671
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.258960573476703e-05,
+      "loss": 0.8831,
+      "step": 1672
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.256720430107528e-05,
+      "loss": 0.1308,
+      "step": 1673
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.254480286738351e-05,
+      "loss": 1.5538,
+      "step": 1674
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.252240143369176e-05,
+      "loss": 0.5135,
+      "step": 1675
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.25e-05,
+      "loss": 1.0863,
+      "step": 1676
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.247759856630825e-05,
+      "loss": 0.0446,
+      "step": 1677
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.245519713261649e-05,
+      "loss": 1.5275,
+      "step": 1678
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.243279569892473e-05,
+      "loss": 0.0057,
+      "step": 1679
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.241039426523298e-05,
+      "loss": 3.101,
+      "step": 1680
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.238799283154122e-05,
+      "loss": 0.1619,
+      "step": 1681
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 6.236559139784946e-05,
+      "loss": 0.0269,
+      "step": 1682
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.234318996415771e-05,
+      "loss": 0.0166,
+      "step": 1683
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.232078853046595e-05,
+      "loss": 0.217,
+      "step": 1684
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.22983870967742e-05,
+      "loss": 1.6271,
+      "step": 1685
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.227598566308244e-05,
+      "loss": 1.4349,
+      "step": 1686
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 6.225358422939068e-05,
+      "loss": 0.1285,
+      "step": 1687
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.223118279569893e-05,
+      "loss": 0.3932,
+      "step": 1688
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.220878136200717e-05,
+      "loss": 2.9136,
+      "step": 1689
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.218637992831542e-05,
+      "loss": 0.2241,
+      "step": 1690
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.216397849462366e-05,
+      "loss": 0.6071,
+      "step": 1691
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.21415770609319e-05,
+      "loss": 0.0143,
+      "step": 1692
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 6.211917562724015e-05,
+      "loss": 1.9497,
+      "step": 1693
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.209677419354839e-05,
+      "loss": 1.5532,
+      "step": 1694
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.207437275985663e-05,
+      "loss": 1.8786,
+      "step": 1695
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.205197132616488e-05,
+      "loss": 0.0281,
+      "step": 1696
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.202956989247312e-05,
+      "loss": 0.055,
+      "step": 1697
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.200716845878137e-05,
+      "loss": 0.0084,
+      "step": 1698
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 6.198476702508961e-05,
+      "loss": 0.3234,
+      "step": 1699
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.196236559139785e-05,
+      "loss": 0.2232,
+      "step": 1700
+    },
+    {
+      "epoch": 3.05,
+      "eval_accuracy": 0.5642856955528259,
+      "eval_loss": 1.4665553569793701,
+      "eval_runtime": 73.21,
+      "eval_samples_per_second": 3.825,
+      "eval_steps_per_second": 3.825,
+      "step": 1700
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.19399641577061e-05,
+      "loss": 0.5841,
+      "step": 1701
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.191756272401434e-05,
+      "loss": 0.1527,
+      "step": 1702
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.189516129032257e-05,
+      "loss": 0.0203,
+      "step": 1703
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.187275985663083e-05,
+      "loss": 0.0678,
+      "step": 1704
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.185035842293907e-05,
+      "loss": 0.0192,
+      "step": 1705
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.182795698924732e-05,
+      "loss": 0.1156,
+      "step": 1706
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.180555555555556e-05,
+      "loss": 0.364,
+      "step": 1707
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.17831541218638e-05,
+      "loss": 0.0715,
+      "step": 1708
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.176075268817204e-05,
+      "loss": 0.0763,
+      "step": 1709
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.173835125448029e-05,
+      "loss": 0.1876,
+      "step": 1710
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.171594982078854e-05,
+      "loss": 0.1646,
+      "step": 1711
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.169354838709678e-05,
+      "loss": 0.1347,
+      "step": 1712
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.167114695340502e-05,
+      "loss": 0.1174,
+      "step": 1713
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.164874551971327e-05,
+      "loss": 1.0829,
+      "step": 1714
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.16263440860215e-05,
+      "loss": 0.3609,
+      "step": 1715
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.160394265232976e-05,
+      "loss": 0.0205,
+      "step": 1716
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.1581541218638e-05,
+      "loss": 3.7051,
+      "step": 1717
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.155913978494624e-05,
+      "loss": 0.0271,
+      "step": 1718
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.153673835125449e-05,
+      "loss": 0.1408,
+      "step": 1719
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.151433691756273e-05,
+      "loss": 0.062,
+      "step": 1720
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.149193548387096e-05,
+      "loss": 0.4429,
+      "step": 1721
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.14695340501792e-05,
+      "loss": 0.0171,
+      "step": 1722
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.144713261648746e-05,
+      "loss": 0.3741,
+      "step": 1723
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.14247311827957e-05,
+      "loss": 0.3173,
+      "step": 1724
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.140232974910395e-05,
+      "loss": 0.2003,
+      "step": 1725
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.13799283154122e-05,
+      "loss": 3.126,
+      "step": 1726
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.135752688172042e-05,
+      "loss": 0.8724,
+      "step": 1727
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.133512544802867e-05,
+      "loss": 0.3535,
+      "step": 1728
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.131272401433693e-05,
+      "loss": 0.2745,
+      "step": 1729
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.129032258064517e-05,
+      "loss": 0.0156,
+      "step": 1730
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.126792114695341e-05,
+      "loss": 0.0808,
+      "step": 1731
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.124551971326166e-05,
+      "loss": 0.099,
+      "step": 1732
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.122311827956989e-05,
+      "loss": 0.0189,
+      "step": 1733
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.120071684587813e-05,
+      "loss": 0.0044,
+      "step": 1734
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.117831541218638e-05,
+      "loss": 0.0162,
+      "step": 1735
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.115591397849463e-05,
+      "loss": 0.0124,
+      "step": 1736
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.113351254480288e-05,
+      "loss": 0.6126,
+      "step": 1737
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.111111111111112e-05,
+      "loss": 0.076,
+      "step": 1738
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.108870967741935e-05,
+      "loss": 1.4089,
+      "step": 1739
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.10663082437276e-05,
+      "loss": 0.1568,
+      "step": 1740
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.104390681003584e-05,
+      "loss": 0.1305,
+      "step": 1741
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.102150537634409e-05,
+      "loss": 0.0058,
+      "step": 1742
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.099910394265233e-05,
+      "loss": 0.051,
+      "step": 1743
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.0976702508960584e-05,
+      "loss": 1.2674,
+      "step": 1744
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.0954301075268814e-05,
+      "loss": 0.0022,
+      "step": 1745
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.093189964157706e-05,
+      "loss": 0.0164,
+      "step": 1746
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.090949820788531e-05,
+      "loss": 0.6284,
+      "step": 1747
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.088709677419355e-05,
+      "loss": 3.0459,
+      "step": 1748
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.0864695340501797e-05,
+      "loss": 0.6746,
+      "step": 1749
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.084229390681005e-05,
+      "loss": 0.1876,
+      "step": 1750
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.081989247311828e-05,
+      "loss": 0.0028,
+      "step": 1751
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.079749103942652e-05,
+      "loss": 0.1351,
+      "step": 1752
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.077508960573477e-05,
+      "loss": 0.3886,
+      "step": 1753
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.0752688172043016e-05,
+      "loss": 0.0165,
+      "step": 1754
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.073028673835126e-05,
+      "loss": 0.3414,
+      "step": 1755
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.0707885304659504e-05,
+      "loss": 0.0131,
+      "step": 1756
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.068548387096774e-05,
+      "loss": 0.0257,
+      "step": 1757
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.0663082437275985e-05,
+      "loss": 0.0053,
+      "step": 1758
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.0640681003584235e-05,
+      "loss": 0.4794,
+      "step": 1759
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.061827956989248e-05,
+      "loss": 0.126,
+      "step": 1760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.059587813620072e-05,
+      "loss": 1.4702,
+      "step": 1761
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.057347670250897e-05,
+      "loss": 0.0107,
+      "step": 1762
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.0551075268817204e-05,
+      "loss": 1.9502,
+      "step": 1763
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.052867383512545e-05,
+      "loss": 0.0049,
+      "step": 1764
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.050627240143369e-05,
+      "loss": 0.0046,
+      "step": 1765
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.048387096774194e-05,
+      "loss": 0.0097,
+      "step": 1766
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.0461469534050186e-05,
+      "loss": 0.569,
+      "step": 1767
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.043906810035843e-05,
+      "loss": 0.2552,
+      "step": 1768
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.041666666666667e-05,
+      "loss": 0.3359,
+      "step": 1769
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.039426523297491e-05,
+      "loss": 0.0459,
+      "step": 1770
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.0371863799283155e-05,
+      "loss": 0.2504,
+      "step": 1771
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.0349462365591405e-05,
+      "loss": 0.0303,
+      "step": 1772
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.032706093189965e-05,
+      "loss": 0.4551,
+      "step": 1773
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.030465949820788e-05,
+      "loss": 0.0942,
+      "step": 1774
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.028225806451613e-05,
+      "loss": 0.4063,
+      "step": 1775
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.0259856630824374e-05,
+      "loss": 0.0036,
+      "step": 1776
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.023745519713262e-05,
+      "loss": 0.0034,
+      "step": 1777
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.021505376344086e-05,
+      "loss": 0.343,
+      "step": 1778
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.019265232974911e-05,
+      "loss": 0.0037,
+      "step": 1779
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.017025089605734e-05,
+      "loss": 0.0058,
+      "step": 1780
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.014784946236559e-05,
+      "loss": 0.004,
+      "step": 1781
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.012544802867384e-05,
+      "loss": 0.0931,
+      "step": 1782
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.010304659498208e-05,
+      "loss": 1.3133,
+      "step": 1783
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0080645161290325e-05,
+      "loss": 0.0039,
+      "step": 1784
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0058243727598576e-05,
+      "loss": 0.0144,
+      "step": 1785
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0035842293906806e-05,
+      "loss": 0.005,
+      "step": 1786
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0013440860215057e-05,
+      "loss": 1.2444,
+      "step": 1787
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.99910394265233e-05,
+      "loss": 0.0043,
+      "step": 1788
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.9968637992831544e-05,
+      "loss": 0.003,
+      "step": 1789
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.994623655913979e-05,
+      "loss": 0.0237,
+      "step": 1790
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.992383512544804e-05,
+      "loss": 0.2136,
+      "step": 1791
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990143369175627e-05,
+      "loss": 0.0044,
+      "step": 1792
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.987903225806451e-05,
+      "loss": 0.2514,
+      "step": 1793
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9856630824372764e-05,
+      "loss": 0.7873,
+      "step": 1794
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.983422939068101e-05,
+      "loss": 0.0033,
+      "step": 1795
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.981182795698925e-05,
+      "loss": 0.0064,
+      "step": 1796
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9789426523297495e-05,
+      "loss": 0.548,
+      "step": 1797
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.976702508960573e-05,
+      "loss": 1.1223,
+      "step": 1798
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9744623655913976e-05,
+      "loss": 3.5831,
+      "step": 1799
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.972222222222223e-05,
+      "loss": 4.417,
+      "step": 1800
+    },
+    {
+      "epoch": 3.23,
+      "eval_accuracy": 0.5821428298950195,
+      "eval_loss": 2.0204527378082275,
+      "eval_runtime": 73.6514,
+      "eval_samples_per_second": 3.802,
+      "eval_steps_per_second": 3.802,
+      "step": 1800
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.969982078853047e-05,
+      "loss": 1.1039,
+      "step": 1801
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9677419354838715e-05,
+      "loss": 1.8254,
+      "step": 1802
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.965501792114696e-05,
+      "loss": 1.7179,
+      "step": 1803
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.9632616487455195e-05,
+      "loss": 0.5615,
+      "step": 1804
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.961021505376344e-05,
+      "loss": 0.002,
+      "step": 1805
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.958781362007168e-05,
+      "loss": 0.703,
+      "step": 1806
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.9565412186379934e-05,
+      "loss": 0.1628,
+      "step": 1807
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.954301075268818e-05,
+      "loss": 2.0153,
+      "step": 1808
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.952060931899642e-05,
+      "loss": 0.0083,
+      "step": 1809
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.949820788530466e-05,
+      "loss": 0.2768,
+      "step": 1810
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.94758064516129e-05,
+      "loss": 2.7355,
+      "step": 1811
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.9453405017921146e-05,
+      "loss": 0.0032,
+      "step": 1812
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.94310035842294e-05,
+      "loss": 0.0211,
+      "step": 1813
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.940860215053764e-05,
+      "loss": 0.6649,
+      "step": 1814
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.9386200716845885e-05,
+      "loss": 0.0055,
+      "step": 1815
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.936379928315412e-05,
+      "loss": 0.048,
+      "step": 1816
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.9341397849462366e-05,
+      "loss": 0.0299,
+      "step": 1817
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.931899641577061e-05,
+      "loss": 0.0045,
+      "step": 1818
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.929659498207886e-05,
+      "loss": 0.7653,
+      "step": 1819
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.9274193548387104e-05,
+      "loss": 0.1348,
+      "step": 1820
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.925179211469535e-05,
+      "loss": 0.0711,
+      "step": 1821
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.9229390681003585e-05,
+      "loss": 0.0083,
+      "step": 1822
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.920698924731183e-05,
+      "loss": 0.05,
+      "step": 1823
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.918458781362007e-05,
+      "loss": 2.3804,
+      "step": 1824
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.9162186379928317e-05,
+      "loss": 0.0058,
+      "step": 1825
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.913978494623657e-05,
+      "loss": 0.0147,
+      "step": 1826
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.911738351254481e-05,
+      "loss": 3.0894,
+      "step": 1827
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.909498207885305e-05,
+      "loss": 0.0157,
+      "step": 1828
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.907258064516129e-05,
+      "loss": 0.0041,
+      "step": 1829
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.9050179211469536e-05,
+      "loss": 0.3412,
+      "step": 1830
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.902777777777778e-05,
+      "loss": 0.0056,
+      "step": 1831
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.900537634408603e-05,
+      "loss": 0.4272,
+      "step": 1832
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.8982974910394274e-05,
+      "loss": 0.3729,
+      "step": 1833
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.8960573476702505e-05,
+      "loss": 0.0035,
+      "step": 1834
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.8938172043010755e-05,
+      "loss": 0.6801,
+      "step": 1835
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.8915770609319e-05,
+      "loss": 0.0116,
+      "step": 1836
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.889336917562724e-05,
+      "loss": 0.2858,
+      "step": 1837
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.887096774193549e-05,
+      "loss": 0.0351,
+      "step": 1838
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.884856630824374e-05,
+      "loss": 0.0504,
+      "step": 1839
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.882616487455197e-05,
+      "loss": 0.0053,
+      "step": 1840
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.880376344086022e-05,
+      "loss": 0.0181,
+      "step": 1841
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.878136200716846e-05,
+      "loss": 2.1116,
+      "step": 1842
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.8758960573476706e-05,
+      "loss": 0.0072,
+      "step": 1843
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.873655913978495e-05,
+      "loss": 0.0933,
+      "step": 1844
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.87141577060932e-05,
+      "loss": 0.0393,
+      "step": 1845
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.869175627240143e-05,
+      "loss": 0.0767,
+      "step": 1846
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.8669354838709675e-05,
+      "loss": 0.0034,
+      "step": 1847
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.8646953405017925e-05,
+      "loss": 0.0429,
+      "step": 1848
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.862455197132617e-05,
+      "loss": 0.5937,
+      "step": 1849
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.860215053763441e-05,
+      "loss": 2.3998,
+      "step": 1850
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.8579749103942664e-05,
+      "loss": 0.0074,
+      "step": 1851
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.8557347670250894e-05,
+      "loss": 0.0027,
+      "step": 1852
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.853494623655914e-05,
+      "loss": 0.5179,
+      "step": 1853
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.851254480286739e-05,
+      "loss": 0.0058,
+      "step": 1854
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.849014336917563e-05,
+      "loss": 0.0502,
+      "step": 1855
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.8467741935483876e-05,
+      "loss": 0.0982,
+      "step": 1856
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.844534050179212e-05,
+      "loss": 0.0044,
+      "step": 1857
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.842293906810036e-05,
+      "loss": 0.0774,
+      "step": 1858
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.84005376344086e-05,
+      "loss": 1.9723,
+      "step": 1859
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.837813620071685e-05,
+      "loss": 0.002,
+      "step": 1860
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.8355734767025096e-05,
+      "loss": 0.0075,
+      "step": 1861
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.833333333333334e-05,
+      "loss": 0.0358,
+      "step": 1862
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.8310931899641577e-05,
+      "loss": 0.4065,
+      "step": 1863
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.828853046594982e-05,
+      "loss": 0.2798,
+      "step": 1864
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.8266129032258064e-05,
+      "loss": 0.0251,
+      "step": 1865
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.824372759856631e-05,
+      "loss": 0.4577,
+      "step": 1866
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.822132616487456e-05,
+      "loss": 0.0309,
+      "step": 1867
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.81989247311828e-05,
+      "loss": 0.1335,
+      "step": 1868
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.817652329749104e-05,
+      "loss": 0.0191,
+      "step": 1869
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.8154121863799284e-05,
+      "loss": 3.3039,
+      "step": 1870
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.813172043010753e-05,
+      "loss": 0.0042,
+      "step": 1871
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.810931899641577e-05,
+      "loss": 0.0048,
+      "step": 1872
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.808691756272402e-05,
+      "loss": 0.0925,
+      "step": 1873
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.8064516129032266e-05,
+      "loss": 0.2282,
+      "step": 1874
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.8042114695340496e-05,
+      "loss": 0.0246,
+      "step": 1875
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.801971326164875e-05,
+      "loss": 3.9005,
+      "step": 1876
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.799731182795699e-05,
+      "loss": 1.1228,
+      "step": 1877
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.7974910394265235e-05,
+      "loss": 1.5536,
+      "step": 1878
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.795250896057348e-05,
+      "loss": 0.1178,
+      "step": 1879
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.793010752688173e-05,
+      "loss": 0.0091,
+      "step": 1880
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.790770609318996e-05,
+      "loss": 3.1866,
+      "step": 1881
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.788530465949821e-05,
+      "loss": 0.0559,
+      "step": 1882
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.7862903225806454e-05,
+      "loss": 1.7508,
+      "step": 1883
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.78405017921147e-05,
+      "loss": 0.0025,
+      "step": 1884
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.781810035842294e-05,
+      "loss": 1.0338,
+      "step": 1885
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.779569892473119e-05,
+      "loss": 2.8042,
+      "step": 1886
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.777329749103942e-05,
+      "loss": 0.2292,
+      "step": 1887
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.775089605734767e-05,
+      "loss": 0.0087,
+      "step": 1888
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.772849462365592e-05,
+      "loss": 0.0131,
+      "step": 1889
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.770609318996416e-05,
+      "loss": 1.0257,
+      "step": 1890
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.7683691756272405e-05,
+      "loss": 0.0054,
+      "step": 1891
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.7661290322580655e-05,
+      "loss": 0.0255,
+      "step": 1892
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.7638888888888886e-05,
+      "loss": 0.1015,
+      "step": 1893
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.761648745519713e-05,
+      "loss": 0.6171,
+      "step": 1894
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.759408602150538e-05,
+      "loss": 1.0701,
+      "step": 1895
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.7571684587813624e-05,
+      "loss": 1.9763,
+      "step": 1896
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.754928315412187e-05,
+      "loss": 0.039,
+      "step": 1897
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.752688172043011e-05,
+      "loss": 0.4372,
+      "step": 1898
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.750448028673835e-05,
+      "loss": 0.0113,
+      "step": 1899
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.748207885304659e-05,
+      "loss": 1.4207,
+      "step": 1900
+    },
+    {
+      "epoch": 3.41,
+      "eval_accuracy": 0.574999988079071,
+      "eval_loss": 1.7889454364776611,
+      "eval_runtime": 73.8533,
+      "eval_samples_per_second": 3.791,
+      "eval_steps_per_second": 3.791,
+      "step": 1900
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.7459677419354843e-05,
+      "loss": 0.2137,
+      "step": 1901
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.743727598566309e-05,
+      "loss": 0.0881,
+      "step": 1902
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.741487455197133e-05,
+      "loss": 0.9055,
+      "step": 1903
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.7392473118279575e-05,
+      "loss": 0.007,
+      "step": 1904
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.737007168458781e-05,
+      "loss": 0.7366,
+      "step": 1905
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.7347670250896056e-05,
+      "loss": 0.0054,
+      "step": 1906
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.73252688172043e-05,
+      "loss": 0.1997,
+      "step": 1907
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.730286738351255e-05,
+      "loss": 0.0858,
+      "step": 1908
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.7280465949820794e-05,
+      "loss": 0.6951,
+      "step": 1909
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.725806451612904e-05,
+      "loss": 0.0275,
+      "step": 1910
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.7235663082437275e-05,
+      "loss": 0.0021,
+      "step": 1911
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.721326164874552e-05,
+      "loss": 0.0231,
+      "step": 1912
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.719086021505376e-05,
+      "loss": 0.0019,
+      "step": 1913
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.7168458781362014e-05,
+      "loss": 0.833,
+      "step": 1914
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.714605734767026e-05,
+      "loss": 0.6074,
+      "step": 1915
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.71236559139785e-05,
+      "loss": 0.0036,
+      "step": 1916
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.710125448028674e-05,
+      "loss": 0.9974,
+      "step": 1917
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.707885304659498e-05,
+      "loss": 0.2051,
+      "step": 1918
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.7056451612903226e-05,
+      "loss": 1.0717,
+      "step": 1919
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.703405017921148e-05,
+      "loss": 2.1875,
+      "step": 1920
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.701164874551972e-05,
+      "loss": 0.0022,
+      "step": 1921
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.6989247311827965e-05,
+      "loss": 0.7121,
+      "step": 1922
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.69668458781362e-05,
+      "loss": 0.0027,
+      "step": 1923
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.6944444444444445e-05,
+      "loss": 0.0017,
+      "step": 1924
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.692204301075269e-05,
+      "loss": 0.0584,
+      "step": 1925
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.689964157706093e-05,
+      "loss": 0.0033,
+      "step": 1926
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 5.6877240143369184e-05,
+      "loss": 0.0917,
+      "step": 1927
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.685483870967743e-05,
+      "loss": 0.0298,
+      "step": 1928
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.6832437275985665e-05,
+      "loss": 0.0025,
+      "step": 1929
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.681003584229391e-05,
+      "loss": 0.4522,
+      "step": 1930
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.678763440860215e-05,
+      "loss": 0.002,
+      "step": 1931
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.6765232974910396e-05,
+      "loss": 1.6821,
+      "step": 1932
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 5.674283154121865e-05,
+      "loss": 1.2817,
+      "step": 1933
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.672043010752689e-05,
+      "loss": 0.0303,
+      "step": 1934
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.669802867383512e-05,
+      "loss": 0.7527,
+      "step": 1935
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.667562724014337e-05,
+      "loss": 0.0042,
+      "step": 1936
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.6653225806451616e-05,
+      "loss": 0.4025,
+      "step": 1937
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.663082437275986e-05,
+      "loss": 0.0029,
+      "step": 1938
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 5.6608422939068103e-05,
+      "loss": 0.008,
+      "step": 1939
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.6586021505376354e-05,
+      "loss": 1.0343,
+      "step": 1940
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.6563620071684584e-05,
+      "loss": 0.0222,
+      "step": 1941
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.6541218637992835e-05,
+      "loss": 0.0089,
+      "step": 1942
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.651881720430108e-05,
+      "loss": 4.1311,
+      "step": 1943
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 5.649641577060932e-05,
+      "loss": 0.0086,
+      "step": 1944
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.6474014336917567e-05,
+      "loss": 1.4846,
+      "step": 1945
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.645161290322582e-05,
+      "loss": 1.9395,
+      "step": 1946
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.642921146953405e-05,
+      "loss": 0.0129,
+      "step": 1947
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.640681003584229e-05,
+      "loss": 0.6686,
+      "step": 1948
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.638440860215054e-05,
+      "loss": 0.0102,
+      "step": 1949
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 5.6362007168458786e-05,
+      "loss": 0.005,
+      "step": 1950
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.633960573476703e-05,
+      "loss": 0.0208,
+      "step": 1951
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.6317204301075274e-05,
+      "loss": 0.0601,
+      "step": 1952
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.629480286738351e-05,
+      "loss": 0.1367,
+      "step": 1953
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.6272401433691755e-05,
+      "loss": 0.0037,
+      "step": 1954
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 0.0116,
+      "step": 1955
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.622759856630825e-05,
+      "loss": 1.134,
+      "step": 1956
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.620519713261649e-05,
+      "loss": 0.2073,
+      "step": 1957
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.618279569892473e-05,
+      "loss": 0.0024,
+      "step": 1958
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.6160394265232974e-05,
+      "loss": 0.0176,
+      "step": 1959
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.613799283154122e-05,
+      "loss": 0.0072,
+      "step": 1960
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.611559139784947e-05,
+      "loss": 3.3782,
+      "step": 1961
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.609318996415771e-05,
+      "loss": 0.0037,
+      "step": 1962
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.6070788530465956e-05,
+      "loss": 0.8563,
+      "step": 1963
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.604838709677419e-05,
+      "loss": 0.7186,
+      "step": 1964
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.602598566308244e-05,
+      "loss": 3.4678,
+      "step": 1965
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.600358422939068e-05,
+      "loss": 1.5629,
+      "step": 1966
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.5981182795698925e-05,
+      "loss": 0.0028,
+      "step": 1967
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.5958781362007175e-05,
+      "loss": 3.751,
+      "step": 1968
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.593637992831542e-05,
+      "loss": 2.6646,
+      "step": 1969
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.5913978494623656e-05,
+      "loss": 0.0192,
+      "step": 1970
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.58915770609319e-05,
+      "loss": 0.0102,
+      "step": 1971
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.5869175627240144e-05,
+      "loss": 0.5797,
+      "step": 1972
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.584677419354839e-05,
+      "loss": 0.0132,
+      "step": 1973
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.582437275985664e-05,
+      "loss": 0.0022,
+      "step": 1974
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.580197132616488e-05,
+      "loss": 0.0524,
+      "step": 1975
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.577956989247311e-05,
+      "loss": 1.8706,
+      "step": 1976
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.5757168458781363e-05,
+      "loss": 0.9671,
+      "step": 1977
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.573476702508961e-05,
+      "loss": 0.8737,
+      "step": 1978
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.571236559139785e-05,
+      "loss": 2.2064,
+      "step": 1979
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.5689964157706095e-05,
+      "loss": 0.0027,
+      "step": 1980
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.5667562724014346e-05,
+      "loss": 0.0022,
+      "step": 1981
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.5645161290322576e-05,
+      "loss": 0.0115,
+      "step": 1982
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.562275985663083e-05,
+      "loss": 0.1373,
+      "step": 1983
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.560035842293907e-05,
+      "loss": 0.0222,
+      "step": 1984
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.5577956989247314e-05,
+      "loss": 2.3237,
+      "step": 1985
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.555555555555556e-05,
+      "loss": 0.1035,
+      "step": 1986
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.553315412186381e-05,
+      "loss": 0.0017,
+      "step": 1987
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.551075268817204e-05,
+      "loss": 0.0027,
+      "step": 1988
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.548835125448028e-05,
+      "loss": 0.0134,
+      "step": 1989
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.5465949820788534e-05,
+      "loss": 0.0121,
+      "step": 1990
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.544354838709678e-05,
+      "loss": 0.0043,
+      "step": 1991
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.542114695340502e-05,
+      "loss": 0.0097,
+      "step": 1992
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.539874551971327e-05,
+      "loss": 0.0028,
+      "step": 1993
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.53763440860215e-05,
+      "loss": 0.0022,
+      "step": 1994
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.5353942652329746e-05,
+      "loss": 3.1664,
+      "step": 1995
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.5331541218638e-05,
+      "loss": 0.0393,
+      "step": 1996
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.530913978494624e-05,
+      "loss": 1.7134,
+      "step": 1997
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.5286738351254485e-05,
+      "loss": 0.4588,
+      "step": 1998
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.526433691756273e-05,
+      "loss": 0.3539,
+      "step": 1999
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.5241935483870966e-05,
+      "loss": 0.0599,
+      "step": 2000
+    },
+    {
+      "epoch": 3.58,
+      "eval_accuracy": 0.6964285969734192,
+      "eval_loss": 1.279900074005127,
+      "eval_runtime": 73.7055,
+      "eval_samples_per_second": 3.799,
+      "eval_steps_per_second": 3.799,
+      "step": 2000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.521953405017921e-05,
+      "loss": 0.0052,
+      "step": 2001
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.519713261648746e-05,
+      "loss": 0.4093,
+      "step": 2002
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.5174731182795704e-05,
+      "loss": 0.0038,
+      "step": 2003
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.515232974910395e-05,
+      "loss": 0.0142,
+      "step": 2004
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.512992831541219e-05,
+      "loss": 0.0893,
+      "step": 2005
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.510752688172043e-05,
+      "loss": 2.0547,
+      "step": 2006
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.508512544802867e-05,
+      "loss": 0.0226,
+      "step": 2007
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5062724014336916e-05,
+      "loss": 0.2471,
+      "step": 2008
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.504032258064517e-05,
+      "loss": 0.003,
+      "step": 2009
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.501792114695341e-05,
+      "loss": 0.0033,
+      "step": 2010
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.4995519713261655e-05,
+      "loss": 0.0238,
+      "step": 2011
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.497311827956989e-05,
+      "loss": 0.0039,
+      "step": 2012
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4950716845878136e-05,
+      "loss": 0.0024,
+      "step": 2013
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.492831541218638e-05,
+      "loss": 0.0274,
+      "step": 2014
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.490591397849463e-05,
+      "loss": 3.8525,
+      "step": 2015
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.4883512544802874e-05,
+      "loss": 1.2731,
+      "step": 2016
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.486111111111112e-05,
+      "loss": 0.4238,
+      "step": 2017
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.4838709677419355e-05,
+      "loss": 0.0025,
+      "step": 2018
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.48163082437276e-05,
+      "loss": 1.1026,
+      "step": 2019
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.479390681003584e-05,
+      "loss": 0.0079,
+      "step": 2020
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.477150537634409e-05,
+      "loss": 0.0018,
+      "step": 2021
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.474910394265234e-05,
+      "loss": 0.0051,
+      "step": 2022
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.472670250896058e-05,
+      "loss": 0.0044,
+      "step": 2023
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.470430107526882e-05,
+      "loss": 0.0089,
+      "step": 2024
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.468189964157706e-05,
+      "loss": 0.0218,
+      "step": 2025
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4659498207885306e-05,
+      "loss": 0.0026,
+      "step": 2026
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.463709677419355e-05,
+      "loss": 0.0024,
+      "step": 2027
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.46146953405018e-05,
+      "loss": 0.0116,
+      "step": 2028
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4592293906810044e-05,
+      "loss": 0.0019,
+      "step": 2029
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.456989247311828e-05,
+      "loss": 0.1176,
+      "step": 2030
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4547491039426525e-05,
+      "loss": 0.002,
+      "step": 2031
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.452508960573477e-05,
+      "loss": 3.1156,
+      "step": 2032
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.450268817204301e-05,
+      "loss": 0.0046,
+      "step": 2033
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.4480286738351264e-05,
+      "loss": 0.4854,
+      "step": 2034
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.445788530465951e-05,
+      "loss": 0.0057,
+      "step": 2035
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.443548387096774e-05,
+      "loss": 0.003,
+      "step": 2036
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.441308243727599e-05,
+      "loss": 0.0257,
+      "step": 2037
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.439068100358423e-05,
+      "loss": 0.0068,
+      "step": 2038
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.4368279569892476e-05,
+      "loss": 0.0023,
+      "step": 2039
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.434587813620072e-05,
+      "loss": 0.1872,
+      "step": 2040
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.432347670250897e-05,
+      "loss": 0.8416,
+      "step": 2041
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.43010752688172e-05,
+      "loss": 0.1013,
+      "step": 2042
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.427867383512545e-05,
+      "loss": 0.0011,
+      "step": 2043
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.4256272401433696e-05,
+      "loss": 0.0008,
+      "step": 2044
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.423387096774194e-05,
+      "loss": 2.0452,
+      "step": 2045
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.421146953405018e-05,
+      "loss": 2.9467,
+      "step": 2046
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.4189068100358434e-05,
+      "loss": 2.7978,
+      "step": 2047
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.4166666666666664e-05,
+      "loss": 0.0016,
+      "step": 2048
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.414426523297491e-05,
+      "loss": 4.064,
+      "step": 2049
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.412186379928316e-05,
+      "loss": 0.0031,
+      "step": 2050
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.40994623655914e-05,
+      "loss": 0.003,
+      "step": 2051
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.4077060931899646e-05,
+      "loss": 0.007,
+      "step": 2052
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.4054659498207883e-05,
+      "loss": 0.0072,
+      "step": 2053
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.403225806451613e-05,
+      "loss": 0.002,
+      "step": 2054
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.400985663082437e-05,
+      "loss": 2.1421,
+      "step": 2055
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.398745519713262e-05,
+      "loss": 0.0038,
+      "step": 2056
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.3965053763440866e-05,
+      "loss": 0.1945,
+      "step": 2057
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.394265232974911e-05,
+      "loss": 2.1797,
+      "step": 2058
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.392025089605735e-05,
+      "loss": 1.0487,
+      "step": 2059
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.389784946236559e-05,
+      "loss": 0.0039,
+      "step": 2060
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.3875448028673834e-05,
+      "loss": 0.0177,
+      "step": 2061
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.3853046594982085e-05,
+      "loss": 0.0053,
+      "step": 2062
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.383064516129033e-05,
+      "loss": 0.0064,
+      "step": 2063
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.380824372759857e-05,
+      "loss": 0.0046,
+      "step": 2064
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.378584229390681e-05,
+      "loss": 0.0107,
+      "step": 2065
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.3763440860215054e-05,
+      "loss": 0.5287,
+      "step": 2066
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.37410394265233e-05,
+      "loss": 0.5329,
+      "step": 2067
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.371863799283154e-05,
+      "loss": 1.4991,
+      "step": 2068
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.369623655913979e-05,
+      "loss": 0.0025,
+      "step": 2069
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.3673835125448036e-05,
+      "loss": 0.027,
+      "step": 2070
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.365143369175627e-05,
+      "loss": 0.0054,
+      "step": 2071
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.362903225806452e-05,
+      "loss": 0.0035,
+      "step": 2072
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.360663082437276e-05,
+      "loss": 0.0027,
+      "step": 2073
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.3584229390681005e-05,
+      "loss": 0.0035,
+      "step": 2074
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.3561827956989255e-05,
+      "loss": 0.0022,
+      "step": 2075
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.35394265232975e-05,
+      "loss": 0.0026,
+      "step": 2076
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.351702508960573e-05,
+      "loss": 0.4492,
+      "step": 2077
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.349462365591398e-05,
+      "loss": 0.1863,
+      "step": 2078
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.3472222222222224e-05,
+      "loss": 0.0009,
+      "step": 2079
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.344982078853047e-05,
+      "loss": 0.5727,
+      "step": 2080
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.342741935483871e-05,
+      "loss": 0.0452,
+      "step": 2081
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.340501792114696e-05,
+      "loss": 0.0033,
+      "step": 2082
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.338261648745519e-05,
+      "loss": 0.0059,
+      "step": 2083
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.336021505376344e-05,
+      "loss": 0.0018,
+      "step": 2084
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.333781362007169e-05,
+      "loss": 1.9292,
+      "step": 2085
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.331541218637993e-05,
+      "loss": 0.129,
+      "step": 2086
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.3293010752688175e-05,
+      "loss": 1.5875,
+      "step": 2087
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.3270609318996425e-05,
+      "loss": 0.0371,
+      "step": 2088
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.3248207885304656e-05,
+      "loss": 1.7235,
+      "step": 2089
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.32258064516129e-05,
+      "loss": 0.0351,
+      "step": 2090
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.320340501792115e-05,
+      "loss": 0.0015,
+      "step": 2091
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.3181003584229394e-05,
+      "loss": 0.007,
+      "step": 2092
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.315860215053764e-05,
+      "loss": 0.0208,
+      "step": 2093
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.313620071684589e-05,
+      "loss": 1.6594,
+      "step": 2094
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.311379928315412e-05,
+      "loss": 0.0144,
+      "step": 2095
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.309139784946236e-05,
+      "loss": 2.7598,
+      "step": 2096
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.3068996415770613e-05,
+      "loss": 1.4443,
+      "step": 2097
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.304659498207886e-05,
+      "loss": 0.0071,
+      "step": 2098
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.30241935483871e-05,
+      "loss": 0.0061,
+      "step": 2099
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 5.3001792114695345e-05,
+      "loss": 0.0049,
+      "step": 2100
+    },
+    {
+      "epoch": 3.76,
+      "eval_accuracy": 0.6107142567634583,
+      "eval_loss": 1.5202769041061401,
+      "eval_runtime": 69.3659,
+      "eval_samples_per_second": 4.037,
+      "eval_steps_per_second": 4.037,
+      "step": 2100
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.297939068100358e-05,
+      "loss": 0.0373,
+      "step": 2101
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.2956989247311826e-05,
+      "loss": 1.6037,
+      "step": 2102
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.293458781362008e-05,
+      "loss": 0.0983,
+      "step": 2103
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.291218637992832e-05,
+      "loss": 0.002,
+      "step": 2104
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.2889784946236564e-05,
+      "loss": 1.9854,
+      "step": 2105
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 5.286738351254481e-05,
+      "loss": 2.578,
+      "step": 2106
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.2844982078853045e-05,
+      "loss": 1.3607,
+      "step": 2107
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.282258064516129e-05,
+      "loss": 0.9426,
+      "step": 2108
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.280017921146953e-05,
+      "loss": 0.128,
+      "step": 2109
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.2777777777777784e-05,
+      "loss": 0.0125,
+      "step": 2110
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.275537634408603e-05,
+      "loss": 0.0014,
+      "step": 2111
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 5.273297491039427e-05,
+      "loss": 0.0324,
+      "step": 2112
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.271057347670251e-05,
+      "loss": 0.0369,
+      "step": 2113
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.268817204301075e-05,
+      "loss": 0.0047,
+      "step": 2114
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.2665770609318996e-05,
+      "loss": 0.0029,
+      "step": 2115
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.264336917562725e-05,
+      "loss": 0.006,
+      "step": 2116
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 5.262096774193549e-05,
+      "loss": 0.0013,
+      "step": 2117
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.2598566308243735e-05,
+      "loss": 0.0034,
+      "step": 2118
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.257616487455197e-05,
+      "loss": 0.0377,
+      "step": 2119
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.2553763440860216e-05,
+      "loss": 0.0019,
+      "step": 2120
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.253136200716846e-05,
+      "loss": 0.0076,
+      "step": 2121
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.25089605734767e-05,
+      "loss": 0.0016,
+      "step": 2122
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 5.2486559139784954e-05,
+      "loss": 3.0204,
+      "step": 2123
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.24641577060932e-05,
+      "loss": 0.0037,
+      "step": 2124
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.2441756272401435e-05,
+      "loss": 4.1026,
+      "step": 2125
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.241935483870968e-05,
+      "loss": 0.1093,
+      "step": 2126
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.239695340501792e-05,
+      "loss": 0.0179,
+      "step": 2127
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 5.2374551971326166e-05,
+      "loss": 0.0121,
+      "step": 2128
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.235215053763442e-05,
+      "loss": 0.0272,
+      "step": 2129
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.232974910394266e-05,
+      "loss": 0.0042,
+      "step": 2130
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.23073476702509e-05,
+      "loss": 0.4437,
+      "step": 2131
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.228494623655914e-05,
+      "loss": 0.0087,
+      "step": 2132
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.2262544802867386e-05,
+      "loss": 2.8117,
+      "step": 2133
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 5.224014336917563e-05,
+      "loss": 1.8841,
+      "step": 2134
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.221774193548388e-05,
+      "loss": 0.1368,
+      "step": 2135
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.2195340501792124e-05,
+      "loss": 0.0168,
+      "step": 2136
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.2172939068100354e-05,
+      "loss": 0.0855,
+      "step": 2137
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.2150537634408605e-05,
+      "loss": 0.0053,
+      "step": 2138
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 5.212813620071685e-05,
+      "loss": 0.0025,
+      "step": 2139
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.210573476702509e-05,
+      "loss": 1.4301,
+      "step": 2140
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.208333333333334e-05,
+      "loss": 0.0019,
+      "step": 2141
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.2060931899641574e-05,
+      "loss": 0.0018,
+      "step": 2142
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.203853046594982e-05,
+      "loss": 0.5242,
+      "step": 2143
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.201612903225807e-05,
+      "loss": 0.0036,
+      "step": 2144
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 5.199372759856631e-05,
+      "loss": 1.8636,
+      "step": 2145
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.1971326164874556e-05,
+      "loss": 2.6082,
+      "step": 2146
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.19489247311828e-05,
+      "loss": 0.001,
+      "step": 2147
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.192652329749104e-05,
+      "loss": 0.0274,
+      "step": 2148
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.190412186379928e-05,
+      "loss": 0.0016,
+      "step": 2149
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.1881720430107525e-05,
+      "loss": 0.1833,
+      "step": 2150
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 5.1859318996415775e-05,
+      "loss": 0.4737,
+      "step": 2151
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.183691756272402e-05,
+      "loss": 1.4693,
+      "step": 2152
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.181451612903226e-05,
+      "loss": 0.0013,
+      "step": 2153
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.17921146953405e-05,
+      "loss": 0.438,
+      "step": 2154
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.1769713261648744e-05,
+      "loss": 0.002,
+      "step": 2155
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 5.174731182795699e-05,
+      "loss": 2.6678,
+      "step": 2156
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.172491039426524e-05,
+      "loss": 0.0018,
+      "step": 2157
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.170250896057348e-05,
+      "loss": 0.0037,
+      "step": 2158
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.1680107526881726e-05,
+      "loss": 0.0014,
+      "step": 2159
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.165770609318996e-05,
+      "loss": 0.0055,
+      "step": 2160
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.163530465949821e-05,
+      "loss": 3.7777,
+      "step": 2161
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.161290322580645e-05,
+      "loss": 0.0035,
+      "step": 2162
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.1590501792114695e-05,
+      "loss": 0.7733,
+      "step": 2163
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.1568100358422946e-05,
+      "loss": 0.0289,
+      "step": 2164
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.154569892473119e-05,
+      "loss": 0.0019,
+      "step": 2165
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.1523297491039426e-05,
+      "loss": 0.0493,
+      "step": 2166
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.150089605734767e-05,
+      "loss": 1.0727,
+      "step": 2167
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.1478494623655914e-05,
+      "loss": 0.0469,
+      "step": 2168
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.145609318996416e-05,
+      "loss": 0.0754,
+      "step": 2169
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.143369175627241e-05,
+      "loss": 0.0009,
+      "step": 2170
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.141129032258065e-05,
+      "loss": 0.9647,
+      "step": 2171
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.138888888888889e-05,
+      "loss": 0.0007,
+      "step": 2172
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 5.1366487455197134e-05,
+      "loss": 0.0028,
+      "step": 2173
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.134408602150538e-05,
+      "loss": 0.0019,
+      "step": 2174
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.132168458781362e-05,
+      "loss": 0.0085,
+      "step": 2175
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.129928315412187e-05,
+      "loss": 0.0016,
+      "step": 2176
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.1276881720430116e-05,
+      "loss": 0.0739,
+      "step": 2177
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 5.1254480286738346e-05,
+      "loss": 2.0294,
+      "step": 2178
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.12320788530466e-05,
+      "loss": 0.024,
+      "step": 2179
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.120967741935484e-05,
+      "loss": 1.9676,
+      "step": 2180
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.1187275985663084e-05,
+      "loss": 0.002,
+      "step": 2181
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.116487455197133e-05,
+      "loss": 1.8031,
+      "step": 2182
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.114247311827958e-05,
+      "loss": 0.0024,
+      "step": 2183
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 5.112007168458781e-05,
+      "loss": 0.0028,
+      "step": 2184
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.109767025089606e-05,
+      "loss": 0.3564,
+      "step": 2185
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.1075268817204304e-05,
+      "loss": 0.0029,
+      "step": 2186
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.105286738351255e-05,
+      "loss": 0.0008,
+      "step": 2187
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.103046594982079e-05,
+      "loss": 0.5006,
+      "step": 2188
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.100806451612904e-05,
+      "loss": 0.0647,
+      "step": 2189
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 5.098566308243727e-05,
+      "loss": 0.0174,
+      "step": 2190
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.0963261648745516e-05,
+      "loss": 0.0018,
+      "step": 2191
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.094086021505377e-05,
+      "loss": 0.0014,
+      "step": 2192
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.091845878136201e-05,
+      "loss": 5.7266,
+      "step": 2193
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.0896057347670255e-05,
+      "loss": 0.228,
+      "step": 2194
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 5.08736559139785e-05,
+      "loss": 2.8969,
+      "step": 2195
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.0851254480286736e-05,
+      "loss": 0.0012,
+      "step": 2196
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.082885304659498e-05,
+      "loss": 0.2089,
+      "step": 2197
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.080645161290323e-05,
+      "loss": 0.0051,
+      "step": 2198
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.0784050179211474e-05,
+      "loss": 0.0049,
+      "step": 2199
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.076164874551972e-05,
+      "loss": 0.0675,
+      "step": 2200
+    },
+    {
+      "epoch": 3.94,
+      "eval_accuracy": 0.6214285492897034,
+      "eval_loss": 1.4654197692871094,
+      "eval_runtime": 73.8536,
+      "eval_samples_per_second": 3.791,
+      "eval_steps_per_second": 3.791,
+      "step": 2200
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 5.073924731182796e-05,
+      "loss": 1.4824,
+      "step": 2201
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.07168458781362e-05,
+      "loss": 0.0042,
+      "step": 2202
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.069444444444444e-05,
+      "loss": 0.007,
+      "step": 2203
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.067204301075269e-05,
+      "loss": 2.7593,
+      "step": 2204
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.064964157706094e-05,
+      "loss": 0.1449,
+      "step": 2205
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 5.062724014336918e-05,
+      "loss": 0.0193,
+      "step": 2206
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.0604838709677425e-05,
+      "loss": 0.0627,
+      "step": 2207
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.058243727598566e-05,
+      "loss": 0.1046,
+      "step": 2208
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.0560035842293906e-05,
+      "loss": 0.0016,
+      "step": 2209
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.053763440860215e-05,
+      "loss": 0.0024,
+      "step": 2210
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.05152329749104e-05,
+      "loss": 0.0156,
+      "step": 2211
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.0492831541218644e-05,
+      "loss": 0.0032,
+      "step": 2212
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.047043010752689e-05,
+      "loss": 0.0011,
+      "step": 2213
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.0448028673835125e-05,
+      "loss": 1.5951,
+      "step": 2214
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.042562724014337e-05,
+      "loss": 0.2686,
+      "step": 2215
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.040322580645161e-05,
+      "loss": 0.0029,
+      "step": 2216
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.0380824372759863e-05,
+      "loss": 0.0049,
+      "step": 2217
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 5.035842293906811e-05,
+      "loss": 0.6988,
+      "step": 2218
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.033602150537635e-05,
+      "loss": 0.0008,
+      "step": 2219
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.031362007168459e-05,
+      "loss": 0.0042,
+      "step": 2220
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.029121863799283e-05,
+      "loss": 0.4707,
+      "step": 2221
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.0268817204301076e-05,
+      "loss": 0.5142,
+      "step": 2222
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 5.024641577060932e-05,
+      "loss": 0.9772,
+      "step": 2223
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.022401433691757e-05,
+      "loss": 0.0333,
+      "step": 2224
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.0201612903225814e-05,
+      "loss": 0.0015,
+      "step": 2225
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.017921146953405e-05,
+      "loss": 0.0345,
+      "step": 2226
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.0156810035842295e-05,
+      "loss": 0.0132,
+      "step": 2227
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.013440860215054e-05,
+      "loss": 0.0985,
+      "step": 2228
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.011200716845878e-05,
+      "loss": 0.0014,
+      "step": 2229
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.0089605734767034e-05,
+      "loss": 3.2193,
+      "step": 2230
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.006720430107528e-05,
+      "loss": 0.018,
+      "step": 2231
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.004480286738351e-05,
+      "loss": 0.0433,
+      "step": 2232
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5.002240143369176e-05,
+      "loss": 0.0408,
+      "step": 2233
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 5e-05,
+      "loss": 0.0059,
+      "step": 2234
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9977598566308246e-05,
+      "loss": 0.0079,
+      "step": 2235
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.995519713261649e-05,
+      "loss": 0.2787,
+      "step": 2236
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9932795698924734e-05,
+      "loss": 0.0012,
+      "step": 2237
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.991039426523298e-05,
+      "loss": 0.0097,
+      "step": 2238
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.988799283154122e-05,
+      "loss": 0.013,
+      "step": 2239
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 4.9865591397849466e-05,
+      "loss": 0.1691,
+      "step": 2240
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.984318996415771e-05,
+      "loss": 0.0057,
+      "step": 2241
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.982078853046595e-05,
+      "loss": 0.0085,
+      "step": 2242
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.97983870967742e-05,
+      "loss": 0.0012,
+      "step": 2243
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.977598566308244e-05,
+      "loss": 0.1866,
+      "step": 2244
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 4.9753584229390685e-05,
+      "loss": 0.0176,
+      "step": 2245
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.973118279569893e-05,
+      "loss": 0.0104,
+      "step": 2246
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.970878136200717e-05,
+      "loss": 3.9378,
+      "step": 2247
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.968637992831541e-05,
+      "loss": 0.0017,
+      "step": 2248
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.966397849462366e-05,
+      "loss": 0.1372,
+      "step": 2249
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.9641577060931904e-05,
+      "loss": 0.0299,
+      "step": 2250
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 4.961917562724014e-05,
+      "loss": 0.0018,
+      "step": 2251
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.959677419354839e-05,
+      "loss": 0.0009,
+      "step": 2252
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.9574372759856636e-05,
+      "loss": 0.0094,
+      "step": 2253
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.955197132616487e-05,
+      "loss": 0.0009,
+      "step": 2254
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.9529569892473124e-05,
+      "loss": 0.0027,
+      "step": 2255
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.950716845878137e-05,
+      "loss": 0.0113,
+      "step": 2256
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 4.9484767025089604e-05,
+      "loss": 0.0009,
+      "step": 2257
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.9462365591397855e-05,
+      "loss": 0.0079,
+      "step": 2258
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.94399641577061e-05,
+      "loss": 0.0387,
+      "step": 2259
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.9417562724014336e-05,
+      "loss": 0.001,
+      "step": 2260
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.939516129032259e-05,
+      "loss": 0.0016,
+      "step": 2261
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 4.937275985663083e-05,
+      "loss": 0.0015,
+      "step": 2262
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.935035842293907e-05,
+      "loss": 0.0015,
+      "step": 2263
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.932795698924731e-05,
+      "loss": 0.048,
+      "step": 2264
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.930555555555556e-05,
+      "loss": 0.3292,
+      "step": 2265
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.92831541218638e-05,
+      "loss": 0.0048,
+      "step": 2266
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.926075268817204e-05,
+      "loss": 0.001,
+      "step": 2267
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 4.9238351254480294e-05,
+      "loss": 0.5711,
+      "step": 2268
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.921594982078853e-05,
+      "loss": 0.0007,
+      "step": 2269
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9193548387096775e-05,
+      "loss": 0.0007,
+      "step": 2270
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9171146953405025e-05,
+      "loss": 0.0301,
+      "step": 2271
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.914874551971326e-05,
+      "loss": 0.2274,
+      "step": 2272
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 4.9126344086021506e-05,
+      "loss": 0.0023,
+      "step": 2273
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.910394265232976e-05,
+      "loss": 0.0032,
+      "step": 2274
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.9081541218637994e-05,
+      "loss": 0.2341,
+      "step": 2275
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.905913978494624e-05,
+      "loss": 0.0022,
+      "step": 2276
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.903673835125449e-05,
+      "loss": 0.0075,
+      "step": 2277
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.9014336917562726e-05,
+      "loss": 0.0008,
+      "step": 2278
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 4.899193548387097e-05,
+      "loss": 0.0438,
+      "step": 2279
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.896953405017921e-05,
+      "loss": 0.0011,
+      "step": 2280
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.894713261648746e-05,
+      "loss": 1.1302,
+      "step": 2281
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.89247311827957e-05,
+      "loss": 0.0017,
+      "step": 2282
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.8902329749103945e-05,
+      "loss": 0.0069,
+      "step": 2283
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.887992831541219e-05,
+      "loss": 0.0299,
+      "step": 2284
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 4.885752688172043e-05,
+      "loss": 0.9449,
+      "step": 2285
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.8835125448028677e-05,
+      "loss": 0.3538,
+      "step": 2286
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.881272401433692e-05,
+      "loss": 0.0061,
+      "step": 2287
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.8790322580645164e-05,
+      "loss": 0.0111,
+      "step": 2288
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.876792114695341e-05,
+      "loss": 0.7068,
+      "step": 2289
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 4.874551971326165e-05,
+      "loss": 0.0008,
+      "step": 2290
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.8723118279569896e-05,
+      "loss": 0.0326,
+      "step": 2291
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.870071684587813e-05,
+      "loss": 0.3148,
+      "step": 2292
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.8678315412186384e-05,
+      "loss": 0.4512,
+      "step": 2293
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.865591397849463e-05,
+      "loss": 0.003,
+      "step": 2294
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.8633512544802864e-05,
+      "loss": 0.0249,
+      "step": 2295
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.0516,
+      "step": 2296
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.858870967741936e-05,
+      "loss": 0.0802,
+      "step": 2297
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.8566308243727596e-05,
+      "loss": 0.0072,
+      "step": 2298
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.854390681003585e-05,
+      "loss": 0.0047,
+      "step": 2299
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.852150537634409e-05,
+      "loss": 0.01,
+      "step": 2300
+    },
+    {
+      "epoch": 4.12,
+      "eval_accuracy": 0.7107142806053162,
+      "eval_loss": 1.283547282218933,
+      "eval_runtime": 73.3443,
+      "eval_samples_per_second": 3.818,
+      "eval_steps_per_second": 3.818,
+      "step": 2300
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 4.849910394265233e-05,
+      "loss": 0.32,
+      "step": 2301
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.847670250896058e-05,
+      "loss": 0.0108,
+      "step": 2302
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.845430107526882e-05,
+      "loss": 0.0016,
+      "step": 2303
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.843189964157706e-05,
+      "loss": 0.0229,
+      "step": 2304
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.840949820788531e-05,
+      "loss": 0.0047,
+      "step": 2305
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.8387096774193554e-05,
+      "loss": 0.0764,
+      "step": 2306
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 4.836469534050179e-05,
+      "loss": 3.1432,
+      "step": 2307
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.8342293906810035e-05,
+      "loss": 0.0007,
+      "step": 2308
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.8319892473118285e-05,
+      "loss": 0.1069,
+      "step": 2309
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.829749103942652e-05,
+      "loss": 0.0028,
+      "step": 2310
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.8275089605734766e-05,
+      "loss": 0.0017,
+      "step": 2311
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 4.825268817204302e-05,
+      "loss": 0.0136,
+      "step": 2312
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.8230286738351254e-05,
+      "loss": 3.3698,
+      "step": 2313
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.82078853046595e-05,
+      "loss": 0.0309,
+      "step": 2314
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.818548387096775e-05,
+      "loss": 0.0012,
+      "step": 2315
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.8163082437275986e-05,
+      "loss": 0.0023,
+      "step": 2316
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.814068100358423e-05,
+      "loss": 0.0011,
+      "step": 2317
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 4.811827956989248e-05,
+      "loss": 0.0017,
+      "step": 2318
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.809587813620072e-05,
+      "loss": 0.0007,
+      "step": 2319
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.807347670250896e-05,
+      "loss": 0.0024,
+      "step": 2320
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.805107526881721e-05,
+      "loss": 0.0008,
+      "step": 2321
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.802867383512545e-05,
+      "loss": 0.0062,
+      "step": 2322
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.800627240143369e-05,
+      "loss": 0.0041,
+      "step": 2323
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 4.7983870967741937e-05,
+      "loss": 0.0028,
+      "step": 2324
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.796146953405018e-05,
+      "loss": 0.0137,
+      "step": 2325
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.7939068100358424e-05,
+      "loss": 2.0355,
+      "step": 2326
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.791666666666667e-05,
+      "loss": 0.0474,
+      "step": 2327
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.789426523297491e-05,
+      "loss": 2.3273,
+      "step": 2328
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 4.7871863799283156e-05,
+      "loss": 0.0012,
+      "step": 2329
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.78494623655914e-05,
+      "loss": 0.0864,
+      "step": 2330
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.7827060931899644e-05,
+      "loss": 1.5513,
+      "step": 2331
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.780465949820789e-05,
+      "loss": 0.0009,
+      "step": 2332
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.778225806451613e-05,
+      "loss": 0.0012,
+      "step": 2333
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.7759856630824375e-05,
+      "loss": 0.0007,
+      "step": 2334
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 4.773745519713262e-05,
+      "loss": 0.0081,
+      "step": 2335
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.771505376344086e-05,
+      "loss": 0.0029,
+      "step": 2336
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.769265232974911e-05,
+      "loss": 0.0008,
+      "step": 2337
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.767025089605735e-05,
+      "loss": 2.4287,
+      "step": 2338
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.7647849462365594e-05,
+      "loss": 0.0596,
+      "step": 2339
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 4.762544802867384e-05,
+      "loss": 0.0015,
+      "step": 2340
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.760304659498208e-05,
+      "loss": 0.0448,
+      "step": 2341
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.7580645161290326e-05,
+      "loss": 0.001,
+      "step": 2342
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.755824372759857e-05,
+      "loss": 0.0006,
+      "step": 2343
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.7535842293906814e-05,
+      "loss": 0.0885,
+      "step": 2344
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.751344086021506e-05,
+      "loss": 0.001,
+      "step": 2345
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 4.74910394265233e-05,
+      "loss": 0.2117,
+      "step": 2346
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.7468637992831545e-05,
+      "loss": 0.0006,
+      "step": 2347
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.744623655913979e-05,
+      "loss": 0.0015,
+      "step": 2348
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.7423835125448026e-05,
+      "loss": 1.286,
+      "step": 2349
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.740143369175628e-05,
+      "loss": 0.0015,
+      "step": 2350
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 4.737903225806452e-05,
+      "loss": 4.1728,
+      "step": 2351
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.735663082437276e-05,
+      "loss": 0.338,
+      "step": 2352
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.733422939068101e-05,
+      "loss": 2.4046,
+      "step": 2353
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.731182795698925e-05,
+      "loss": 0.0027,
+      "step": 2354
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.728942652329749e-05,
+      "loss": 0.0014,
+      "step": 2355
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.726702508960574e-05,
+      "loss": 0.032,
+      "step": 2356
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 4.7244623655913984e-05,
+      "loss": 0.0015,
+      "step": 2357
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 0.0021,
+      "step": 2358
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.719982078853047e-05,
+      "loss": 0.0571,
+      "step": 2359
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.7177419354838716e-05,
+      "loss": 0.0041,
+      "step": 2360
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.715501792114695e-05,
+      "loss": 0.0236,
+      "step": 2361
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.71326164874552e-05,
+      "loss": 0.0026,
+      "step": 2362
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 4.711021505376345e-05,
+      "loss": 1.0004,
+      "step": 2363
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.7087813620071684e-05,
+      "loss": 0.0009,
+      "step": 2364
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.706541218637993e-05,
+      "loss": 0.001,
+      "step": 2365
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.704301075268818e-05,
+      "loss": 0.0015,
+      "step": 2366
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.7020609318996416e-05,
+      "loss": 0.0151,
+      "step": 2367
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 4.699820788530466e-05,
+      "loss": 0.0009,
+      "step": 2368
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.697580645161291e-05,
+      "loss": 0.0009,
+      "step": 2369
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.695340501792115e-05,
+      "loss": 3.9732,
+      "step": 2370
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.693100358422939e-05,
+      "loss": 0.036,
+      "step": 2371
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.690860215053764e-05,
+      "loss": 0.001,
+      "step": 2372
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.688620071684588e-05,
+      "loss": 0.0265,
+      "step": 2373
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 4.686379928315412e-05,
+      "loss": 0.0013,
+      "step": 2374
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.684139784946237e-05,
+      "loss": 0.0007,
+      "step": 2375
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.681899641577061e-05,
+      "loss": 0.1097,
+      "step": 2376
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.6796594982078854e-05,
+      "loss": 0.0007,
+      "step": 2377
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.67741935483871e-05,
+      "loss": 0.0025,
+      "step": 2378
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 4.675179211469534e-05,
+      "loss": 0.0014,
+      "step": 2379
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.6729390681003586e-05,
+      "loss": 0.0017,
+      "step": 2380
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.670698924731183e-05,
+      "loss": 0.3251,
+      "step": 2381
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.6684587813620074e-05,
+      "loss": 0.0749,
+      "step": 2382
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.666218637992832e-05,
+      "loss": 0.6085,
+      "step": 2383
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.663978494623656e-05,
+      "loss": 2.469,
+      "step": 2384
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 4.6617383512544805e-05,
+      "loss": 0.0014,
+      "step": 2385
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.659498207885305e-05,
+      "loss": 0.0004,
+      "step": 2386
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.657258064516129e-05,
+      "loss": 0.0311,
+      "step": 2387
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.655017921146954e-05,
+      "loss": 0.0153,
+      "step": 2388
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.652777777777778e-05,
+      "loss": 0.0018,
+      "step": 2389
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.650537634408602e-05,
+      "loss": 0.0015,
+      "step": 2390
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 4.648297491039427e-05,
+      "loss": 0.0068,
+      "step": 2391
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.646057347670251e-05,
+      "loss": 0.0012,
+      "step": 2392
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.643817204301075e-05,
+      "loss": 0.0025,
+      "step": 2393
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.6415770609319e-05,
+      "loss": 0.0574,
+      "step": 2394
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.6393369175627244e-05,
+      "loss": 0.0033,
+      "step": 2395
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 4.637096774193548e-05,
+      "loss": 0.0441,
+      "step": 2396
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.634856630824373e-05,
+      "loss": 0.0029,
+      "step": 2397
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.6326164874551976e-05,
+      "loss": 0.0012,
+      "step": 2398
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.630376344086021e-05,
+      "loss": 2.8771,
+      "step": 2399
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.628136200716846e-05,
+      "loss": 3.9507,
+      "step": 2400
+    },
+    {
+      "epoch": 4.3,
+      "eval_accuracy": 0.6357142925262451,
+      "eval_loss": 1.6841779947280884,
+      "eval_runtime": 73.5174,
+      "eval_samples_per_second": 3.809,
+      "eval_steps_per_second": 3.809,
+      "step": 2400
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.625896057347671e-05,
+      "loss": 0.0045,
+      "step": 2401
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 4.6236559139784944e-05,
+      "loss": 1.4038,
+      "step": 2402
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.6214157706093195e-05,
+      "loss": 2.1527,
+      "step": 2403
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.619175627240144e-05,
+      "loss": 0.0017,
+      "step": 2404
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.6169354838709676e-05,
+      "loss": 0.2794,
+      "step": 2405
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.614695340501792e-05,
+      "loss": 1.8836,
+      "step": 2406
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.612455197132617e-05,
+      "loss": 0.0009,
+      "step": 2407
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.610215053763441e-05,
+      "loss": 0.1977,
+      "step": 2408
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.607974910394265e-05,
+      "loss": 0.0031,
+      "step": 2409
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.60573476702509e-05,
+      "loss": 0.0031,
+      "step": 2410
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.603494623655914e-05,
+      "loss": 0.0768,
+      "step": 2411
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.601254480286738e-05,
+      "loss": 0.001,
+      "step": 2412
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.5990143369175634e-05,
+      "loss": 0.0572,
+      "step": 2413
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.596774193548387e-05,
+      "loss": 0.6276,
+      "step": 2414
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.5945340501792115e-05,
+      "loss": 0.0109,
+      "step": 2415
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.5922939068100365e-05,
+      "loss": 0.0018,
+      "step": 2416
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.59005376344086e-05,
+      "loss": 0.009,
+      "step": 2417
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.5878136200716846e-05,
+      "loss": 0.0017,
+      "step": 2418
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.58557347670251e-05,
+      "loss": 0.001,
+      "step": 2419
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.0006,
+      "step": 2420
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.581093189964158e-05,
+      "loss": 0.0005,
+      "step": 2421
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.578853046594982e-05,
+      "loss": 0.0013,
+      "step": 2422
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.5766129032258065e-05,
+      "loss": 0.0009,
+      "step": 2423
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.574372759856631e-05,
+      "loss": 0.0009,
+      "step": 2424
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.572132616487455e-05,
+      "loss": 0.0201,
+      "step": 2425
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.56989247311828e-05,
+      "loss": 0.0073,
+      "step": 2426
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.567652329749104e-05,
+      "loss": 0.0052,
+      "step": 2427
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.5654121863799285e-05,
+      "loss": 1.2477,
+      "step": 2428
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.563172043010753e-05,
+      "loss": 0.2262,
+      "step": 2429
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.560931899641577e-05,
+      "loss": 0.0008,
+      "step": 2430
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.5586917562724016e-05,
+      "loss": 0.0012,
+      "step": 2431
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.556451612903226e-05,
+      "loss": 0.1875,
+      "step": 2432
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.5542114695340504e-05,
+      "loss": 0.4245,
+      "step": 2433
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.551971326164875e-05,
+      "loss": 0.0008,
+      "step": 2434
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.549731182795699e-05,
+      "loss": 0.7461,
+      "step": 2435
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.5474910394265236e-05,
+      "loss": 1.7211,
+      "step": 2436
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.545250896057348e-05,
+      "loss": 0.0011,
+      "step": 2437
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.543010752688172e-05,
+      "loss": 0.0007,
+      "step": 2438
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.540770609318997e-05,
+      "loss": 0.3995,
+      "step": 2439
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.538530465949821e-05,
+      "loss": 0.004,
+      "step": 2440
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.5362903225806455e-05,
+      "loss": 0.0662,
+      "step": 2441
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.53405017921147e-05,
+      "loss": 0.0013,
+      "step": 2442
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.531810035842294e-05,
+      "loss": 0.0414,
+      "step": 2443
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.5295698924731187e-05,
+      "loss": 0.0658,
+      "step": 2444
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.527329749103943e-05,
+      "loss": 0.0017,
+      "step": 2445
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.5250896057347674e-05,
+      "loss": 0.2541,
+      "step": 2446
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.522849462365592e-05,
+      "loss": 0.0041,
+      "step": 2447
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.520609318996416e-05,
+      "loss": 0.0055,
+      "step": 2448
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.5183691756272406e-05,
+      "loss": 0.001,
+      "step": 2449
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.516129032258064e-05,
+      "loss": 0.0008,
+      "step": 2450
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.5138888888888894e-05,
+      "loss": 0.0036,
+      "step": 2451
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.511648745519714e-05,
+      "loss": 0.0031,
+      "step": 2452
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.5094086021505375e-05,
+      "loss": 0.0007,
+      "step": 2453
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.5071684587813625e-05,
+      "loss": 0.0081,
+      "step": 2454
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.504928315412187e-05,
+      "loss": 0.0011,
+      "step": 2455
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.5026881720430106e-05,
+      "loss": 0.0018,
+      "step": 2456
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.500448028673836e-05,
+      "loss": 0.0084,
+      "step": 2457
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.49820788530466e-05,
+      "loss": 0.001,
+      "step": 2458
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.495967741935484e-05,
+      "loss": 0.0258,
+      "step": 2459
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.493727598566309e-05,
+      "loss": 3.4167,
+      "step": 2460
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.491487455197133e-05,
+      "loss": 0.0014,
+      "step": 2461
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.489247311827957e-05,
+      "loss": 0.0096,
+      "step": 2462
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.487007168458782e-05,
+      "loss": 0.0008,
+      "step": 2463
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.4847670250896064e-05,
+      "loss": 0.0007,
+      "step": 2464
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.48252688172043e-05,
+      "loss": 1.7141,
+      "step": 2465
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.4802867383512545e-05,
+      "loss": 0.412,
+      "step": 2466
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.478046594982079e-05,
+      "loss": 0.0011,
+      "step": 2467
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.475806451612903e-05,
+      "loss": 0.0007,
+      "step": 2468
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.4735663082437276e-05,
+      "loss": 0.0268,
+      "step": 2469
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.471326164874552e-05,
+      "loss": 0.0033,
+      "step": 2470
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.4690860215053764e-05,
+      "loss": 0.0237,
+      "step": 2471
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.466845878136201e-05,
+      "loss": 0.0025,
+      "step": 2472
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.464605734767025e-05,
+      "loss": 1.9975,
+      "step": 2473
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.4623655913978496e-05,
+      "loss": 0.006,
+      "step": 2474
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.460125448028674e-05,
+      "loss": 0.0021,
+      "step": 2475
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.4578853046594983e-05,
+      "loss": 0.0008,
+      "step": 2476
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.455645161290323e-05,
+      "loss": 0.0019,
+      "step": 2477
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.453405017921147e-05,
+      "loss": 0.0015,
+      "step": 2478
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.4511648745519715e-05,
+      "loss": 0.0009,
+      "step": 2479
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.448924731182796e-05,
+      "loss": 0.0006,
+      "step": 2480
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.44668458781362e-05,
+      "loss": 0.001,
+      "step": 2481
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.0095,
+      "step": 2482
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.442204301075269e-05,
+      "loss": 0.0227,
+      "step": 2483
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.4399641577060934e-05,
+      "loss": 0.1191,
+      "step": 2484
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.437724014336918e-05,
+      "loss": 0.0021,
+      "step": 2485
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.435483870967742e-05,
+      "loss": 0.0166,
+      "step": 2486
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.4332437275985666e-05,
+      "loss": 1.3538,
+      "step": 2487
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.431003584229391e-05,
+      "loss": 0.0014,
+      "step": 2488
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.4287634408602154e-05,
+      "loss": 0.0044,
+      "step": 2489
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.42652329749104e-05,
+      "loss": 0.005,
+      "step": 2490
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.4242831541218635e-05,
+      "loss": 0.001,
+      "step": 2491
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.4220430107526885e-05,
+      "loss": 0.0025,
+      "step": 2492
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.419802867383513e-05,
+      "loss": 0.0665,
+      "step": 2493
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.4175627240143366e-05,
+      "loss": 0.0008,
+      "step": 2494
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.415322580645162e-05,
+      "loss": 0.0008,
+      "step": 2495
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.413082437275986e-05,
+      "loss": 4.1609,
+      "step": 2496
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.41084229390681e-05,
+      "loss": 0.0396,
+      "step": 2497
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.408602150537635e-05,
+      "loss": 1.9105,
+      "step": 2498
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.406362007168459e-05,
+      "loss": 0.0037,
+      "step": 2499
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.404121863799283e-05,
+      "loss": 0.0018,
+      "step": 2500
+    },
+    {
+      "epoch": 4.48,
+      "eval_accuracy": 0.675000011920929,
+      "eval_loss": 1.5358456373214722,
+      "eval_runtime": 74.0048,
+      "eval_samples_per_second": 3.784,
+      "eval_steps_per_second": 3.784,
+      "step": 2500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.401881720430108e-05,
+      "loss": 1.7893,
+      "step": 2501
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.3996415770609324e-05,
+      "loss": 0.0221,
+      "step": 2502
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.397401433691756e-05,
+      "loss": 0.001,
+      "step": 2503
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.395161290322581e-05,
+      "loss": 1.3303,
+      "step": 2504
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.3929211469534055e-05,
+      "loss": 0.0016,
+      "step": 2505
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.390681003584229e-05,
+      "loss": 0.0008,
+      "step": 2506
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.3884408602150536e-05,
+      "loss": 0.0084,
+      "step": 2507
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.386200716845879e-05,
+      "loss": 2.9384,
+      "step": 2508
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.3839605734767024e-05,
+      "loss": 0.0027,
+      "step": 2509
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.381720430107527e-05,
+      "loss": 0.007,
+      "step": 2510
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.379480286738352e-05,
+      "loss": 0.0016,
+      "step": 2511
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.3772401433691756e-05,
+      "loss": 0.0847,
+      "step": 2512
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.375e-05,
+      "loss": 0.0009,
+      "step": 2513
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.372759856630825e-05,
+      "loss": 0.0056,
+      "step": 2514
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.370519713261649e-05,
+      "loss": 0.0005,
+      "step": 2515
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.368279569892473e-05,
+      "loss": 0.0008,
+      "step": 2516
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.366039426523298e-05,
+      "loss": 0.0012,
+      "step": 2517
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.363799283154122e-05,
+      "loss": 0.5674,
+      "step": 2518
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.361559139784946e-05,
+      "loss": 0.0016,
+      "step": 2519
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.359318996415771e-05,
+      "loss": 0.122,
+      "step": 2520
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.357078853046595e-05,
+      "loss": 0.0115,
+      "step": 2521
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.3548387096774194e-05,
+      "loss": 0.0012,
+      "step": 2522
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.352598566308244e-05,
+      "loss": 0.0015,
+      "step": 2523
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.350358422939068e-05,
+      "loss": 0.0009,
+      "step": 2524
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.3481182795698926e-05,
+      "loss": 0.0162,
+      "step": 2525
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.345878136200717e-05,
+      "loss": 0.0006,
+      "step": 2526
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.3436379928315414e-05,
+      "loss": 0.0011,
+      "step": 2527
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.341397849462366e-05,
+      "loss": 0.0108,
+      "step": 2528
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.33915770609319e-05,
+      "loss": 0.0055,
+      "step": 2529
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.3369175627240145e-05,
+      "loss": 0.0004,
+      "step": 2530
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.334677419354839e-05,
+      "loss": 0.0033,
+      "step": 2531
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.332437275985663e-05,
+      "loss": 1.866,
+      "step": 2532
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.330197132616488e-05,
+      "loss": 0.0007,
+      "step": 2533
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.327956989247312e-05,
+      "loss": 0.0008,
+      "step": 2534
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3257168458781365e-05,
+      "loss": 0.0033,
+      "step": 2535
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.323476702508961e-05,
+      "loss": 0.0031,
+      "step": 2536
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.321236559139785e-05,
+      "loss": 0.0167,
+      "step": 2537
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3189964157706096e-05,
+      "loss": 0.5907,
+      "step": 2538
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.316756272401434e-05,
+      "loss": 0.1156,
+      "step": 2539
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.3145161290322584e-05,
+      "loss": 0.0014,
+      "step": 2540
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.312275985663083e-05,
+      "loss": 0.0008,
+      "step": 2541
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.310035842293907e-05,
+      "loss": 0.0017,
+      "step": 2542
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.3077956989247315e-05,
+      "loss": 0.0036,
+      "step": 2543
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 0.0005,
+      "step": 2544
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.30331541218638e-05,
+      "loss": 1.679,
+      "step": 2545
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.301075268817205e-05,
+      "loss": 3.0315,
+      "step": 2546
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.298835125448029e-05,
+      "loss": 0.0011,
+      "step": 2547
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.296594982078853e-05,
+      "loss": 0.0015,
+      "step": 2548
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.294354838709678e-05,
+      "loss": 0.0006,
+      "step": 2549
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.292114695340502e-05,
+      "loss": 0.3655,
+      "step": 2550
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.289874551971326e-05,
+      "loss": 0.0004,
+      "step": 2551
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.287634408602151e-05,
+      "loss": 0.0007,
+      "step": 2552
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2853942652329754e-05,
+      "loss": 0.0026,
+      "step": 2553
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.283154121863799e-05,
+      "loss": 0.0754,
+      "step": 2554
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.280913978494624e-05,
+      "loss": 0.0007,
+      "step": 2555
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2786738351254486e-05,
+      "loss": 3.6303,
+      "step": 2556
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.276433691756272e-05,
+      "loss": 0.0084,
+      "step": 2557
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.2741935483870973e-05,
+      "loss": 0.0006,
+      "step": 2558
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.271953405017922e-05,
+      "loss": 0.0016,
+      "step": 2559
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2697132616487454e-05,
+      "loss": 0.0695,
+      "step": 2560
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2674731182795705e-05,
+      "loss": 0.0016,
+      "step": 2561
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.265232974910394e-05,
+      "loss": 0.0096,
+      "step": 2562
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.2629928315412186e-05,
+      "loss": 0.8969,
+      "step": 2563
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 4.260752688172043e-05,
+      "loss": 0.0006,
+      "step": 2564
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2585125448028674e-05,
+      "loss": 0.0058,
+      "step": 2565
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.256272401433692e-05,
+      "loss": 0.2016,
+      "step": 2566
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.254032258064516e-05,
+      "loss": 0.007,
+      "step": 2567
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.2517921146953405e-05,
+      "loss": 0.0006,
+      "step": 2568
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.249551971326165e-05,
+      "loss": 0.0007,
+      "step": 2569
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.247311827956989e-05,
+      "loss": 0.0014,
+      "step": 2570
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.245071684587814e-05,
+      "loss": 0.001,
+      "step": 2571
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.242831541218638e-05,
+      "loss": 0.0037,
+      "step": 2572
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.2405913978494625e-05,
+      "loss": 0.0013,
+      "step": 2573
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.238351254480287e-05,
+      "loss": 0.0007,
+      "step": 2574
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 4.236111111111111e-05,
+      "loss": 0.002,
+      "step": 2575
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2338709677419356e-05,
+      "loss": 0.3764,
+      "step": 2576
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.23163082437276e-05,
+      "loss": 0.0243,
+      "step": 2577
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.2293906810035844e-05,
+      "loss": 0.0135,
+      "step": 2578
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.227150537634409e-05,
+      "loss": 0.0006,
+      "step": 2579
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.224910394265233e-05,
+      "loss": 0.0181,
+      "step": 2580
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2226702508960575e-05,
+      "loss": 0.0007,
+      "step": 2581
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.220430107526882e-05,
+      "loss": 0.0005,
+      "step": 2582
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.218189964157706e-05,
+      "loss": 0.0006,
+      "step": 2583
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.215949820788531e-05,
+      "loss": 0.0008,
+      "step": 2584
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.213709677419355e-05,
+      "loss": 0.0009,
+      "step": 2585
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.2114695340501795e-05,
+      "loss": 0.4914,
+      "step": 2586
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.209229390681004e-05,
+      "loss": 0.0004,
+      "step": 2587
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.206989247311828e-05,
+      "loss": 2.9995,
+      "step": 2588
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2047491039426526e-05,
+      "loss": 0.0023,
+      "step": 2589
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.202508960573477e-05,
+      "loss": 0.0007,
+      "step": 2590
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 4.2002688172043014e-05,
+      "loss": 0.0005,
+      "step": 2591
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.198028673835125e-05,
+      "loss": 0.0048,
+      "step": 2592
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.19578853046595e-05,
+      "loss": 0.0006,
+      "step": 2593
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.1935483870967746e-05,
+      "loss": 0.7339,
+      "step": 2594
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.191308243727598e-05,
+      "loss": 0.0009,
+      "step": 2595
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.1890681003584233e-05,
+      "loss": 0.0006,
+      "step": 2596
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.186827956989248e-05,
+      "loss": 0.1167,
+      "step": 2597
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.1845878136200714e-05,
+      "loss": 0.0006,
+      "step": 2598
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.1823476702508965e-05,
+      "loss": 0.0023,
+      "step": 2599
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.180107526881721e-05,
+      "loss": 0.0626,
+      "step": 2600
+    },
+    {
+      "epoch": 4.66,
+      "eval_accuracy": 0.6857143044471741,
+      "eval_loss": 1.4614109992980957,
+      "eval_runtime": 69.881,
+      "eval_samples_per_second": 4.007,
+      "eval_steps_per_second": 4.007,
+      "step": 2600
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.1778673835125446e-05,
+      "loss": 0.0008,
+      "step": 2601
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.1756272401433697e-05,
+      "loss": 0.0006,
+      "step": 2602
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.173387096774194e-05,
+      "loss": 0.0064,
+      "step": 2603
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.171146953405018e-05,
+      "loss": 0.0035,
+      "step": 2604
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.168906810035843e-05,
+      "loss": 0.0013,
+      "step": 2605
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.0004,
+      "step": 2606
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.164426523297491e-05,
+      "loss": 0.0005,
+      "step": 2607
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 4.162186379928315e-05,
+      "loss": 0.0016,
+      "step": 2608
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.1599462365591404e-05,
+      "loss": 0.0013,
+      "step": 2609
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.157706093189964e-05,
+      "loss": 0.0007,
+      "step": 2610
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.1554659498207885e-05,
+      "loss": 0.0022,
+      "step": 2611
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.1532258064516135e-05,
+      "loss": 0.0039,
+      "step": 2612
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.150985663082437e-05,
+      "loss": 0.0008,
+      "step": 2613
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.1487455197132616e-05,
+      "loss": 0.0008,
+      "step": 2614
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.146505376344087e-05,
+      "loss": 0.0006,
+      "step": 2615
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.1442652329749104e-05,
+      "loss": 0.1961,
+      "step": 2616
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.142025089605735e-05,
+      "loss": 0.0013,
+      "step": 2617
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.13978494623656e-05,
+      "loss": 0.0011,
+      "step": 2618
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 4.1375448028673835e-05,
+      "loss": 0.0005,
+      "step": 2619
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.135304659498208e-05,
+      "loss": 0.0225,
+      "step": 2620
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.133064516129033e-05,
+      "loss": 0.0004,
+      "step": 2621
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.130824372759857e-05,
+      "loss": 0.3503,
+      "step": 2622
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.128584229390681e-05,
+      "loss": 0.0009,
+      "step": 2623
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.1263440860215055e-05,
+      "loss": 0.001,
+      "step": 2624
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.12410394265233e-05,
+      "loss": 3.6282,
+      "step": 2625
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.121863799283154e-05,
+      "loss": 0.001,
+      "step": 2626
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.1196236559139786e-05,
+      "loss": 0.081,
+      "step": 2627
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.117383512544803e-05,
+      "loss": 0.0008,
+      "step": 2628
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.1151433691756274e-05,
+      "loss": 1.6556,
+      "step": 2629
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.112903225806452e-05,
+      "loss": 1.3859,
+      "step": 2630
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.110663082437276e-05,
+      "loss": 0.0026,
+      "step": 2631
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1084229390681006e-05,
+      "loss": 0.0004,
+      "step": 2632
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.106182795698925e-05,
+      "loss": 0.0005,
+      "step": 2633
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.1039426523297493e-05,
+      "loss": 0.4907,
+      "step": 2634
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.101702508960574e-05,
+      "loss": 1.8229,
+      "step": 2635
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 4.099462365591398e-05,
+      "loss": 0.0019,
+      "step": 2636
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.0972222222222225e-05,
+      "loss": 0.1208,
+      "step": 2637
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.094982078853047e-05,
+      "loss": 0.0009,
+      "step": 2638
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.092741935483871e-05,
+      "loss": 0.1674,
+      "step": 2639
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.090501792114696e-05,
+      "loss": 0.0007,
+      "step": 2640
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.08826164874552e-05,
+      "loss": 0.8555,
+      "step": 2641
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.0860215053763444e-05,
+      "loss": 0.1349,
+      "step": 2642
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.083781362007169e-05,
+      "loss": 0.0009,
+      "step": 2643
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.081541218637993e-05,
+      "loss": 0.0083,
+      "step": 2644
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.0793010752688176e-05,
+      "loss": 0.1108,
+      "step": 2645
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.077060931899642e-05,
+      "loss": 0.0004,
+      "step": 2646
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.0748207885304664e-05,
+      "loss": 5.1332,
+      "step": 2647
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.072580645161291e-05,
+      "loss": 0.0119,
+      "step": 2648
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.0703405017921145e-05,
+      "loss": 0.0101,
+      "step": 2649
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.0681003584229395e-05,
+      "loss": 0.0094,
+      "step": 2650
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.065860215053764e-05,
+      "loss": 0.001,
+      "step": 2651
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.0636200716845876e-05,
+      "loss": 0.0005,
+      "step": 2652
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 4.061379928315413e-05,
+      "loss": 1.8081,
+      "step": 2653
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.0591397849462364e-05,
+      "loss": 0.0068,
+      "step": 2654
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.056899641577061e-05,
+      "loss": 0.0015,
+      "step": 2655
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.054659498207886e-05,
+      "loss": 0.0015,
+      "step": 2656
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.0524193548387096e-05,
+      "loss": 0.0009,
+      "step": 2657
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.050179211469534e-05,
+      "loss": 0.0007,
+      "step": 2658
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.047939068100359e-05,
+      "loss": 0.0009,
+      "step": 2659
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.045698924731183e-05,
+      "loss": 0.7424,
+      "step": 2660
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.043458781362007e-05,
+      "loss": 0.0015,
+      "step": 2661
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.041218637992832e-05,
+      "loss": 0.0008,
+      "step": 2662
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.038978494623656e-05,
+      "loss": 0.0661,
+      "step": 2663
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 4.03673835125448e-05,
+      "loss": 0.0173,
+      "step": 2664
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.0344982078853046e-05,
+      "loss": 0.0151,
+      "step": 2665
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.032258064516129e-05,
+      "loss": 0.0034,
+      "step": 2666
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.0300179211469534e-05,
+      "loss": 0.001,
+      "step": 2667
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.0005,
+      "step": 2668
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.025537634408602e-05,
+      "loss": 0.0076,
+      "step": 2669
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.0232974910394266e-05,
+      "loss": 0.0005,
+      "step": 2670
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.021057347670251e-05,
+      "loss": 0.0013,
+      "step": 2671
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.0188172043010753e-05,
+      "loss": 0.0008,
+      "step": 2672
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.0165770609319e-05,
+      "loss": 0.0006,
+      "step": 2673
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.014336917562724e-05,
+      "loss": 0.0008,
+      "step": 2674
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.0120967741935485e-05,
+      "loss": 0.0008,
+      "step": 2675
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.009856630824373e-05,
+      "loss": 0.0009,
+      "step": 2676
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.007616487455197e-05,
+      "loss": 0.0007,
+      "step": 2677
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.005376344086022e-05,
+      "loss": 0.0009,
+      "step": 2678
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.003136200716846e-05,
+      "loss": 0.0014,
+      "step": 2679
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.0008960573476704e-05,
+      "loss": 0.0031,
+      "step": 2680
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 3.998655913978495e-05,
+      "loss": 0.0008,
+      "step": 2681
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.996415770609319e-05,
+      "loss": 0.0094,
+      "step": 2682
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.9941756272401436e-05,
+      "loss": 0.3319,
+      "step": 2683
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.991935483870968e-05,
+      "loss": 0.0005,
+      "step": 2684
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.9896953405017924e-05,
+      "loss": 0.0216,
+      "step": 2685
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.987455197132617e-05,
+      "loss": 0.0005,
+      "step": 2686
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.985215053763441e-05,
+      "loss": 0.0011,
+      "step": 2687
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.985215053763441e-05,
+      "loss": 1.6067,
+      "step": 2688
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.9829749103942655e-05,
+      "loss": 0.0006,
+      "step": 2689
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.98073476702509e-05,
+      "loss": 0.0055,
+      "step": 2690
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.978494623655914e-05,
+      "loss": 0.012,
+      "step": 2691
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.976254480286739e-05,
+      "loss": 0.0007,
+      "step": 2692
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.974014336917563e-05,
+      "loss": 0.0006,
+      "step": 2693
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.971774193548387e-05,
+      "loss": 0.0009,
+      "step": 2694
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.969534050179212e-05,
+      "loss": 0.0006,
+      "step": 2695
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.967293906810036e-05,
+      "loss": 0.0008,
+      "step": 2696
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.96505376344086e-05,
+      "loss": 0.0006,
+      "step": 2697
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.962813620071685e-05,
+      "loss": 0.0005,
+      "step": 2698
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.9605734767025094e-05,
+      "loss": 0.0006,
+      "step": 2699
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.958333333333333e-05,
+      "loss": 0.0008,
+      "step": 2700
+    },
+    {
+      "epoch": 4.84,
+      "eval_accuracy": 0.6928571462631226,
+      "eval_loss": 1.6478432416915894,
+      "eval_runtime": 73.5728,
+      "eval_samples_per_second": 3.806,
+      "eval_steps_per_second": 3.806,
+      "step": 2700
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.956093189964158e-05,
+      "loss": 0.001,
+      "step": 2701
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.9538530465949825e-05,
+      "loss": 0.001,
+      "step": 2702
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.951612903225806e-05,
+      "loss": 0.0005,
+      "step": 2703
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.949372759856631e-05,
+      "loss": 0.0005,
+      "step": 2704
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.947132616487456e-05,
+      "loss": 0.0005,
+      "step": 2705
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.9448924731182794e-05,
+      "loss": 0.0008,
+      "step": 2706
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.9426523297491045e-05,
+      "loss": 0.0004,
+      "step": 2707
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.940412186379929e-05,
+      "loss": 0.0009,
+      "step": 2708
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 3.9381720430107526e-05,
+      "loss": 0.0004,
+      "step": 2709
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.935931899641577e-05,
+      "loss": 0.0013,
+      "step": 2710
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.933691756272402e-05,
+      "loss": 0.9943,
+      "step": 2711
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.931451612903226e-05,
+      "loss": 0.0007,
+      "step": 2712
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.92921146953405e-05,
+      "loss": 0.0004,
+      "step": 2713
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.926971326164875e-05,
+      "loss": 0.0006,
+      "step": 2714
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.924731182795699e-05,
+      "loss": 2.9109,
+      "step": 2715
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.922491039426523e-05,
+      "loss": 0.0075,
+      "step": 2716
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.9202508960573483e-05,
+      "loss": 0.0008,
+      "step": 2717
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.918010752688172e-05,
+      "loss": 0.0006,
+      "step": 2718
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.9157706093189964e-05,
+      "loss": 0.0006,
+      "step": 2719
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.9135304659498215e-05,
+      "loss": 0.0899,
+      "step": 2720
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.911290322580645e-05,
+      "loss": 0.8874,
+      "step": 2721
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.9090501792114696e-05,
+      "loss": 0.0012,
+      "step": 2722
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.906810035842295e-05,
+      "loss": 0.0005,
+      "step": 2723
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.9045698924731184e-05,
+      "loss": 0.0214,
+      "step": 2724
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.902329749103943e-05,
+      "loss": 0.8248,
+      "step": 2725
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.900089605734767e-05,
+      "loss": 0.0005,
+      "step": 2726
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.8978494623655915e-05,
+      "loss": 0.0013,
+      "step": 2727
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.895609318996416e-05,
+      "loss": 0.0004,
+      "step": 2728
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.89336917562724e-05,
+      "loss": 1.7834,
+      "step": 2729
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.891129032258065e-05,
+      "loss": 0.0004,
+      "step": 2730
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.0006,
+      "step": 2731
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.8866487455197135e-05,
+      "loss": 0.0003,
+      "step": 2732
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.884408602150538e-05,
+      "loss": 0.001,
+      "step": 2733
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.882168458781362e-05,
+      "loss": 0.0005,
+      "step": 2734
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.8799283154121866e-05,
+      "loss": 0.0009,
+      "step": 2735
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.877688172043011e-05,
+      "loss": 0.0105,
+      "step": 2736
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.8754480286738354e-05,
+      "loss": 0.0009,
+      "step": 2737
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.87320788530466e-05,
+      "loss": 0.0006,
+      "step": 2738
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.870967741935484e-05,
+      "loss": 0.1754,
+      "step": 2739
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.8687275985663086e-05,
+      "loss": 0.0011,
+      "step": 2740
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.866487455197133e-05,
+      "loss": 0.0011,
+      "step": 2741
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.864247311827957e-05,
+      "loss": 0.0007,
+      "step": 2742
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.862007168458782e-05,
+      "loss": 0.0006,
+      "step": 2743
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.859767025089606e-05,
+      "loss": 0.0005,
+      "step": 2744
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.8575268817204305e-05,
+      "loss": 1.0237,
+      "step": 2745
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.855286738351255e-05,
+      "loss": 2.4709,
+      "step": 2746
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.8530465949820786e-05,
+      "loss": 0.0011,
+      "step": 2747
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.8508064516129036e-05,
+      "loss": 0.001,
+      "step": 2748
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.848566308243728e-05,
+      "loss": 0.0005,
+      "step": 2749
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.846326164874552e-05,
+      "loss": 0.0008,
+      "step": 2750
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.844086021505376e-05,
+      "loss": 0.0095,
+      "step": 2751
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.841845878136201e-05,
+      "loss": 0.0011,
+      "step": 2752
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 3.839605734767025e-05,
+      "loss": 1.7736,
+      "step": 2753
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.837365591397849e-05,
+      "loss": 0.0012,
+      "step": 2754
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.8351254480286743e-05,
+      "loss": 0.0004,
+      "step": 2755
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.832885304659498e-05,
+      "loss": 0.0005,
+      "step": 2756
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.8306451612903224e-05,
+      "loss": 0.1702,
+      "step": 2757
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.8284050179211475e-05,
+      "loss": 0.0021,
+      "step": 2758
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.826164874551971e-05,
+      "loss": 0.0005,
+      "step": 2759
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.8239247311827956e-05,
+      "loss": 0.5549,
+      "step": 2760
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.821684587813621e-05,
+      "loss": 0.0007,
+      "step": 2761
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.8194444444444444e-05,
+      "loss": 1.0647,
+      "step": 2762
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.817204301075269e-05,
+      "loss": 0.0006,
+      "step": 2763
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.814964157706094e-05,
+      "loss": 0.0005,
+      "step": 2764
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8127240143369175e-05,
+      "loss": 0.0009,
+      "step": 2765
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.810483870967742e-05,
+      "loss": 0.0045,
+      "step": 2766
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.808243727598566e-05,
+      "loss": 0.0008,
+      "step": 2767
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.806003584229391e-05,
+      "loss": 0.0013,
+      "step": 2768
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.803763440860215e-05,
+      "loss": 0.0024,
+      "step": 2769
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8015232974910395e-05,
+      "loss": 0.0009,
+      "step": 2770
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.799283154121864e-05,
+      "loss": 0.0004,
+      "step": 2771
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.797043010752688e-05,
+      "loss": 0.0004,
+      "step": 2772
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.7948028673835126e-05,
+      "loss": 0.0007,
+      "step": 2773
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.792562724014337e-05,
+      "loss": 0.0029,
+      "step": 2774
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.7903225806451614e-05,
+      "loss": 1.3122,
+      "step": 2775
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.788082437275986e-05,
+      "loss": 2.6969,
+      "step": 2776
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.78584229390681e-05,
+      "loss": 0.0004,
+      "step": 2777
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.7836021505376346e-05,
+      "loss": 3.2991,
+      "step": 2778
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.781362007168459e-05,
+      "loss": 0.0008,
+      "step": 2779
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.779121863799283e-05,
+      "loss": 0.001,
+      "step": 2780
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.776881720430108e-05,
+      "loss": 0.0004,
+      "step": 2781
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.774641577060932e-05,
+      "loss": 0.0019,
+      "step": 2782
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.7724014336917565e-05,
+      "loss": 0.0093,
+      "step": 2783
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.770161290322581e-05,
+      "loss": 2.9764,
+      "step": 2784
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.767921146953405e-05,
+      "loss": 0.001,
+      "step": 2785
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.7656810035842296e-05,
+      "loss": 0.0006,
+      "step": 2786
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.763440860215054e-05,
+      "loss": 1.5191,
+      "step": 2787
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.7612007168458784e-05,
+      "loss": 1.5257,
+      "step": 2788
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.758960573476703e-05,
+      "loss": 0.0003,
+      "step": 2789
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.756720430107527e-05,
+      "loss": 0.9546,
+      "step": 2790
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.7544802867383516e-05,
+      "loss": 0.0031,
+      "step": 2791
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.752240143369175e-05,
+      "loss": 0.0496,
+      "step": 2792
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.0176,
+      "step": 2793
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.747759856630825e-05,
+      "loss": 0.0017,
+      "step": 2794
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.7455197132616484e-05,
+      "loss": 0.0006,
+      "step": 2795
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.7432795698924735e-05,
+      "loss": 0.0005,
+      "step": 2796
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.741039426523298e-05,
+      "loss": 0.0172,
+      "step": 2797
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 3.7387992831541216e-05,
+      "loss": 0.0004,
+      "step": 2798
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.736559139784947e-05,
+      "loss": 0.02,
+      "step": 2799
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.734318996415771e-05,
+      "loss": 0.0189,
+      "step": 2800
+    },
+    {
+      "epoch": 5.02,
+      "eval_accuracy": 0.6857143044471741,
+      "eval_loss": 1.5654765367507935,
+      "eval_runtime": 74.2978,
+      "eval_samples_per_second": 3.769,
+      "eval_steps_per_second": 3.769,
+      "step": 2800
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.732078853046595e-05,
+      "loss": 2.5551,
+      "step": 2801
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.72983870967742e-05,
+      "loss": 0.0004,
+      "step": 2802
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.727598566308244e-05,
+      "loss": 0.0006,
+      "step": 2803
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.725358422939068e-05,
+      "loss": 0.0028,
+      "step": 2804
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.723118279569893e-05,
+      "loss": 0.0008,
+      "step": 2805
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.7208781362007174e-05,
+      "loss": 0.0005,
+      "step": 2806
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.718637992831541e-05,
+      "loss": 2.5656,
+      "step": 2807
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.7163978494623655e-05,
+      "loss": 0.0017,
+      "step": 2808
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.7141577060931905e-05,
+      "loss": 0.0005,
+      "step": 2809
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.711917562724014e-05,
+      "loss": 0.0006,
+      "step": 2810
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 0.0005,
+      "step": 2811
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.707437275985664e-05,
+      "loss": 0.0014,
+      "step": 2812
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.7051971326164874e-05,
+      "loss": 0.0008,
+      "step": 2813
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.702956989247312e-05,
+      "loss": 0.0005,
+      "step": 2814
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 3.700716845878137e-05,
+      "loss": 0.0005,
+      "step": 2815
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.6984767025089606e-05,
+      "loss": 0.0004,
+      "step": 2816
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.696236559139785e-05,
+      "loss": 0.0006,
+      "step": 2817
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.69399641577061e-05,
+      "loss": 0.0142,
+      "step": 2818
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.691756272401434e-05,
+      "loss": 0.002,
+      "step": 2819
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.689516129032258e-05,
+      "loss": 0.0008,
+      "step": 2820
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.687275985663083e-05,
+      "loss": 0.0005,
+      "step": 2821
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.685035842293907e-05,
+      "loss": 0.0018,
+      "step": 2822
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.682795698924731e-05,
+      "loss": 0.0016,
+      "step": 2823
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.6805555555555556e-05,
+      "loss": 0.0005,
+      "step": 2824
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.67831541218638e-05,
+      "loss": 0.0047,
+      "step": 2825
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.6760752688172044e-05,
+      "loss": 0.0007,
+      "step": 2826
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.673835125448029e-05,
+      "loss": 0.5937,
+      "step": 2827
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.671594982078853e-05,
+      "loss": 0.0016,
+      "step": 2828
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.6693548387096776e-05,
+      "loss": 0.0007,
+      "step": 2829
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.667114695340502e-05,
+      "loss": 0.0003,
+      "step": 2830
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 3.6648745519713264e-05,
+      "loss": 0.0005,
+      "step": 2831
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.662634408602151e-05,
+      "loss": 0.0032,
+      "step": 2832
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.660394265232975e-05,
+      "loss": 0.0004,
+      "step": 2833
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.6581541218637995e-05,
+      "loss": 0.0063,
+      "step": 2834
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.655913978494624e-05,
+      "loss": 0.0005,
+      "step": 2835
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.653673835125448e-05,
+      "loss": 1.5316,
+      "step": 2836
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.651433691756273e-05,
+      "loss": 0.0007,
+      "step": 2837
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.649193548387097e-05,
+      "loss": 0.0077,
+      "step": 2838
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.6469534050179214e-05,
+      "loss": 0.0727,
+      "step": 2839
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.644713261648746e-05,
+      "loss": 0.0011,
+      "step": 2840
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.64247311827957e-05,
+      "loss": 0.0011,
+      "step": 2841
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.640232974910394e-05,
+      "loss": 0.132,
+      "step": 2842
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 3.637992831541219e-05,
+      "loss": 0.0005,
+      "step": 2843
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.6357526881720434e-05,
+      "loss": 0.0005,
+      "step": 2844
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.633512544802867e-05,
+      "loss": 0.0007,
+      "step": 2845
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.631272401433692e-05,
+      "loss": 0.0006,
+      "step": 2846
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.6290322580645165e-05,
+      "loss": 0.0005,
+      "step": 2847
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.62679211469534e-05,
+      "loss": 0.0005,
+      "step": 2848
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.624551971326165e-05,
+      "loss": 0.0005,
+      "step": 2849
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.62231182795699e-05,
+      "loss": 0.0006,
+      "step": 2850
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.6200716845878134e-05,
+      "loss": 0.0006,
+      "step": 2851
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.617831541218638e-05,
+      "loss": 0.0003,
+      "step": 2852
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.615591397849463e-05,
+      "loss": 0.0114,
+      "step": 2853
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 3.6133512544802866e-05,
+      "loss": 1.297,
+      "step": 2854
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.0016,
+      "step": 2855
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.608870967741936e-05,
+      "loss": 0.0006,
+      "step": 2856
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.60663082437276e-05,
+      "loss": 0.0769,
+      "step": 2857
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.604390681003584e-05,
+      "loss": 0.0006,
+      "step": 2858
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 3.602150537634409e-05,
+      "loss": 0.0116,
+      "step": 2859
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.599910394265233e-05,
+      "loss": 0.0009,
+      "step": 2860
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.597670250896057e-05,
+      "loss": 0.3105,
+      "step": 2861
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.595430107526882e-05,
+      "loss": 0.0004,
+      "step": 2862
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.593189964157706e-05,
+      "loss": 0.0034,
+      "step": 2863
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.5909498207885304e-05,
+      "loss": 0.0011,
+      "step": 2864
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 3.5887096774193555e-05,
+      "loss": 0.0007,
+      "step": 2865
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.586469534050179e-05,
+      "loss": 0.0007,
+      "step": 2866
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.5842293906810036e-05,
+      "loss": 0.0008,
+      "step": 2867
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.581989247311828e-05,
+      "loss": 3.3676,
+      "step": 2868
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.5797491039426524e-05,
+      "loss": 0.0005,
+      "step": 2869
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 3.577508960573477e-05,
+      "loss": 0.0183,
+      "step": 2870
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.575268817204301e-05,
+      "loss": 0.0006,
+      "step": 2871
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.5730286738351255e-05,
+      "loss": 0.0006,
+      "step": 2872
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.57078853046595e-05,
+      "loss": 0.0004,
+      "step": 2873
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.568548387096774e-05,
+      "loss": 0.0109,
+      "step": 2874
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.566308243727599e-05,
+      "loss": 0.0005,
+      "step": 2875
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 3.564068100358423e-05,
+      "loss": 0.0004,
+      "step": 2876
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.5618279569892474e-05,
+      "loss": 0.0012,
+      "step": 2877
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.559587813620072e-05,
+      "loss": 0.0004,
+      "step": 2878
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.557347670250896e-05,
+      "loss": 0.0003,
+      "step": 2879
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.5551075268817206e-05,
+      "loss": 1.9357,
+      "step": 2880
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.552867383512545e-05,
+      "loss": 0.0006,
+      "step": 2881
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 3.5506272401433694e-05,
+      "loss": 0.0039,
+      "step": 2882
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.548387096774194e-05,
+      "loss": 0.0028,
+      "step": 2883
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.546146953405018e-05,
+      "loss": 0.0003,
+      "step": 2884
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.5439068100358425e-05,
+      "loss": 0.0007,
+      "step": 2885
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.541666666666667e-05,
+      "loss": 0.0005,
+      "step": 2886
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 3.539426523297491e-05,
+      "loss": 0.0005,
+      "step": 2887
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.537186379928316e-05,
+      "loss": 0.093,
+      "step": 2888
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.53494623655914e-05,
+      "loss": 0.001,
+      "step": 2889
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.5327060931899645e-05,
+      "loss": 0.0004,
+      "step": 2890
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.530465949820789e-05,
+      "loss": 0.0005,
+      "step": 2891
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.528225806451613e-05,
+      "loss": 0.0062,
+      "step": 2892
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 3.525985663082437e-05,
+      "loss": 0.0006,
+      "step": 2893
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.523745519713262e-05,
+      "loss": 0.0194,
+      "step": 2894
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.5215053763440864e-05,
+      "loss": 0.3527,
+      "step": 2895
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.51926523297491e-05,
+      "loss": 0.0005,
+      "step": 2896
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.517025089605735e-05,
+      "loss": 0.0005,
+      "step": 2897
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 3.5147849462365596e-05,
+      "loss": 0.0003,
+      "step": 2898
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.512544802867383e-05,
+      "loss": 0.0027,
+      "step": 2899
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.510304659498208e-05,
+      "loss": 0.7838,
+      "step": 2900
+    },
+    {
+      "epoch": 5.2,
+      "eval_accuracy": 0.7107142806053162,
+      "eval_loss": 1.6127811670303345,
+      "eval_runtime": 74.0774,
+      "eval_samples_per_second": 3.78,
+      "eval_steps_per_second": 3.78,
+      "step": 2900
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.508064516129033e-05,
+      "loss": 0.0006,
+      "step": 2901
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.5058243727598564e-05,
+      "loss": 0.0004,
+      "step": 2902
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.5035842293906815e-05,
+      "loss": 0.0014,
+      "step": 2903
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 3.501344086021506e-05,
+      "loss": 0.0007,
+      "step": 2904
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.4991039426523296e-05,
+      "loss": 0.0005,
+      "step": 2905
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.4968637992831546e-05,
+      "loss": 0.0167,
+      "step": 2906
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.494623655913979e-05,
+      "loss": 0.0004,
+      "step": 2907
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.492383512544803e-05,
+      "loss": 0.0003,
+      "step": 2908
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 3.490143369175627e-05,
+      "loss": 0.0005,
+      "step": 2909
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.487903225806452e-05,
+      "loss": 0.0011,
+      "step": 2910
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.485663082437276e-05,
+      "loss": 0.0838,
+      "step": 2911
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.4834229390681e-05,
+      "loss": 1.5313,
+      "step": 2912
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.4811827956989254e-05,
+      "loss": 0.7172,
+      "step": 2913
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.478942652329749e-05,
+      "loss": 0.1075,
+      "step": 2914
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 3.4767025089605734e-05,
+      "loss": 0.0004,
+      "step": 2915
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.4744623655913985e-05,
+      "loss": 0.0003,
+      "step": 2916
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.0197,
+      "step": 2917
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.4699820788530466e-05,
+      "loss": 0.0007,
+      "step": 2918
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.467741935483872e-05,
+      "loss": 0.0007,
+      "step": 2919
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.4655017921146954e-05,
+      "loss": 0.0003,
+      "step": 2920
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 3.46326164874552e-05,
+      "loss": 0.0005,
+      "step": 2921
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.461021505376345e-05,
+      "loss": 0.0022,
+      "step": 2922
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.4587813620071685e-05,
+      "loss": 0.0007,
+      "step": 2923
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.456541218637993e-05,
+      "loss": 0.0006,
+      "step": 2924
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.454301075268817e-05,
+      "loss": 0.0009,
+      "step": 2925
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 3.452060931899642e-05,
+      "loss": 0.0004,
+      "step": 2926
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.449820788530466e-05,
+      "loss": 0.0005,
+      "step": 2927
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.4475806451612905e-05,
+      "loss": 0.0003,
+      "step": 2928
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.445340501792115e-05,
+      "loss": 0.0005,
+      "step": 2929
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.443100358422939e-05,
+      "loss": 0.0005,
+      "step": 2930
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.4408602150537636e-05,
+      "loss": 0.0003,
+      "step": 2931
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 3.438620071684588e-05,
+      "loss": 0.0006,
+      "step": 2932
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.4363799283154124e-05,
+      "loss": 0.0008,
+      "step": 2933
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.434139784946237e-05,
+      "loss": 0.8923,
+      "step": 2934
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.431899641577061e-05,
+      "loss": 0.0003,
+      "step": 2935
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.4296594982078856e-05,
+      "loss": 0.0422,
+      "step": 2936
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 3.427419354838709e-05,
+      "loss": 0.0003,
+      "step": 2937
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.425179211469534e-05,
+      "loss": 0.0004,
+      "step": 2938
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.422939068100359e-05,
+      "loss": 0.0004,
+      "step": 2939
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.4206989247311824e-05,
+      "loss": 0.0017,
+      "step": 2940
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.4184587813620075e-05,
+      "loss": 1.2482,
+      "step": 2941
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.416218637992832e-05,
+      "loss": 0.0036,
+      "step": 2942
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 3.4139784946236556e-05,
+      "loss": 0.0004,
+      "step": 2943
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.4117383512544806e-05,
+      "loss": 0.0079,
+      "step": 2944
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.409498207885305e-05,
+      "loss": 0.0006,
+      "step": 2945
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.407258064516129e-05,
+      "loss": 0.0005,
+      "step": 2946
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.405017921146954e-05,
+      "loss": 0.0026,
+      "step": 2947
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.402777777777778e-05,
+      "loss": 0.0005,
+      "step": 2948
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 3.400537634408602e-05,
+      "loss": 0.4054,
+      "step": 2949
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.398297491039426e-05,
+      "loss": 0.0004,
+      "step": 2950
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.3960573476702514e-05,
+      "loss": 0.001,
+      "step": 2951
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.393817204301075e-05,
+      "loss": 0.0004,
+      "step": 2952
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.3915770609318994e-05,
+      "loss": 0.0288,
+      "step": 2953
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 3.3893369175627245e-05,
+      "loss": 0.0004,
+      "step": 2954
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.387096774193548e-05,
+      "loss": 0.0008,
+      "step": 2955
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.3848566308243726e-05,
+      "loss": 0.0625,
+      "step": 2956
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.382616487455198e-05,
+      "loss": 0.0004,
+      "step": 2957
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.3803763440860214e-05,
+      "loss": 0.0004,
+      "step": 2958
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.378136200716846e-05,
+      "loss": 0.0012,
+      "step": 2959
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 3.375896057347671e-05,
+      "loss": 0.0004,
+      "step": 2960
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.3736559139784945e-05,
+      "loss": 0.0005,
+      "step": 2961
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.371415770609319e-05,
+      "loss": 0.0008,
+      "step": 2962
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.369175627240144e-05,
+      "loss": 0.0003,
+      "step": 2963
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.366935483870968e-05,
+      "loss": 0.0007,
+      "step": 2964
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 3.364695340501792e-05,
+      "loss": 0.0052,
+      "step": 2965
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.3624551971326165e-05,
+      "loss": 0.0063,
+      "step": 2966
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.360215053763441e-05,
+      "loss": 0.0009,
+      "step": 2967
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.357974910394265e-05,
+      "loss": 0.0004,
+      "step": 2968
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.3557347670250896e-05,
+      "loss": 0.0003,
+      "step": 2969
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.353494623655914e-05,
+      "loss": 0.0004,
+      "step": 2970
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 3.3512544802867384e-05,
+      "loss": 0.0004,
+      "step": 2971
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.349014336917563e-05,
+      "loss": 0.001,
+      "step": 2972
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.346774193548387e-05,
+      "loss": 0.0005,
+      "step": 2973
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.3445340501792116e-05,
+      "loss": 0.0014,
+      "step": 2974
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.342293906810036e-05,
+      "loss": 0.0857,
+      "step": 2975
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 3.34005376344086e-05,
+      "loss": 0.0011,
+      "step": 2976
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.337813620071685e-05,
+      "loss": 0.0037,
+      "step": 2977
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.335573476702509e-05,
+      "loss": 0.0006,
+      "step": 2978
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.0005,
+      "step": 2979
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.331093189964158e-05,
+      "loss": 0.0125,
+      "step": 2980
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.328853046594982e-05,
+      "loss": 0.0004,
+      "step": 2981
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 3.3266129032258067e-05,
+      "loss": 0.0004,
+      "step": 2982
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.324372759856631e-05,
+      "loss": 0.0006,
+      "step": 2983
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.3221326164874554e-05,
+      "loss": 0.0004,
+      "step": 2984
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.31989247311828e-05,
+      "loss": 0.0215,
+      "step": 2985
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.317652329749104e-05,
+      "loss": 0.0071,
+      "step": 2986
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.3154121863799286e-05,
+      "loss": 0.0005,
+      "step": 2987
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 3.313172043010753e-05,
+      "loss": 0.0013,
+      "step": 2988
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.3109318996415774e-05,
+      "loss": 0.0003,
+      "step": 2989
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.308691756272402e-05,
+      "loss": 0.002,
+      "step": 2990
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.306451612903226e-05,
+      "loss": 0.001,
+      "step": 2991
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.3042114695340505e-05,
+      "loss": 0.0006,
+      "step": 2992
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 3.301971326164875e-05,
+      "loss": 0.0003,
+      "step": 2993
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.2997311827956986e-05,
+      "loss": 2.1468,
+      "step": 2994
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.297491039426524e-05,
+      "loss": 0.0004,
+      "step": 2995
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.295250896057348e-05,
+      "loss": 0.0909,
+      "step": 2996
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.293010752688172e-05,
+      "loss": 1.2581,
+      "step": 2997
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.290770609318997e-05,
+      "loss": 0.0005,
+      "step": 2998
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 3.288530465949821e-05,
+      "loss": 0.0006,
+      "step": 2999
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.286290322580645e-05,
+      "loss": 0.0274,
+      "step": 3000
+    },
+    {
+      "epoch": 5.38,
+      "eval_accuracy": 0.7035714387893677,
+      "eval_loss": 1.620468020439148,
+      "eval_runtime": 73.5589,
+      "eval_samples_per_second": 3.806,
+      "eval_steps_per_second": 3.806,
+      "step": 3000
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.28405017921147e-05,
+      "loss": 0.0004,
+      "step": 3001
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.2818100358422944e-05,
+      "loss": 0.0007,
+      "step": 3002
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.279569892473118e-05,
+      "loss": 0.0149,
+      "step": 3003
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 3.277329749103943e-05,
+      "loss": 0.0003,
+      "step": 3004
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.2750896057347675e-05,
+      "loss": 0.0003,
+      "step": 3005
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.272849462365591e-05,
+      "loss": 0.0086,
+      "step": 3006
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.270609318996416e-05,
+      "loss": 0.0003,
+      "step": 3007
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.268369175627241e-05,
+      "loss": 0.0012,
+      "step": 3008
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.2661290322580644e-05,
+      "loss": 0.0015,
+      "step": 3009
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 3.263888888888889e-05,
+      "loss": 0.0005,
+      "step": 3010
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.261648745519714e-05,
+      "loss": 0.0034,
+      "step": 3011
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.2594086021505376e-05,
+      "loss": 0.0007,
+      "step": 3012
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.257168458781362e-05,
+      "loss": 0.0018,
+      "step": 3013
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.254928315412187e-05,
+      "loss": 0.0004,
+      "step": 3014
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 3.252688172043011e-05,
+      "loss": 0.0021,
+      "step": 3015
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.250448028673835e-05,
+      "loss": 0.0005,
+      "step": 3016
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.24820788530466e-05,
+      "loss": 0.0005,
+      "step": 3017
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.245967741935484e-05,
+      "loss": 0.0004,
+      "step": 3018
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.243727598566308e-05,
+      "loss": 0.0003,
+      "step": 3019
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.241487455197133e-05,
+      "loss": 0.0003,
+      "step": 3020
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 3.239247311827957e-05,
+      "loss": 0.0004,
+      "step": 3021
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.2370071684587814e-05,
+      "loss": 0.0004,
+      "step": 3022
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.2347670250896065e-05,
+      "loss": 0.0002,
+      "step": 3023
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.23252688172043e-05,
+      "loss": 0.0002,
+      "step": 3024
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.2302867383512546e-05,
+      "loss": 0.0003,
+      "step": 3025
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.228046594982079e-05,
+      "loss": 0.0019,
+      "step": 3026
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 3.2258064516129034e-05,
+      "loss": 0.0006,
+      "step": 3027
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.223566308243728e-05,
+      "loss": 0.0004,
+      "step": 3028
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.221326164874552e-05,
+      "loss": 0.0007,
+      "step": 3029
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.2190860215053765e-05,
+      "loss": 0.001,
+      "step": 3030
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.216845878136201e-05,
+      "loss": 0.0013,
+      "step": 3031
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 3.214605734767025e-05,
+      "loss": 0.0015,
+      "step": 3032
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.21236559139785e-05,
+      "loss": 0.0989,
+      "step": 3033
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.210125448028674e-05,
+      "loss": 2.4395,
+      "step": 3034
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.207885304659498e-05,
+      "loss": 0.0004,
+      "step": 3035
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.205645161290323e-05,
+      "loss": 0.0032,
+      "step": 3036
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.203405017921147e-05,
+      "loss": 0.0004,
+      "step": 3037
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 3.201164874551971e-05,
+      "loss": 0.0004,
+      "step": 3038
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.198924731182796e-05,
+      "loss": 0.0003,
+      "step": 3039
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.1966845878136204e-05,
+      "loss": 0.7205,
+      "step": 3040
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.0004,
+      "step": 3041
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.192204301075269e-05,
+      "loss": 0.0004,
+      "step": 3042
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 3.1899641577060935e-05,
+      "loss": 0.0265,
+      "step": 3043
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.187724014336917e-05,
+      "loss": 0.0341,
+      "step": 3044
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.185483870967742e-05,
+      "loss": 0.0003,
+      "step": 3045
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.183243727598567e-05,
+      "loss": 0.0003,
+      "step": 3046
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1810035842293904e-05,
+      "loss": 0.0746,
+      "step": 3047
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.1787634408602155e-05,
+      "loss": 0.0003,
+      "step": 3048
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 3.17652329749104e-05,
+      "loss": 0.0009,
+      "step": 3049
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.1742831541218636e-05,
+      "loss": 0.2729,
+      "step": 3050
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.172043010752688e-05,
+      "loss": 0.0004,
+      "step": 3051
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.169802867383513e-05,
+      "loss": 0.0003,
+      "step": 3052
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.167562724014337e-05,
+      "loss": 0.0028,
+      "step": 3053
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.165322580645161e-05,
+      "loss": 4.6329,
+      "step": 3054
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 3.163082437275986e-05,
+      "loss": 0.0003,
+      "step": 3055
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.16084229390681e-05,
+      "loss": 0.0027,
+      "step": 3056
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.158602150537634e-05,
+      "loss": 0.0005,
+      "step": 3057
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.156362007168459e-05,
+      "loss": 0.0006,
+      "step": 3058
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.154121863799283e-05,
+      "loss": 0.0005,
+      "step": 3059
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 3.1518817204301074e-05,
+      "loss": 0.0003,
+      "step": 3060
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1496415770609325e-05,
+      "loss": 0.0004,
+      "step": 3061
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.147401433691756e-05,
+      "loss": 0.0007,
+      "step": 3062
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1451612903225806e-05,
+      "loss": 0.0005,
+      "step": 3063
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1429211469534057e-05,
+      "loss": 0.0004,
+      "step": 3064
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.1406810035842294e-05,
+      "loss": 0.0003,
+      "step": 3065
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 3.138440860215054e-05,
+      "loss": 0.0009,
+      "step": 3066
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.136200716845878e-05,
+      "loss": 0.0005,
+      "step": 3067
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.1339605734767025e-05,
+      "loss": 0.0003,
+      "step": 3068
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.131720430107527e-05,
+      "loss": 0.0005,
+      "step": 3069
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.129480286738351e-05,
+      "loss": 0.0004,
+      "step": 3070
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 3.127240143369176e-05,
+      "loss": 0.0014,
+      "step": 3071
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.125e-05,
+      "loss": 1.0011,
+      "step": 3072
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.1227598566308245e-05,
+      "loss": 0.0015,
+      "step": 3073
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.120519713261649e-05,
+      "loss": 0.8447,
+      "step": 3074
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.118279569892473e-05,
+      "loss": 0.0004,
+      "step": 3075
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.1160394265232976e-05,
+      "loss": 0.0003,
+      "step": 3076
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 3.113799283154122e-05,
+      "loss": 0.0004,
+      "step": 3077
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1115591397849464e-05,
+      "loss": 0.0022,
+      "step": 3078
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.109318996415771e-05,
+      "loss": 0.0004,
+      "step": 3079
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.107078853046595e-05,
+      "loss": 0.0003,
+      "step": 3080
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.1048387096774195e-05,
+      "loss": 0.01,
+      "step": 3081
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 3.102598566308244e-05,
+      "loss": 0.0049,
+      "step": 3082
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.100358422939068e-05,
+      "loss": 0.0003,
+      "step": 3083
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.098118279569893e-05,
+      "loss": 0.0006,
+      "step": 3084
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.095878136200717e-05,
+      "loss": 0.0004,
+      "step": 3085
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.0936379928315415e-05,
+      "loss": 0.0005,
+      "step": 3086
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.091397849462366e-05,
+      "loss": 0.0005,
+      "step": 3087
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 3.08915770609319e-05,
+      "loss": 0.0006,
+      "step": 3088
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0869175627240146e-05,
+      "loss": 0.0082,
+      "step": 3089
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.084677419354839e-05,
+      "loss": 0.0005,
+      "step": 3090
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0824372759856634e-05,
+      "loss": 0.0003,
+      "step": 3091
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.080197132616488e-05,
+      "loss": 0.0002,
+      "step": 3092
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.077956989247312e-05,
+      "loss": 0.0017,
+      "step": 3093
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 3.0757168458781366e-05,
+      "loss": 0.0003,
+      "step": 3094
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.07347670250896e-05,
+      "loss": 0.0004,
+      "step": 3095
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.071236559139785e-05,
+      "loss": 0.001,
+      "step": 3096
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.06899641577061e-05,
+      "loss": 0.0005,
+      "step": 3097
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.0667562724014334e-05,
+      "loss": 1.3811,
+      "step": 3098
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 3.0645161290322585e-05,
+      "loss": 0.0006,
+      "step": 3099
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.062275985663083e-05,
+      "loss": 0.0003,
+      "step": 3100
+    },
+    {
+      "epoch": 5.56,
+      "eval_accuracy": 0.6857143044471741,
+      "eval_loss": 1.7564141750335693,
+      "eval_runtime": 69.5836,
+      "eval_samples_per_second": 4.024,
+      "eval_steps_per_second": 4.024,
+      "step": 3100
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.0600358422939066e-05,
+      "loss": 0.0004,
+      "step": 3101
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.0577956989247317e-05,
+      "loss": 0.0015,
+      "step": 3102
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.055555555555556e-05,
+      "loss": 0.0014,
+      "step": 3103
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.05331541218638e-05,
+      "loss": 0.0005,
+      "step": 3104
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 3.0510752688172045e-05,
+      "loss": 0.0007,
+      "step": 3105
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.0488351254480292e-05,
+      "loss": 0.0003,
+      "step": 3106
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.046594982078853e-05,
+      "loss": 0.0003,
+      "step": 3107
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.0443548387096776e-05,
+      "loss": 0.0015,
+      "step": 3108
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.0421146953405024e-05,
+      "loss": 0.0003,
+      "step": 3109
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 3.039874551971326e-05,
+      "loss": 0.0003,
+      "step": 3110
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.0376344086021508e-05,
+      "loss": 0.0004,
+      "step": 3111
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.0353942652329752e-05,
+      "loss": 0.0003,
+      "step": 3112
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.0331541218637992e-05,
+      "loss": 0.0003,
+      "step": 3113
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.030913978494624e-05,
+      "loss": 0.0003,
+      "step": 3114
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.0286738351254483e-05,
+      "loss": 0.0004,
+      "step": 3115
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 3.0264336917562724e-05,
+      "loss": 0.0005,
+      "step": 3116
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.024193548387097e-05,
+      "loss": 0.0003,
+      "step": 3117
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.0219534050179215e-05,
+      "loss": 0.0003,
+      "step": 3118
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.0197132616487455e-05,
+      "loss": 0.0004,
+      "step": 3119
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.0174731182795703e-05,
+      "loss": 0.0047,
+      "step": 3120
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.015232974910394e-05,
+      "loss": 0.0003,
+      "step": 3121
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 3.0129928315412187e-05,
+      "loss": 0.0003,
+      "step": 3122
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.010752688172043e-05,
+      "loss": 0.0563,
+      "step": 3123
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.008512544802867e-05,
+      "loss": 0.0005,
+      "step": 3124
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.006272401433692e-05,
+      "loss": 0.0006,
+      "step": 3125
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.0040322580645162e-05,
+      "loss": 0.0005,
+      "step": 3126
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 3.0017921146953403e-05,
+      "loss": 0.0004,
+      "step": 3127
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.999551971326165e-05,
+      "loss": 0.2361,
+      "step": 3128
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.9973118279569894e-05,
+      "loss": 0.0005,
+      "step": 3129
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.9950716845878135e-05,
+      "loss": 0.0006,
+      "step": 3130
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.9928315412186382e-05,
+      "loss": 0.0006,
+      "step": 3131
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.9905913978494626e-05,
+      "loss": 0.0004,
+      "step": 3132
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 2.9883512544802866e-05,
+      "loss": 0.0006,
+      "step": 3133
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.9861111111111113e-05,
+      "loss": 0.0003,
+      "step": 3134
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.9838709677419357e-05,
+      "loss": 0.0005,
+      "step": 3135
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.9816308243727598e-05,
+      "loss": 0.0006,
+      "step": 3136
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.979390681003584e-05,
+      "loss": 0.0003,
+      "step": 3137
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.977150537634409e-05,
+      "loss": 0.0005,
+      "step": 3138
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.974910394265233e-05,
+      "loss": 0.0003,
+      "step": 3139
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.9726702508960573e-05,
+      "loss": 0.0003,
+      "step": 3140
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.970430107526882e-05,
+      "loss": 0.2402,
+      "step": 3141
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.968189964157706e-05,
+      "loss": 0.0012,
+      "step": 3142
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.9659498207885305e-05,
+      "loss": 0.0008,
+      "step": 3143
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.9637096774193552e-05,
+      "loss": 0.0007,
+      "step": 3144
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.9614695340501792e-05,
+      "loss": 0.0008,
+      "step": 3145
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.9592293906810036e-05,
+      "loss": 0.0025,
+      "step": 3146
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.9569892473118284e-05,
+      "loss": 0.0009,
+      "step": 3147
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.9547491039426524e-05,
+      "loss": 0.0005,
+      "step": 3148
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.9525089605734768e-05,
+      "loss": 0.003,
+      "step": 3149
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.9502688172043015e-05,
+      "loss": 0.0002,
+      "step": 3150
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.9480286738351252e-05,
+      "loss": 0.0004,
+      "step": 3151
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.94578853046595e-05,
+      "loss": 0.0054,
+      "step": 3152
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.9435483870967743e-05,
+      "loss": 0.0004,
+      "step": 3153
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.9413082437275984e-05,
+      "loss": 0.0007,
+      "step": 3154
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.939068100358423e-05,
+      "loss": 0.0005,
+      "step": 3155
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.9368279569892475e-05,
+      "loss": 0.0003,
+      "step": 3156
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.9345878136200715e-05,
+      "loss": 0.0004,
+      "step": 3157
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.9323476702508963e-05,
+      "loss": 0.0002,
+      "step": 3158
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.9301075268817207e-05,
+      "loss": 0.0003,
+      "step": 3159
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.9278673835125447e-05,
+      "loss": 0.0003,
+      "step": 3160
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.9256272401433694e-05,
+      "loss": 0.089,
+      "step": 3161
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.9233870967741938e-05,
+      "loss": 0.0004,
+      "step": 3162
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.921146953405018e-05,
+      "loss": 0.0007,
+      "step": 3163
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.9189068100358426e-05,
+      "loss": 0.0005,
+      "step": 3164
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.0009,
+      "step": 3165
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.914426523297491e-05,
+      "loss": 0.0006,
+      "step": 3166
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.9121863799283154e-05,
+      "loss": 0.0008,
+      "step": 3167
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.90994623655914e-05,
+      "loss": 0.0003,
+      "step": 3168
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.9077060931899642e-05,
+      "loss": 0.0004,
+      "step": 3169
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.9054659498207886e-05,
+      "loss": 0.034,
+      "step": 3170
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.9032258064516133e-05,
+      "loss": 0.0003,
+      "step": 3171
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.9009856630824373e-05,
+      "loss": 0.0003,
+      "step": 3172
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.8987455197132617e-05,
+      "loss": 0.0002,
+      "step": 3173
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.8965053763440865e-05,
+      "loss": 0.0004,
+      "step": 3174
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.8942652329749105e-05,
+      "loss": 0.004,
+      "step": 3175
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.892025089605735e-05,
+      "loss": 0.0005,
+      "step": 3176
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.8897849462365596e-05,
+      "loss": 0.0006,
+      "step": 3177
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.8875448028673837e-05,
+      "loss": 0.0004,
+      "step": 3178
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.885304659498208e-05,
+      "loss": 0.0004,
+      "step": 3179
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.8830645161290328e-05,
+      "loss": 0.0012,
+      "step": 3180
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.8808243727598565e-05,
+      "loss": 0.0004,
+      "step": 3181
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.8785842293906812e-05,
+      "loss": 0.0007,
+      "step": 3182
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.8763440860215056e-05,
+      "loss": 0.0004,
+      "step": 3183
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.8741039426523296e-05,
+      "loss": 1.4748,
+      "step": 3184
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.8718637992831544e-05,
+      "loss": 0.0888,
+      "step": 3185
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.8696236559139787e-05,
+      "loss": 0.0012,
+      "step": 3186
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.8673835125448028e-05,
+      "loss": 0.0003,
+      "step": 3187
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.8651433691756275e-05,
+      "loss": 0.0005,
+      "step": 3188
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.862903225806452e-05,
+      "loss": 0.057,
+      "step": 3189
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.860663082437276e-05,
+      "loss": 0.0015,
+      "step": 3190
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.8584229390681007e-05,
+      "loss": 0.0003,
+      "step": 3191
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.856182795698925e-05,
+      "loss": 0.0005,
+      "step": 3192
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.853942652329749e-05,
+      "loss": 0.0005,
+      "step": 3193
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.851702508960574e-05,
+      "loss": 0.0004,
+      "step": 3194
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.8494623655913982e-05,
+      "loss": 0.0005,
+      "step": 3195
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.8472222222222223e-05,
+      "loss": 0.0003,
+      "step": 3196
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.8449820788530467e-05,
+      "loss": 0.0003,
+      "step": 3197
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.8427419354838714e-05,
+      "loss": 0.0003,
+      "step": 3198
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.8405017921146954e-05,
+      "loss": 0.0002,
+      "step": 3199
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.8382616487455198e-05,
+      "loss": 0.0007,
+      "step": 3200
+    },
+    {
+      "epoch": 5.73,
+      "eval_accuracy": 0.6821428537368774,
+      "eval_loss": 1.9412034749984741,
+      "eval_runtime": 74.5735,
+      "eval_samples_per_second": 3.755,
+      "eval_steps_per_second": 3.755,
+      "step": 3200
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.8360215053763445e-05,
+      "loss": 0.0009,
+      "step": 3201
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.8337813620071686e-05,
+      "loss": 0.0005,
+      "step": 3202
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.831541218637993e-05,
+      "loss": 0.0116,
+      "step": 3203
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.8293010752688177e-05,
+      "loss": 0.0008,
+      "step": 3204
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.8270609318996418e-05,
+      "loss": 0.0003,
+      "step": 3205
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.824820788530466e-05,
+      "loss": 0.0015,
+      "step": 3206
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.822580645161291e-05,
+      "loss": 0.0003,
+      "step": 3207
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.8203405017921146e-05,
+      "loss": 0.0002,
+      "step": 3208
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.8181003584229393e-05,
+      "loss": 0.0004,
+      "step": 3209
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.8158602150537637e-05,
+      "loss": 3.1218,
+      "step": 3210
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.8136200716845877e-05,
+      "loss": 0.0007,
+      "step": 3211
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.8113799283154125e-05,
+      "loss": 0.0013,
+      "step": 3212
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.8091397849462365e-05,
+      "loss": 0.6754,
+      "step": 3213
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.806899641577061e-05,
+      "loss": 0.0004,
+      "step": 3214
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.8046594982078856e-05,
+      "loss": 0.0003,
+      "step": 3215
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.8024193548387097e-05,
+      "loss": 0.0004,
+      "step": 3216
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.800179211469534e-05,
+      "loss": 0.0027,
+      "step": 3217
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.7979390681003588e-05,
+      "loss": 0.0004,
+      "step": 3218
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.7956989247311828e-05,
+      "loss": 0.0087,
+      "step": 3219
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.7934587813620072e-05,
+      "loss": 0.0002,
+      "step": 3220
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.791218637992832e-05,
+      "loss": 0.0003,
+      "step": 3221
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.7889784946236556e-05,
+      "loss": 0.0007,
+      "step": 3222
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.7867383512544804e-05,
+      "loss": 0.0003,
+      "step": 3223
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.7844982078853048e-05,
+      "loss": 0.0004,
+      "step": 3224
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.7822580645161288e-05,
+      "loss": 0.0013,
+      "step": 3225
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.7800179211469535e-05,
+      "loss": 0.0008,
+      "step": 3226
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.0003,
+      "step": 3227
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.775537634408602e-05,
+      "loss": 0.0009,
+      "step": 3228
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.7732974910394267e-05,
+      "loss": 0.0006,
+      "step": 3229
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.771057347670251e-05,
+      "loss": 0.0003,
+      "step": 3230
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.768817204301075e-05,
+      "loss": 0.0004,
+      "step": 3231
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.7665770609319e-05,
+      "loss": 0.0003,
+      "step": 3232
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.7643369175627242e-05,
+      "loss": 0.0002,
+      "step": 3233
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.7620967741935483e-05,
+      "loss": 0.0004,
+      "step": 3234
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.759856630824373e-05,
+      "loss": 0.0004,
+      "step": 3235
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.7576164874551974e-05,
+      "loss": 0.0003,
+      "step": 3236
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.7553763440860214e-05,
+      "loss": 0.0005,
+      "step": 3237
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.7531362007168458e-05,
+      "loss": 0.0003,
+      "step": 3238
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.7508960573476705e-05,
+      "loss": 0.0003,
+      "step": 3239
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.7486559139784946e-05,
+      "loss": 0.0008,
+      "step": 3240
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.746415770609319e-05,
+      "loss": 0.0014,
+      "step": 3241
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.7441756272401437e-05,
+      "loss": 0.0006,
+      "step": 3242
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.7419354838709678e-05,
+      "loss": 0.0003,
+      "step": 3243
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.739695340501792e-05,
+      "loss": 0.0003,
+      "step": 3244
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.737455197132617e-05,
+      "loss": 0.0006,
+      "step": 3245
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.735215053763441e-05,
+      "loss": 0.0004,
+      "step": 3246
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.7329749103942653e-05,
+      "loss": 0.005,
+      "step": 3247
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.73073476702509e-05,
+      "loss": 0.0005,
+      "step": 3248
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.728494623655914e-05,
+      "loss": 0.0004,
+      "step": 3249
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.7262544802867385e-05,
+      "loss": 0.0003,
+      "step": 3250
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.7240143369175632e-05,
+      "loss": 0.0004,
+      "step": 3251
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.721774193548387e-05,
+      "loss": 0.0004,
+      "step": 3252
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.7195340501792116e-05,
+      "loss": 0.0003,
+      "step": 3253
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.717293906810036e-05,
+      "loss": 2.7384,
+      "step": 3254
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.71505376344086e-05,
+      "loss": 0.0003,
+      "step": 3255
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.7128136200716848e-05,
+      "loss": 0.0032,
+      "step": 3256
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.710573476702509e-05,
+      "loss": 0.0003,
+      "step": 3257
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.7083333333333332e-05,
+      "loss": 0.0019,
+      "step": 3258
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.706093189964158e-05,
+      "loss": 0.0017,
+      "step": 3259
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.7038530465949823e-05,
+      "loss": 0.0002,
+      "step": 3260
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 2.7016129032258064e-05,
+      "loss": 0.0003,
+      "step": 3261
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.699372759856631e-05,
+      "loss": 0.0003,
+      "step": 3262
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.6971326164874555e-05,
+      "loss": 0.0005,
+      "step": 3263
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.6948924731182795e-05,
+      "loss": 0.0004,
+      "step": 3264
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.6926523297491043e-05,
+      "loss": 0.0006,
+      "step": 3265
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.6904121863799286e-05,
+      "loss": 0.0003,
+      "step": 3266
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.6881720430107527e-05,
+      "loss": 0.0003,
+      "step": 3267
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.685931899641577e-05,
+      "loss": 0.0004,
+      "step": 3268
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.6836917562724018e-05,
+      "loss": 0.0429,
+      "step": 3269
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.681451612903226e-05,
+      "loss": 0.0012,
+      "step": 3270
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.6792114695340502e-05,
+      "loss": 0.0032,
+      "step": 3271
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.676971326164875e-05,
+      "loss": 0.0002,
+      "step": 3272
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.674731182795699e-05,
+      "loss": 0.0013,
+      "step": 3273
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.6724910394265234e-05,
+      "loss": 0.0003,
+      "step": 3274
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.670250896057348e-05,
+      "loss": 0.0003,
+      "step": 3275
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.668010752688172e-05,
+      "loss": 0.2656,
+      "step": 3276
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.6657706093189965e-05,
+      "loss": 0.0025,
+      "step": 3277
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.6635304659498213e-05,
+      "loss": 0.0006,
+      "step": 3278
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.661290322580645e-05,
+      "loss": 0.0003,
+      "step": 3279
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.6590501792114697e-05,
+      "loss": 0.0004,
+      "step": 3280
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.6568100358422944e-05,
+      "loss": 0.0003,
+      "step": 3281
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.654569892473118e-05,
+      "loss": 0.0005,
+      "step": 3282
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.652329749103943e-05,
+      "loss": 0.0005,
+      "step": 3283
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.6500896057347673e-05,
+      "loss": 0.0006,
+      "step": 3284
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.6478494623655913e-05,
+      "loss": 0.0008,
+      "step": 3285
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.645609318996416e-05,
+      "loss": 0.0015,
+      "step": 3286
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.6433691756272404e-05,
+      "loss": 0.0028,
+      "step": 3287
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.6411290322580645e-05,
+      "loss": 0.0005,
+      "step": 3288
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 0.0007,
+      "step": 3289
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.6366487455197136e-05,
+      "loss": 3.7365,
+      "step": 3290
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.6344086021505376e-05,
+      "loss": 0.0003,
+      "step": 3291
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.6321684587813623e-05,
+      "loss": 0.0002,
+      "step": 3292
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.6299283154121867e-05,
+      "loss": 0.0003,
+      "step": 3293
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.6276881720430108e-05,
+      "loss": 0.0004,
+      "step": 3294
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.625448028673835e-05,
+      "loss": 0.0092,
+      "step": 3295
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.62320788530466e-05,
+      "loss": 0.0005,
+      "step": 3296
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.620967741935484e-05,
+      "loss": 0.0002,
+      "step": 3297
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.6187275985663083e-05,
+      "loss": 0.0003,
+      "step": 3298
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.616487455197133e-05,
+      "loss": 0.0004,
+      "step": 3299
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.614247311827957e-05,
+      "loss": 0.0002,
+      "step": 3300
+    },
+    {
+      "epoch": 5.91,
+      "eval_accuracy": 0.7178571224212646,
+      "eval_loss": 1.6739420890808105,
+      "eval_runtime": 73.8573,
+      "eval_samples_per_second": 3.791,
+      "eval_steps_per_second": 3.791,
+      "step": 3300
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.6120071684587815e-05,
+      "loss": 0.0003,
+      "step": 3301
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.6097670250896062e-05,
+      "loss": 0.0003,
+      "step": 3302
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.6075268817204303e-05,
+      "loss": 0.0004,
+      "step": 3303
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.6052867383512546e-05,
+      "loss": 0.0095,
+      "step": 3304
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.6030465949820787e-05,
+      "loss": 0.0004,
+      "step": 3305
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.6008064516129034e-05,
+      "loss": 0.0004,
+      "step": 3306
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.5985663082437278e-05,
+      "loss": 0.0005,
+      "step": 3307
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.596326164874552e-05,
+      "loss": 0.0005,
+      "step": 3308
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.5940860215053762e-05,
+      "loss": 0.0013,
+      "step": 3309
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.591845878136201e-05,
+      "loss": 0.0606,
+      "step": 3310
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.589605734767025e-05,
+      "loss": 0.0004,
+      "step": 3311
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.5873655913978494e-05,
+      "loss": 2.1427,
+      "step": 3312
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.585125448028674e-05,
+      "loss": 0.0003,
+      "step": 3313
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.582885304659498e-05,
+      "loss": 0.0004,
+      "step": 3314
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.5806451612903226e-05,
+      "loss": 0.0004,
+      "step": 3315
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.5784050179211473e-05,
+      "loss": 0.0006,
+      "step": 3316
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.5761648745519713e-05,
+      "loss": 0.0003,
+      "step": 3317
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.5739247311827957e-05,
+      "loss": 0.0004,
+      "step": 3318
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.5716845878136204e-05,
+      "loss": 0.0004,
+      "step": 3319
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.5694444444444445e-05,
+      "loss": 0.0003,
+      "step": 3320
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.567204301075269e-05,
+      "loss": 0.0023,
+      "step": 3321
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.5649641577060936e-05,
+      "loss": 0.0014,
+      "step": 3322
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.5627240143369173e-05,
+      "loss": 0.0003,
+      "step": 3323
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.560483870967742e-05,
+      "loss": 0.0002,
+      "step": 3324
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.5582437275985664e-05,
+      "loss": 0.0012,
+      "step": 3325
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.5560035842293905e-05,
+      "loss": 0.0002,
+      "step": 3326
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.5537634408602152e-05,
+      "loss": 0.0002,
+      "step": 3327
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.5515232974910396e-05,
+      "loss": 0.0003,
+      "step": 3328
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.5492831541218636e-05,
+      "loss": 0.0008,
+      "step": 3329
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.5470430107526883e-05,
+      "loss": 0.0004,
+      "step": 3330
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.5448028673835127e-05,
+      "loss": 0.0002,
+      "step": 3331
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.5425627240143368e-05,
+      "loss": 0.0004,
+      "step": 3332
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.5403225806451615e-05,
+      "loss": 0.0005,
+      "step": 3333
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.538082437275986e-05,
+      "loss": 0.0002,
+      "step": 3334
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.53584229390681e-05,
+      "loss": 0.0003,
+      "step": 3335
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.5336021505376347e-05,
+      "loss": 0.0023,
+      "step": 3336
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.531362007168459e-05,
+      "loss": 0.0197,
+      "step": 3337
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.529121863799283e-05,
+      "loss": 0.001,
+      "step": 3338
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.5268817204301075e-05,
+      "loss": 0.0003,
+      "step": 3339
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.5246415770609322e-05,
+      "loss": 0.0004,
+      "step": 3340
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.5224014336917563e-05,
+      "loss": 0.0007,
+      "step": 3341
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.5201612903225806e-05,
+      "loss": 2.6369,
+      "step": 3342
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.5179211469534054e-05,
+      "loss": 0.0086,
+      "step": 3343
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.5156810035842294e-05,
+      "loss": 0.0002,
+      "step": 3344
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.5134408602150538e-05,
+      "loss": 0.0003,
+      "step": 3345
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.5112007168458785e-05,
+      "loss": 0.0005,
+      "step": 3346
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.5089605734767026e-05,
+      "loss": 0.0005,
+      "step": 3347
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.506720430107527e-05,
+      "loss": 2.8147,
+      "step": 3348
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.5044802867383517e-05,
+      "loss": 0.0002,
+      "step": 3349
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.5022401433691754e-05,
+      "loss": 0.0004,
+      "step": 3350
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0006,
+      "step": 3351
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.4977598566308245e-05,
+      "loss": 0.0003,
+      "step": 3352
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.495519713261649e-05,
+      "loss": 0.0003,
+      "step": 3353
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.4932795698924733e-05,
+      "loss": 0.0005,
+      "step": 3354
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.4910394265232977e-05,
+      "loss": 0.0002,
+      "step": 3355
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 2.488799283154122e-05,
+      "loss": 0.0043,
+      "step": 3356
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.4865591397849464e-05,
+      "loss": 0.0004,
+      "step": 3357
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.4843189964157705e-05,
+      "loss": 0.0005,
+      "step": 3358
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.4820788530465952e-05,
+      "loss": 0.3395,
+      "step": 3359
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.4798387096774196e-05,
+      "loss": 0.0003,
+      "step": 3360
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.4775985663082436e-05,
+      "loss": 0.0002,
+      "step": 3361
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.4753584229390684e-05,
+      "loss": 0.0034,
+      "step": 3362
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.4731182795698928e-05,
+      "loss": 0.0002,
+      "step": 3363
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.4708781362007168e-05,
+      "loss": 0.0003,
+      "step": 3364
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.4686379928315415e-05,
+      "loss": 0.0003,
+      "step": 3365
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.4663978494623656e-05,
+      "loss": 0.0003,
+      "step": 3366
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 2.46415770609319e-05,
+      "loss": 0.0002,
+      "step": 3367
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.4619175627240147e-05,
+      "loss": 0.0003,
+      "step": 3368
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.4596774193548387e-05,
+      "loss": 0.0003,
+      "step": 3369
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.457437275985663e-05,
+      "loss": 0.0004,
+      "step": 3370
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.455197132616488e-05,
+      "loss": 0.0244,
+      "step": 3371
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.452956989247312e-05,
+      "loss": 0.0003,
+      "step": 3372
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.4507168458781363e-05,
+      "loss": 0.001,
+      "step": 3373
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.4484767025089607e-05,
+      "loss": 0.0004,
+      "step": 3374
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.446236559139785e-05,
+      "loss": 0.0008,
+      "step": 3375
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.4439964157706094e-05,
+      "loss": 0.0003,
+      "step": 3376
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.4417562724014338e-05,
+      "loss": 4.2345,
+      "step": 3377
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 2.4395161290322582e-05,
+      "loss": 0.0005,
+      "step": 3378
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.4372759856630826e-05,
+      "loss": 0.0004,
+      "step": 3379
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.4350358422939066e-05,
+      "loss": 0.0004,
+      "step": 3380
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.4327956989247314e-05,
+      "loss": 0.0003,
+      "step": 3381
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.4305555555555558e-05,
+      "loss": 0.0003,
+      "step": 3382
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.4283154121863798e-05,
+      "loss": 0.0002,
+      "step": 3383
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 2.4260752688172045e-05,
+      "loss": 0.0015,
+      "step": 3384
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.423835125448029e-05,
+      "loss": 0.0003,
+      "step": 3385
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.421594982078853e-05,
+      "loss": 0.0003,
+      "step": 3386
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.4193548387096777e-05,
+      "loss": 0.0002,
+      "step": 3387
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.4171146953405017e-05,
+      "loss": 0.0002,
+      "step": 3388
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.414874551971326e-05,
+      "loss": 0.0003,
+      "step": 3389
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.412634408602151e-05,
+      "loss": 0.003,
+      "step": 3390
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.410394265232975e-05,
+      "loss": 0.0005,
+      "step": 3391
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.4081541218637993e-05,
+      "loss": 0.0004,
+      "step": 3392
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.405913978494624e-05,
+      "loss": 0.0002,
+      "step": 3393
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.403673835125448e-05,
+      "loss": 0.0003,
+      "step": 3394
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 2.4014336917562724e-05,
+      "loss": 0.0002,
+      "step": 3395
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.3991935483870968e-05,
+      "loss": 0.0012,
+      "step": 3396
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.3969534050179212e-05,
+      "loss": 0.0005,
+      "step": 3397
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.3947132616487456e-05,
+      "loss": 0.0002,
+      "step": 3398
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.39247311827957e-05,
+      "loss": 0.0002,
+      "step": 3399
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.3902329749103944e-05,
+      "loss": 0.0003,
+      "step": 3400
+    },
+    {
+      "epoch": 6.09,
+      "eval_accuracy": 0.7107142806053162,
+      "eval_loss": 1.5762828588485718,
+      "eval_runtime": 73.8591,
+      "eval_samples_per_second": 3.791,
+      "eval_steps_per_second": 3.791,
+      "step": 3400
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.3879928315412188e-05,
+      "loss": 0.0005,
+      "step": 3401
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.385752688172043e-05,
+      "loss": 0.0006,
+      "step": 3402
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.3835125448028675e-05,
+      "loss": 0.0004,
+      "step": 3403
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.381272401433692e-05,
+      "loss": 0.0003,
+      "step": 3404
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.3790322580645163e-05,
+      "loss": 0.0004,
+      "step": 3405
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 2.3767921146953407e-05,
+      "loss": 0.0002,
+      "step": 3406
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.374551971326165e-05,
+      "loss": 0.0005,
+      "step": 3407
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.3723118279569895e-05,
+      "loss": 0.0889,
+      "step": 3408
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.370071684587814e-05,
+      "loss": 0.0006,
+      "step": 3409
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.367831541218638e-05,
+      "loss": 0.0071,
+      "step": 3410
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.3655913978494626e-05,
+      "loss": 0.0004,
+      "step": 3411
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.363351254480287e-05,
+      "loss": 0.0003,
+      "step": 3412
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.361111111111111e-05,
+      "loss": 0.0005,
+      "step": 3413
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.3588709677419358e-05,
+      "loss": 0.0002,
+      "step": 3414
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.35663082437276e-05,
+      "loss": 0.0002,
+      "step": 3415
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.3543906810035842e-05,
+      "loss": 0.0012,
+      "step": 3416
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 2.352150537634409e-05,
+      "loss": 0.0003,
+      "step": 3417
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.349910394265233e-05,
+      "loss": 0.0005,
+      "step": 3418
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.3476702508960574e-05,
+      "loss": 0.0002,
+      "step": 3419
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.345430107526882e-05,
+      "loss": 0.0003,
+      "step": 3420
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.343189964157706e-05,
+      "loss": 0.0004,
+      "step": 3421
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.3409498207885305e-05,
+      "loss": 0.0012,
+      "step": 3422
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 2.338709677419355e-05,
+      "loss": 0.0003,
+      "step": 3423
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.3364695340501793e-05,
+      "loss": 0.0004,
+      "step": 3424
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.3342293906810037e-05,
+      "loss": 0.0003,
+      "step": 3425
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.331989247311828e-05,
+      "loss": 0.0012,
+      "step": 3426
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.3297491039426525e-05,
+      "loss": 0.0003,
+      "step": 3427
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.327508960573477e-05,
+      "loss": 0.0004,
+      "step": 3428
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.325268817204301e-05,
+      "loss": 0.0002,
+      "step": 3429
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.3230286738351256e-05,
+      "loss": 0.0002,
+      "step": 3430
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.32078853046595e-05,
+      "loss": 0.0007,
+      "step": 3431
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.318548387096774e-05,
+      "loss": 0.0004,
+      "step": 3432
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.3163082437275988e-05,
+      "loss": 0.0002,
+      "step": 3433
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 2.314068100358423e-05,
+      "loss": 0.0004,
+      "step": 3434
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.3118279569892472e-05,
+      "loss": 0.0002,
+      "step": 3435
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.309587813620072e-05,
+      "loss": 0.0004,
+      "step": 3436
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.307347670250896e-05,
+      "loss": 0.0002,
+      "step": 3437
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.3051075268817204e-05,
+      "loss": 0.0004,
+      "step": 3438
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.302867383512545e-05,
+      "loss": 0.0005,
+      "step": 3439
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.300627240143369e-05,
+      "loss": 0.0005,
+      "step": 3440
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.2983870967741935e-05,
+      "loss": 0.0012,
+      "step": 3441
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.2961469534050183e-05,
+      "loss": 0.0005,
+      "step": 3442
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.2939068100358423e-05,
+      "loss": 0.001,
+      "step": 3443
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.2916666666666667e-05,
+      "loss": 0.0015,
+      "step": 3444
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 2.289426523297491e-05,
+      "loss": 0.0004,
+      "step": 3445
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.2871863799283155e-05,
+      "loss": 0.0003,
+      "step": 3446
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.28494623655914e-05,
+      "loss": 0.0002,
+      "step": 3447
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.2827060931899642e-05,
+      "loss": 0.0002,
+      "step": 3448
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.2804659498207886e-05,
+      "loss": 0.002,
+      "step": 3449
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.278225806451613e-05,
+      "loss": 0.0003,
+      "step": 3450
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.2759856630824374e-05,
+      "loss": 0.023,
+      "step": 3451
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.2737455197132618e-05,
+      "loss": 0.0006,
+      "step": 3452
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.271505376344086e-05,
+      "loss": 0.0003,
+      "step": 3453
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.2692652329749106e-05,
+      "loss": 0.0004,
+      "step": 3454
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.267025089605735e-05,
+      "loss": 0.0003,
+      "step": 3455
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 2.2647849462365593e-05,
+      "loss": 0.0003,
+      "step": 3456
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.2625448028673837e-05,
+      "loss": 0.0004,
+      "step": 3457
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.260304659498208e-05,
+      "loss": 0.0009,
+      "step": 3458
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.258064516129032e-05,
+      "loss": 0.0012,
+      "step": 3459
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.255824372759857e-05,
+      "loss": 0.0002,
+      "step": 3460
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.2535842293906813e-05,
+      "loss": 0.0002,
+      "step": 3461
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 2.2513440860215053e-05,
+      "loss": 0.0005,
+      "step": 3462
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.24910394265233e-05,
+      "loss": 0.0002,
+      "step": 3463
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.2468637992831544e-05,
+      "loss": 0.0009,
+      "step": 3464
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.2446236559139785e-05,
+      "loss": 0.004,
+      "step": 3465
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.2423835125448032e-05,
+      "loss": 0.0003,
+      "step": 3466
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.2401433691756272e-05,
+      "loss": 0.0006,
+      "step": 3467
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.2379032258064516e-05,
+      "loss": 0.0003,
+      "step": 3468
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.235663082437276e-05,
+      "loss": 0.0003,
+      "step": 3469
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.2334229390681004e-05,
+      "loss": 0.0004,
+      "step": 3470
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.2311827956989248e-05,
+      "loss": 0.0003,
+      "step": 3471
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.2289426523297492e-05,
+      "loss": 0.0009,
+      "step": 3472
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 2.2267025089605736e-05,
+      "loss": 0.0003,
+      "step": 3473
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.224462365591398e-05,
+      "loss": 0.0003,
+      "step": 3474
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.001,
+      "step": 3475
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.2199820788530467e-05,
+      "loss": 0.0003,
+      "step": 3476
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.217741935483871e-05,
+      "loss": 0.0003,
+      "step": 3477
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.2155017921146955e-05,
+      "loss": 0.0008,
+      "step": 3478
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.21326164874552e-05,
+      "loss": 0.0013,
+      "step": 3479
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.2110215053763443e-05,
+      "loss": 1.589,
+      "step": 3480
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.2087813620071683e-05,
+      "loss": 0.0004,
+      "step": 3481
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.206541218637993e-05,
+      "loss": 0.0004,
+      "step": 3482
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.2043010752688174e-05,
+      "loss": 0.0002,
+      "step": 3483
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 2.2020609318996415e-05,
+      "loss": 0.0004,
+      "step": 3484
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.1998207885304662e-05,
+      "loss": 0.0011,
+      "step": 3485
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.1975806451612906e-05,
+      "loss": 0.0003,
+      "step": 3486
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.1953405017921146e-05,
+      "loss": 0.0003,
+      "step": 3487
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.1931003584229394e-05,
+      "loss": 0.0002,
+      "step": 3488
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.1908602150537634e-05,
+      "loss": 0.0002,
+      "step": 3489
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.1886200716845878e-05,
+      "loss": 0.0003,
+      "step": 3490
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.1863799283154125e-05,
+      "loss": 0.0003,
+      "step": 3491
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.1841397849462366e-05,
+      "loss": 0.0005,
+      "step": 3492
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.181899641577061e-05,
+      "loss": 0.0003,
+      "step": 3493
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.1796594982078857e-05,
+      "loss": 0.0003,
+      "step": 3494
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 2.1774193548387097e-05,
+      "loss": 3.5747,
+      "step": 3495
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.175179211469534e-05,
+      "loss": 0.0008,
+      "step": 3496
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.1729390681003585e-05,
+      "loss": 0.0003,
+      "step": 3497
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.170698924731183e-05,
+      "loss": 0.0002,
+      "step": 3498
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.1684587813620073e-05,
+      "loss": 0.0003,
+      "step": 3499
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.1662186379928316e-05,
+      "loss": 0.0004,
+      "step": 3500
+    },
+    {
+      "epoch": 6.27,
+      "eval_accuracy": 0.7071428298950195,
+      "eval_loss": 1.698424220085144,
+      "eval_runtime": 73.7138,
+      "eval_samples_per_second": 3.798,
+      "eval_steps_per_second": 3.798,
+      "step": 3500
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 2.163978494623656e-05,
+      "loss": 0.0013,
+      "step": 3501
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.1617383512544804e-05,
+      "loss": 0.0003,
+      "step": 3502
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.1594982078853048e-05,
+      "loss": 0.0002,
+      "step": 3503
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.1572580645161292e-05,
+      "loss": 0.0004,
+      "step": 3504
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.1550179211469536e-05,
+      "loss": 0.0002,
+      "step": 3505
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.152777777777778e-05,
+      "loss": 0.0003,
+      "step": 3506
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.1505376344086024e-05,
+      "loss": 0.0004,
+      "step": 3507
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.1482974910394264e-05,
+      "loss": 0.0008,
+      "step": 3508
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.146057347670251e-05,
+      "loss": 0.0003,
+      "step": 3509
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.1438172043010755e-05,
+      "loss": 0.0156,
+      "step": 3510
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.1415770609318996e-05,
+      "loss": 0.0003,
+      "step": 3511
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 2.1393369175627243e-05,
+      "loss": 0.0003,
+      "step": 3512
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.1370967741935487e-05,
+      "loss": 0.0003,
+      "step": 3513
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.1348566308243727e-05,
+      "loss": 0.0004,
+      "step": 3514
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.132616487455197e-05,
+      "loss": 0.0033,
+      "step": 3515
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.1303763440860215e-05,
+      "loss": 0.0017,
+      "step": 3516
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.128136200716846e-05,
+      "loss": 0.0005,
+      "step": 3517
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.1258960573476703e-05,
+      "loss": 0.0004,
+      "step": 3518
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.1236559139784946e-05,
+      "loss": 0.0002,
+      "step": 3519
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.121415770609319e-05,
+      "loss": 0.0002,
+      "step": 3520
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.1191756272401434e-05,
+      "loss": 0.0007,
+      "step": 3521
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.1169354838709678e-05,
+      "loss": 0.0003,
+      "step": 3522
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 2.1146953405017922e-05,
+      "loss": 0.0003,
+      "step": 3523
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.1124551971326166e-05,
+      "loss": 0.0012,
+      "step": 3524
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.110215053763441e-05,
+      "loss": 0.0003,
+      "step": 3525
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.1079749103942654e-05,
+      "loss": 0.0004,
+      "step": 3526
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.1057347670250897e-05,
+      "loss": 0.0004,
+      "step": 3527
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.103494623655914e-05,
+      "loss": 0.0003,
+      "step": 3528
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.1012544802867385e-05,
+      "loss": 0.002,
+      "step": 3529
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.0990143369175626e-05,
+      "loss": 0.0003,
+      "step": 3530
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.0967741935483873e-05,
+      "loss": 0.0003,
+      "step": 3531
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.0945340501792117e-05,
+      "loss": 0.0004,
+      "step": 3532
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.0922939068100357e-05,
+      "loss": 0.0003,
+      "step": 3533
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 2.0900537634408604e-05,
+      "loss": 0.0002,
+      "step": 3534
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.0878136200716848e-05,
+      "loss": 0.0004,
+      "step": 3535
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.085573476702509e-05,
+      "loss": 0.0003,
+      "step": 3536
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.0003,
+      "step": 3537
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.0810931899641577e-05,
+      "loss": 0.0003,
+      "step": 3538
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.078853046594982e-05,
+      "loss": 0.0004,
+      "step": 3539
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 2.0766129032258068e-05,
+      "loss": 0.0002,
+      "step": 3540
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.0743727598566308e-05,
+      "loss": 0.0004,
+      "step": 3541
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.0721326164874552e-05,
+      "loss": 0.0036,
+      "step": 3542
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.06989247311828e-05,
+      "loss": 0.0003,
+      "step": 3543
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.067652329749104e-05,
+      "loss": 0.0003,
+      "step": 3544
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.0654121863799284e-05,
+      "loss": 0.0064,
+      "step": 3545
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.0631720430107527e-05,
+      "loss": 0.0003,
+      "step": 3546
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.060931899641577e-05,
+      "loss": 0.0003,
+      "step": 3547
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.0586917562724015e-05,
+      "loss": 0.0002,
+      "step": 3548
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.056451612903226e-05,
+      "loss": 0.0005,
+      "step": 3549
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.0542114695340503e-05,
+      "loss": 0.0002,
+      "step": 3550
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 2.0519713261648747e-05,
+      "loss": 1.9201,
+      "step": 3551
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.049731182795699e-05,
+      "loss": 0.0007,
+      "step": 3552
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.0474910394265234e-05,
+      "loss": 0.0005,
+      "step": 3553
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.045250896057348e-05,
+      "loss": 0.0002,
+      "step": 3554
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.0430107526881722e-05,
+      "loss": 0.0002,
+      "step": 3555
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.0407706093189966e-05,
+      "loss": 0.0026,
+      "step": 3556
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.038530465949821e-05,
+      "loss": 0.0004,
+      "step": 3557
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.0362903225806454e-05,
+      "loss": 0.0002,
+      "step": 3558
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.0340501792114698e-05,
+      "loss": 0.0004,
+      "step": 3559
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.0318100358422938e-05,
+      "loss": 0.0002,
+      "step": 3560
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.0295698924731182e-05,
+      "loss": 0.0002,
+      "step": 3561
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 2.027329749103943e-05,
+      "loss": 0.0002,
+      "step": 3562
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.025089605734767e-05,
+      "loss": 0.0003,
+      "step": 3563
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.0228494623655914e-05,
+      "loss": 0.0017,
+      "step": 3564
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.020609318996416e-05,
+      "loss": 0.0008,
+      "step": 3565
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.01836917562724e-05,
+      "loss": 0.0004,
+      "step": 3566
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.0161290322580645e-05,
+      "loss": 0.0003,
+      "step": 3567
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 2.013888888888889e-05,
+      "loss": 0.0003,
+      "step": 3568
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.0116487455197133e-05,
+      "loss": 0.0002,
+      "step": 3569
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.0094086021505377e-05,
+      "loss": 0.0002,
+      "step": 3570
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.007168458781362e-05,
+      "loss": 0.0013,
+      "step": 3571
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.0049283154121864e-05,
+      "loss": 0.0003,
+      "step": 3572
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 2.002688172043011e-05,
+      "loss": 0.0002,
+      "step": 3573
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 2.0004480286738352e-05,
+      "loss": 0.0012,
+      "step": 3574
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.9982078853046596e-05,
+      "loss": 0.0011,
+      "step": 3575
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.995967741935484e-05,
+      "loss": 0.0004,
+      "step": 3576
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.9937275985663084e-05,
+      "loss": 0.0005,
+      "step": 3577
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.9914874551971328e-05,
+      "loss": 0.0004,
+      "step": 3578
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.989247311827957e-05,
+      "loss": 0.0002,
+      "step": 3579
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.9870071684587815e-05,
+      "loss": 0.0003,
+      "step": 3580
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.984767025089606e-05,
+      "loss": 0.0011,
+      "step": 3581
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.98252688172043e-05,
+      "loss": 0.0003,
+      "step": 3582
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.9802867383512547e-05,
+      "loss": 0.0002,
+      "step": 3583
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.978046594982079e-05,
+      "loss": 0.0003,
+      "step": 3584
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.975806451612903e-05,
+      "loss": 0.0002,
+      "step": 3585
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.973566308243728e-05,
+      "loss": 0.0002,
+      "step": 3586
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.9713261648745522e-05,
+      "loss": 0.206,
+      "step": 3587
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.9690860215053763e-05,
+      "loss": 0.0004,
+      "step": 3588
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.966845878136201e-05,
+      "loss": 0.0002,
+      "step": 3589
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.964605734767025e-05,
+      "loss": 0.0003,
+      "step": 3590
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.9623655913978494e-05,
+      "loss": 0.0003,
+      "step": 3591
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.9601254480286742e-05,
+      "loss": 0.0004,
+      "step": 3592
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.9578853046594982e-05,
+      "loss": 0.01,
+      "step": 3593
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.9556451612903226e-05,
+      "loss": 0.0004,
+      "step": 3594
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.9534050179211473e-05,
+      "loss": 0.0006,
+      "step": 3595
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.9511648745519714e-05,
+      "loss": 0.592,
+      "step": 3596
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.9489247311827958e-05,
+      "loss": 0.0002,
+      "step": 3597
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.94668458781362e-05,
+      "loss": 0.0003,
+      "step": 3598
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.0004,
+      "step": 3599
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.942204301075269e-05,
+      "loss": 0.0008,
+      "step": 3600
+    },
+    {
+      "epoch": 6.45,
+      "eval_accuracy": 0.699999988079071,
+      "eval_loss": 1.7147274017333984,
+      "eval_runtime": 69.752,
+      "eval_samples_per_second": 4.014,
+      "eval_steps_per_second": 4.014,
+      "step": 3600
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.9399641577060933e-05,
+      "loss": 0.0722,
+      "step": 3601
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.9377240143369177e-05,
+      "loss": 0.0003,
+      "step": 3602
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.935483870967742e-05,
+      "loss": 0.0003,
+      "step": 3603
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.9332437275985665e-05,
+      "loss": 0.0003,
+      "step": 3604
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.931003584229391e-05,
+      "loss": 0.0006,
+      "step": 3605
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.9287634408602152e-05,
+      "loss": 3.4748,
+      "step": 3606
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.9265232974910393e-05,
+      "loss": 0.0003,
+      "step": 3607
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.924283154121864e-05,
+      "loss": 0.0006,
+      "step": 3608
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.922043010752688e-05,
+      "loss": 0.0003,
+      "step": 3609
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.9198028673835124e-05,
+      "loss": 0.0004,
+      "step": 3610
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.9175627240143372e-05,
+      "loss": 0.0004,
+      "step": 3611
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.9153225806451612e-05,
+      "loss": 0.0002,
+      "step": 3612
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.9130824372759856e-05,
+      "loss": 0.0409,
+      "step": 3613
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.9108422939068103e-05,
+      "loss": 0.0005,
+      "step": 3614
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.9086021505376344e-05,
+      "loss": 0.0003,
+      "step": 3615
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.9063620071684588e-05,
+      "loss": 0.0003,
+      "step": 3616
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.904121863799283e-05,
+      "loss": 0.0002,
+      "step": 3617
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.9018817204301075e-05,
+      "loss": 5.0392,
+      "step": 3618
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.899641577060932e-05,
+      "loss": 0.0003,
+      "step": 3619
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.8974014336917563e-05,
+      "loss": 0.0003,
+      "step": 3620
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.8951612903225807e-05,
+      "loss": 0.0003,
+      "step": 3621
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.892921146953405e-05,
+      "loss": 0.0002,
+      "step": 3622
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.8906810035842295e-05,
+      "loss": 0.0011,
+      "step": 3623
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.888440860215054e-05,
+      "loss": 0.0004,
+      "step": 3624
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.8862007168458782e-05,
+      "loss": 0.0002,
+      "step": 3625
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.8839605734767026e-05,
+      "loss": 0.0003,
+      "step": 3626
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.881720430107527e-05,
+      "loss": 0.0002,
+      "step": 3627
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.8794802867383514e-05,
+      "loss": 0.0005,
+      "step": 3628
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.8772401433691758e-05,
+      "loss": 0.0003,
+      "step": 3629
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.0004,
+      "step": 3630
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.8727598566308242e-05,
+      "loss": 0.0019,
+      "step": 3631
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.870519713261649e-05,
+      "loss": 0.0014,
+      "step": 3632
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.8682795698924733e-05,
+      "loss": 0.0003,
+      "step": 3633
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.8660394265232974e-05,
+      "loss": 0.0002,
+      "step": 3634
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.863799283154122e-05,
+      "loss": 0.0003,
+      "step": 3635
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.8615591397849465e-05,
+      "loss": 0.0005,
+      "step": 3636
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.8593189964157705e-05,
+      "loss": 0.0003,
+      "step": 3637
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.8570788530465953e-05,
+      "loss": 0.0008,
+      "step": 3638
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.8548387096774193e-05,
+      "loss": 0.0787,
+      "step": 3639
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 1.8525985663082437e-05,
+      "loss": 0.0002,
+      "step": 3640
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.8503584229390684e-05,
+      "loss": 0.0004,
+      "step": 3641
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.8481182795698925e-05,
+      "loss": 0.0003,
+      "step": 3642
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.845878136200717e-05,
+      "loss": 0.0013,
+      "step": 3643
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.8436379928315416e-05,
+      "loss": 0.0002,
+      "step": 3644
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.8413978494623656e-05,
+      "loss": 0.0009,
+      "step": 3645
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.83915770609319e-05,
+      "loss": 0.0002,
+      "step": 3646
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.8369175627240144e-05,
+      "loss": 0.0003,
+      "step": 3647
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.8346774193548388e-05,
+      "loss": 0.0002,
+      "step": 3648
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.8324372759856632e-05,
+      "loss": 0.0002,
+      "step": 3649
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.8301971326164876e-05,
+      "loss": 0.0018,
+      "step": 3650
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.827956989247312e-05,
+      "loss": 0.0214,
+      "step": 3651
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.8257168458781363e-05,
+      "loss": 0.0002,
+      "step": 3652
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.8234767025089607e-05,
+      "loss": 0.0002,
+      "step": 3653
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.821236559139785e-05,
+      "loss": 0.0002,
+      "step": 3654
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.8189964157706095e-05,
+      "loss": 0.0005,
+      "step": 3655
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.8167562724014335e-05,
+      "loss": 0.0002,
+      "step": 3656
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.8145161290322583e-05,
+      "loss": 0.0057,
+      "step": 3657
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.8122759856630827e-05,
+      "loss": 0.0003,
+      "step": 3658
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.8100358422939067e-05,
+      "loss": 0.0012,
+      "step": 3659
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.8077956989247314e-05,
+      "loss": 0.0002,
+      "step": 3660
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.8055555555555555e-05,
+      "loss": 0.0002,
+      "step": 3661
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.80331541218638e-05,
+      "loss": 0.0004,
+      "step": 3662
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.8010752688172046e-05,
+      "loss": 0.0004,
+      "step": 3663
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7988351254480286e-05,
+      "loss": 0.0003,
+      "step": 3664
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.796594982078853e-05,
+      "loss": 0.0004,
+      "step": 3665
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7943548387096777e-05,
+      "loss": 0.0006,
+      "step": 3666
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7921146953405018e-05,
+      "loss": 0.0003,
+      "step": 3667
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7898745519713262e-05,
+      "loss": 0.0003,
+      "step": 3668
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7876344086021506e-05,
+      "loss": 0.0008,
+      "step": 3669
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.785394265232975e-05,
+      "loss": 0.0003,
+      "step": 3670
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7831541218637993e-05,
+      "loss": 0.0003,
+      "step": 3671
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7809139784946237e-05,
+      "loss": 0.0002,
+      "step": 3672
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.778673835125448e-05,
+      "loss": 0.0008,
+      "step": 3673
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7764336917562725e-05,
+      "loss": 0.0002,
+      "step": 3674
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.774193548387097e-05,
+      "loss": 0.0003,
+      "step": 3675
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7719534050179213e-05,
+      "loss": 0.0002,
+      "step": 3676
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7697132616487457e-05,
+      "loss": 0.0003,
+      "step": 3677
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.76747311827957e-05,
+      "loss": 0.0006,
+      "step": 3678
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7652329749103944e-05,
+      "loss": 0.0002,
+      "step": 3679
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7629928315412185e-05,
+      "loss": 0.0002,
+      "step": 3680
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7607526881720432e-05,
+      "loss": 0.0002,
+      "step": 3681
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7585125448028676e-05,
+      "loss": 0.0003,
+      "step": 3682
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7562724014336916e-05,
+      "loss": 0.0006,
+      "step": 3683
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7540322580645164e-05,
+      "loss": 0.0002,
+      "step": 3684
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7517921146953407e-05,
+      "loss": 0.0004,
+      "step": 3685
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.7495519713261648e-05,
+      "loss": 0.0005,
+      "step": 3686
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.7473118279569895e-05,
+      "loss": 0.0003,
+      "step": 3687
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.7450716845878136e-05,
+      "loss": 0.0003,
+      "step": 3688
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.742831541218638e-05,
+      "loss": 0.0004,
+      "step": 3689
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.7405913978494627e-05,
+      "loss": 0.0005,
+      "step": 3690
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.7383512544802867e-05,
+      "loss": 0.0393,
+      "step": 3691
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.736111111111111e-05,
+      "loss": 0.0004,
+      "step": 3692
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.733870967741936e-05,
+      "loss": 0.0003,
+      "step": 3693
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.73163082437276e-05,
+      "loss": 0.0004,
+      "step": 3694
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.7293906810035843e-05,
+      "loss": 0.0005,
+      "step": 3695
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.7271505376344087e-05,
+      "loss": 0.0003,
+      "step": 3696
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.724910394265233e-05,
+      "loss": 0.0004,
+      "step": 3697
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.7226702508960574e-05,
+      "loss": 0.0006,
+      "step": 3698
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.7204301075268818e-05,
+      "loss": 0.0006,
+      "step": 3699
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.7181899641577062e-05,
+      "loss": 0.0004,
+      "step": 3700
+    },
+    {
+      "epoch": 6.63,
+      "eval_accuracy": 0.7142857313156128,
+      "eval_loss": 1.7310985326766968,
+      "eval_runtime": 73.5722,
+      "eval_samples_per_second": 3.806,
+      "eval_steps_per_second": 3.806,
+      "step": 3700
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.7159498207885306e-05,
+      "loss": 0.002,
+      "step": 3701
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.7137096774193546e-05,
+      "loss": 2.3204,
+      "step": 3702
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.7114695340501794e-05,
+      "loss": 0.0006,
+      "step": 3703
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.7092293906810037e-05,
+      "loss": 0.0007,
+      "step": 3704
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.7069892473118278e-05,
+      "loss": 0.0002,
+      "step": 3705
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.7047491039426525e-05,
+      "loss": 0.0003,
+      "step": 3706
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.702508960573477e-05,
+      "loss": 0.0002,
+      "step": 3707
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.700268817204301e-05,
+      "loss": 0.0002,
+      "step": 3708
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6980286738351257e-05,
+      "loss": 0.0003,
+      "step": 3709
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.6957885304659497e-05,
+      "loss": 0.5109,
+      "step": 3710
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.693548387096774e-05,
+      "loss": 0.0004,
+      "step": 3711
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.691308243727599e-05,
+      "loss": 4.1212,
+      "step": 3712
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.689068100358423e-05,
+      "loss": 0.0008,
+      "step": 3713
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6868279569892473e-05,
+      "loss": 0.0005,
+      "step": 3714
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.684587813620072e-05,
+      "loss": 0.0003,
+      "step": 3715
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.682347670250896e-05,
+      "loss": 0.0002,
+      "step": 3716
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6801075268817204e-05,
+      "loss": 0.0003,
+      "step": 3717
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6778673835125448e-05,
+      "loss": 0.0004,
+      "step": 3718
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6756272401433692e-05,
+      "loss": 0.0004,
+      "step": 3719
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6733870967741936e-05,
+      "loss": 0.0003,
+      "step": 3720
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.671146953405018e-05,
+      "loss": 0.0007,
+      "step": 3721
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6689068100358424e-05,
+      "loss": 0.0339,
+      "step": 3722
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.0006,
+      "step": 3723
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.664426523297491e-05,
+      "loss": 0.0003,
+      "step": 3724
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6621863799283155e-05,
+      "loss": 0.0004,
+      "step": 3725
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.65994623655914e-05,
+      "loss": 0.0005,
+      "step": 3726
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6577060931899643e-05,
+      "loss": 0.0003,
+      "step": 3727
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6554659498207887e-05,
+      "loss": 0.0186,
+      "step": 3728
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.653225806451613e-05,
+      "loss": 0.0002,
+      "step": 3729
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.6509856630824375e-05,
+      "loss": 0.0003,
+      "step": 3730
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.648745519713262e-05,
+      "loss": 0.0005,
+      "step": 3731
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.646505376344086e-05,
+      "loss": 0.0031,
+      "step": 3732
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.6442652329749106e-05,
+      "loss": 0.0007,
+      "step": 3733
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.642025089605735e-05,
+      "loss": 0.0006,
+      "step": 3734
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.639784946236559e-05,
+      "loss": 0.0003,
+      "step": 3735
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6375448028673838e-05,
+      "loss": 0.0003,
+      "step": 3736
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.635304659498208e-05,
+      "loss": 0.001,
+      "step": 3737
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6330645161290322e-05,
+      "loss": 0.0002,
+      "step": 3738
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.630824372759857e-05,
+      "loss": 0.0002,
+      "step": 3739
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.628584229390681e-05,
+      "loss": 0.0002,
+      "step": 3740
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6263440860215054e-05,
+      "loss": 0.0003,
+      "step": 3741
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.62410394265233e-05,
+      "loss": 0.0005,
+      "step": 3742
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.621863799283154e-05,
+      "loss": 0.0903,
+      "step": 3743
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6196236559139785e-05,
+      "loss": 0.0003,
+      "step": 3744
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6173835125448032e-05,
+      "loss": 0.0012,
+      "step": 3745
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6151433691756273e-05,
+      "loss": 0.0004,
+      "step": 3746
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6129032258064517e-05,
+      "loss": 0.0006,
+      "step": 3747
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.610663082437276e-05,
+      "loss": 0.0002,
+      "step": 3748
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6084229390681005e-05,
+      "loss": 0.001,
+      "step": 3749
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.606182795698925e-05,
+      "loss": 0.0002,
+      "step": 3750
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.603942652329749e-05,
+      "loss": 0.0033,
+      "step": 3751
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6017025089605736e-05,
+      "loss": 0.0002,
+      "step": 3752
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.599462365591398e-05,
+      "loss": 0.0002,
+      "step": 3753
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.597222222222222e-05,
+      "loss": 0.0003,
+      "step": 3754
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.5949820788530468e-05,
+      "loss": 0.0003,
+      "step": 3755
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.592741935483871e-05,
+      "loss": 0.0005,
+      "step": 3756
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.5905017921146952e-05,
+      "loss": 0.0056,
+      "step": 3757
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.58826164874552e-05,
+      "loss": 0.0003,
+      "step": 3758
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.586021505376344e-05,
+      "loss": 0.0003,
+      "step": 3759
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.5837813620071684e-05,
+      "loss": 0.0002,
+      "step": 3760
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.581541218637993e-05,
+      "loss": 0.0003,
+      "step": 3761
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.579301075268817e-05,
+      "loss": 0.0002,
+      "step": 3762
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.5770609318996415e-05,
+      "loss": 0.0003,
+      "step": 3763
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5748207885304662e-05,
+      "loss": 0.0003,
+      "step": 3764
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5725806451612903e-05,
+      "loss": 0.0002,
+      "step": 3765
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5703405017921147e-05,
+      "loss": 0.0002,
+      "step": 3766
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.568100358422939e-05,
+      "loss": 0.0002,
+      "step": 3767
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.5658602150537635e-05,
+      "loss": 0.0003,
+      "step": 3768
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 1.563620071684588e-05,
+      "loss": 0.0002,
+      "step": 3769
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5613799283154122e-05,
+      "loss": 0.0003,
+      "step": 3770
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5591397849462366e-05,
+      "loss": 0.0003,
+      "step": 3771
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.556899641577061e-05,
+      "loss": 0.0003,
+      "step": 3772
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5546594982078854e-05,
+      "loss": 0.0002,
+      "step": 3773
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.5524193548387098e-05,
+      "loss": 0.0005,
+      "step": 3774
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.550179211469534e-05,
+      "loss": 0.0002,
+      "step": 3775
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.5479390681003585e-05,
+      "loss": 0.0006,
+      "step": 3776
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.545698924731183e-05,
+      "loss": 0.0003,
+      "step": 3777
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.5434587813620073e-05,
+      "loss": 0.0003,
+      "step": 3778
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.5412186379928317e-05,
+      "loss": 0.0002,
+      "step": 3779
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.538978494623656e-05,
+      "loss": 0.0002,
+      "step": 3780
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.53673835125448e-05,
+      "loss": 0.0002,
+      "step": 3781
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.534498207885305e-05,
+      "loss": 0.0004,
+      "step": 3782
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.5322580645161292e-05,
+      "loss": 0.0006,
+      "step": 3783
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.5300179211469533e-05,
+      "loss": 0.0002,
+      "step": 3784
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.527777777777778e-05,
+      "loss": 0.0002,
+      "step": 3785
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.5255376344086022e-05,
+      "loss": 0.0021,
+      "step": 3786
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5232974910394265e-05,
+      "loss": 0.0002,
+      "step": 3787
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5210573476702512e-05,
+      "loss": 0.0003,
+      "step": 3788
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5188172043010754e-05,
+      "loss": 0.0002,
+      "step": 3789
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5165770609318996e-05,
+      "loss": 0.0003,
+      "step": 3790
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.5143369175627242e-05,
+      "loss": 0.0003,
+      "step": 3791
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5120967741935486e-05,
+      "loss": 0.0003,
+      "step": 3792
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5098566308243728e-05,
+      "loss": 0.0008,
+      "step": 3793
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.507616487455197e-05,
+      "loss": 0.0002,
+      "step": 3794
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5053763440860215e-05,
+      "loss": 0.0017,
+      "step": 3795
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.503136200716846e-05,
+      "loss": 0.0003,
+      "step": 3796
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.5008960573476701e-05,
+      "loss": 0.0045,
+      "step": 3797
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4986559139784947e-05,
+      "loss": 0.0006,
+      "step": 3798
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4964157706093191e-05,
+      "loss": 0.0003,
+      "step": 3799
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4941756272401433e-05,
+      "loss": 0.0002,
+      "step": 3800
+    },
+    {
+      "epoch": 6.81,
+      "eval_accuracy": 0.7071428298950195,
+      "eval_loss": 1.784097671508789,
+      "eval_runtime": 73.6953,
+      "eval_samples_per_second": 3.799,
+      "eval_steps_per_second": 3.799,
+      "step": 3800
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.4919354838709679e-05,
+      "loss": 0.0002,
+      "step": 3801
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.489695340501792e-05,
+      "loss": 0.0002,
+      "step": 3802
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.4874551971326165e-05,
+      "loss": 0.0005,
+      "step": 3803
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.485215053763441e-05,
+      "loss": 0.0002,
+      "step": 3804
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.4829749103942652e-05,
+      "loss": 0.0002,
+      "step": 3805
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.4807347670250896e-05,
+      "loss": 0.0002,
+      "step": 3806
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.4784946236559142e-05,
+      "loss": 0.0003,
+      "step": 3807
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.4762544802867384e-05,
+      "loss": 0.2326,
+      "step": 3808
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.4740143369175626e-05,
+      "loss": 0.0002,
+      "step": 3809
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.4717741935483872e-05,
+      "loss": 0.0002,
+      "step": 3810
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.4695340501792116e-05,
+      "loss": 0.0003,
+      "step": 3811
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.4672939068100358e-05,
+      "loss": 0.0002,
+      "step": 3812
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.4650537634408603e-05,
+      "loss": 0.0004,
+      "step": 3813
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4628136200716847e-05,
+      "loss": 0.0003,
+      "step": 3814
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.460573476702509e-05,
+      "loss": 0.0006,
+      "step": 3815
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4583333333333335e-05,
+      "loss": 0.0002,
+      "step": 3816
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4560931899641577e-05,
+      "loss": 0.0003,
+      "step": 3817
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4538530465949821e-05,
+      "loss": 4.5471,
+      "step": 3818
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.4516129032258066e-05,
+      "loss": 0.1715,
+      "step": 3819
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4493727598566309e-05,
+      "loss": 0.0003,
+      "step": 3820
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4471326164874553e-05,
+      "loss": 0.0003,
+      "step": 3821
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4448924731182798e-05,
+      "loss": 0.0003,
+      "step": 3822
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.442652329749104e-05,
+      "loss": 0.0002,
+      "step": 3823
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4404121863799282e-05,
+      "loss": 0.0003,
+      "step": 3824
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.4381720430107528e-05,
+      "loss": 0.0004,
+      "step": 3825
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.4359318996415772e-05,
+      "loss": 0.0003,
+      "step": 3826
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.4336917562724014e-05,
+      "loss": 0.0003,
+      "step": 3827
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.431451612903226e-05,
+      "loss": 0.0002,
+      "step": 3828
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.4292114695340503e-05,
+      "loss": 0.0002,
+      "step": 3829
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.4269713261648746e-05,
+      "loss": 0.0002,
+      "step": 3830
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4247311827956991e-05,
+      "loss": 0.0002,
+      "step": 3831
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4224910394265233e-05,
+      "loss": 0.0004,
+      "step": 3832
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4202508960573477e-05,
+      "loss": 0.0003,
+      "step": 3833
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4180107526881723e-05,
+      "loss": 0.0002,
+      "step": 3834
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4157706093189965e-05,
+      "loss": 0.0002,
+      "step": 3835
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.4135304659498209e-05,
+      "loss": 0.0007,
+      "step": 3836
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4112903225806454e-05,
+      "loss": 0.0002,
+      "step": 3837
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4090501792114696e-05,
+      "loss": 0.0002,
+      "step": 3838
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4068100358422939e-05,
+      "loss": 0.0002,
+      "step": 3839
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4045698924731183e-05,
+      "loss": 0.0013,
+      "step": 3840
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.4023297491039428e-05,
+      "loss": 0.0002,
+      "step": 3841
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.400089605734767e-05,
+      "loss": 0.0002,
+      "step": 3842
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.3978494623655914e-05,
+      "loss": 0.0005,
+      "step": 3843
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.395609318996416e-05,
+      "loss": 0.0006,
+      "step": 3844
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.3933691756272402e-05,
+      "loss": 0.0042,
+      "step": 3845
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.3911290322580644e-05,
+      "loss": 0.0002,
+      "step": 3846
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.1544,
+      "step": 3847
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.3866487455197133e-05,
+      "loss": 0.0002,
+      "step": 3848
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.3844086021505376e-05,
+      "loss": 0.001,
+      "step": 3849
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.3821684587813621e-05,
+      "loss": 0.0002,
+      "step": 3850
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.3799283154121865e-05,
+      "loss": 0.0002,
+      "step": 3851
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.3776881720430107e-05,
+      "loss": 0.0003,
+      "step": 3852
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.3754480286738353e-05,
+      "loss": 0.0004,
+      "step": 3853
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.3732078853046595e-05,
+      "loss": 0.0019,
+      "step": 3854
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.3709677419354839e-05,
+      "loss": 0.0002,
+      "step": 3855
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.3687275985663084e-05,
+      "loss": 0.0002,
+      "step": 3856
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.3664874551971326e-05,
+      "loss": 2.2071,
+      "step": 3857
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.364247311827957e-05,
+      "loss": 0.0002,
+      "step": 3858
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.3620071684587816e-05,
+      "loss": 0.0002,
+      "step": 3859
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.3597670250896058e-05,
+      "loss": 0.0005,
+      "step": 3860
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.35752688172043e-05,
+      "loss": 0.0002,
+      "step": 3861
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.3552867383512546e-05,
+      "loss": 0.0019,
+      "step": 3862
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.353046594982079e-05,
+      "loss": 0.0005,
+      "step": 3863
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.3508064516129032e-05,
+      "loss": 0.0022,
+      "step": 3864
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.3485663082437277e-05,
+      "loss": 0.0002,
+      "step": 3865
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.3463261648745521e-05,
+      "loss": 0.0003,
+      "step": 3866
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.3440860215053763e-05,
+      "loss": 0.0002,
+      "step": 3867
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.3418458781362009e-05,
+      "loss": 0.0002,
+      "step": 3868
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.3396057347670251e-05,
+      "loss": 0.0002,
+      "step": 3869
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.3373655913978495e-05,
+      "loss": 0.0003,
+      "step": 3870
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.335125448028674e-05,
+      "loss": 0.0003,
+      "step": 3871
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.3328853046594983e-05,
+      "loss": 0.0003,
+      "step": 3872
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.3306451612903225e-05,
+      "loss": 0.0002,
+      "step": 3873
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.3284050179211472e-05,
+      "loss": 0.0004,
+      "step": 3874
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.3261648745519714e-05,
+      "loss": 0.0002,
+      "step": 3875
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.3239247311827957e-05,
+      "loss": 0.0001,
+      "step": 3876
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.3216845878136202e-05,
+      "loss": 0.0003,
+      "step": 3877
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.3194444444444446e-05,
+      "loss": 0.0002,
+      "step": 3878
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.3172043010752688e-05,
+      "loss": 0.0003,
+      "step": 3879
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.3149641577060934e-05,
+      "loss": 0.6223,
+      "step": 3880
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.3127240143369176e-05,
+      "loss": 0.0004,
+      "step": 3881
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.310483870967742e-05,
+      "loss": 0.0002,
+      "step": 3882
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.3082437275985665e-05,
+      "loss": 0.0003,
+      "step": 3883
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.3060035842293907e-05,
+      "loss": 0.0002,
+      "step": 3884
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.3037634408602151e-05,
+      "loss": 0.0002,
+      "step": 3885
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.3015232974910393e-05,
+      "loss": 0.0003,
+      "step": 3886
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.2992831541218639e-05,
+      "loss": 0.0003,
+      "step": 3887
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.2970430107526881e-05,
+      "loss": 0.0003,
+      "step": 3888
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.2948028673835125e-05,
+      "loss": 0.0006,
+      "step": 3889
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.292562724014337e-05,
+      "loss": 0.0002,
+      "step": 3890
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.2903225806451613e-05,
+      "loss": 0.0002,
+      "step": 3891
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.2880824372759857e-05,
+      "loss": 0.0003,
+      "step": 3892
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.2858422939068102e-05,
+      "loss": 0.0035,
+      "step": 3893
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.2836021505376344e-05,
+      "loss": 0.0003,
+      "step": 3894
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.2813620071684587e-05,
+      "loss": 0.0002,
+      "step": 3895
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.2791218637992832e-05,
+      "loss": 1.1427,
+      "step": 3896
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.2768817204301076e-05,
+      "loss": 0.0002,
+      "step": 3897
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.2746415770609318e-05,
+      "loss": 0.0003,
+      "step": 3898
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.2724014336917564e-05,
+      "loss": 0.0002,
+      "step": 3899
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.2701612903225808e-05,
+      "loss": 0.0004,
+      "step": 3900
+    },
+    {
+      "epoch": 6.99,
+      "eval_accuracy": 0.7142857313156128,
+      "eval_loss": 1.7812868356704712,
+      "eval_runtime": 73.7369,
+      "eval_samples_per_second": 3.797,
+      "eval_steps_per_second": 3.797,
+      "step": 3900
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.267921146953405e-05,
+      "loss": 0.0011,
+      "step": 3901
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.2656810035842295e-05,
+      "loss": 0.0003,
+      "step": 3902
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.2634408602150537e-05,
+      "loss": 0.0003,
+      "step": 3903
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.2612007168458781e-05,
+      "loss": 0.0002,
+      "step": 3904
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.2589605734767027e-05,
+      "loss": 0.0003,
+      "step": 3905
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.2567204301075269e-05,
+      "loss": 0.0005,
+      "step": 3906
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.2544802867383513e-05,
+      "loss": 0.0002,
+      "step": 3907
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.2522401433691758e-05,
+      "loss": 0.0004,
+      "step": 3908
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.25e-05,
+      "loss": 0.0005,
+      "step": 3909
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.2477598566308244e-05,
+      "loss": 0.0004,
+      "step": 3910
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.2455197132616488e-05,
+      "loss": 0.0003,
+      "step": 3911
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.2432795698924732e-05,
+      "loss": 0.0005,
+      "step": 3912
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.2410394265232976e-05,
+      "loss": 0.0003,
+      "step": 3913
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.2387992831541218e-05,
+      "loss": 0.0002,
+      "step": 3914
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.2365591397849464e-05,
+      "loss": 0.0003,
+      "step": 3915
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.2343189964157708e-05,
+      "loss": 0.0003,
+      "step": 3916
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.232078853046595e-05,
+      "loss": 0.0002,
+      "step": 3917
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.2298387096774194e-05,
+      "loss": 0.0004,
+      "step": 3918
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.227598566308244e-05,
+      "loss": 0.0002,
+      "step": 3919
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2253584229390681e-05,
+      "loss": 0.0003,
+      "step": 3920
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2231182795698925e-05,
+      "loss": 0.0002,
+      "step": 3921
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2208781362007169e-05,
+      "loss": 0.0004,
+      "step": 3922
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2186379928315413e-05,
+      "loss": 0.0005,
+      "step": 3923
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2163978494623657e-05,
+      "loss": 0.0006,
+      "step": 3924
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.2141577060931899e-05,
+      "loss": 0.0002,
+      "step": 3925
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2119175627240145e-05,
+      "loss": 0.0006,
+      "step": 3926
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2096774193548388e-05,
+      "loss": 0.0004,
+      "step": 3927
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.207437275985663e-05,
+      "loss": 0.0002,
+      "step": 3928
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2051971326164874e-05,
+      "loss": 0.0002,
+      "step": 3929
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.202956989247312e-05,
+      "loss": 0.0002,
+      "step": 3930
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.2007168458781362e-05,
+      "loss": 0.0004,
+      "step": 3931
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1984767025089606e-05,
+      "loss": 0.0002,
+      "step": 3932
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.196236559139785e-05,
+      "loss": 0.0002,
+      "step": 3933
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1939964157706094e-05,
+      "loss": 0.0002,
+      "step": 3934
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1917562724014338e-05,
+      "loss": 0.0005,
+      "step": 3935
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.1895161290322582e-05,
+      "loss": 0.0004,
+      "step": 3936
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.1872759856630825e-05,
+      "loss": 0.0002,
+      "step": 3937
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.185035842293907e-05,
+      "loss": 0.0004,
+      "step": 3938
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.1827956989247313e-05,
+      "loss": 0.0002,
+      "step": 3939
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.1805555555555555e-05,
+      "loss": 0.0004,
+      "step": 3940
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.17831541218638e-05,
+      "loss": 0.0003,
+      "step": 3941
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.1760752688172045e-05,
+      "loss": 4.8048,
+      "step": 3942
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1738351254480287e-05,
+      "loss": 0.0004,
+      "step": 3943
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.171594982078853e-05,
+      "loss": 0.0002,
+      "step": 3944
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1693548387096775e-05,
+      "loss": 0.0003,
+      "step": 3945
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1671146953405018e-05,
+      "loss": 0.0002,
+      "step": 3946
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.1648745519713262e-05,
+      "loss": 0.0003,
+      "step": 3947
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1626344086021504e-05,
+      "loss": 0.0003,
+      "step": 3948
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.160394265232975e-05,
+      "loss": 0.0002,
+      "step": 3949
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1581541218637994e-05,
+      "loss": 0.0002,
+      "step": 3950
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1559139784946236e-05,
+      "loss": 0.0003,
+      "step": 3951
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.153673835125448e-05,
+      "loss": 0.0003,
+      "step": 3952
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.1514336917562725e-05,
+      "loss": 0.0002,
+      "step": 3953
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1491935483870968e-05,
+      "loss": 0.0002,
+      "step": 3954
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1469534050179212e-05,
+      "loss": 0.0081,
+      "step": 3955
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1447132616487455e-05,
+      "loss": 0.0002,
+      "step": 3956
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.14247311827957e-05,
+      "loss": 0.0003,
+      "step": 3957
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1402329749103943e-05,
+      "loss": 0.0084,
+      "step": 3958
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.1379928315412187e-05,
+      "loss": 0.0004,
+      "step": 3959
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.135752688172043e-05,
+      "loss": 0.8575,
+      "step": 3960
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1335125448028675e-05,
+      "loss": 0.0003,
+      "step": 3961
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1312724014336919e-05,
+      "loss": 0.0021,
+      "step": 3962
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.129032258064516e-05,
+      "loss": 0.0006,
+      "step": 3963
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.1267921146953406e-05,
+      "loss": 0.0002,
+      "step": 3964
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.124551971326165e-05,
+      "loss": 0.0003,
+      "step": 3965
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1223118279569892e-05,
+      "loss": 0.0007,
+      "step": 3966
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1200716845878136e-05,
+      "loss": 0.0003,
+      "step": 3967
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.117831541218638e-05,
+      "loss": 0.0003,
+      "step": 3968
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1155913978494624e-05,
+      "loss": 0.0002,
+      "step": 3969
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.1133512544802868e-05,
+      "loss": 0.0013,
+      "step": 3970
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.0005,
+      "step": 3971
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1088709677419356e-05,
+      "loss": 0.0002,
+      "step": 3972
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.10663082437276e-05,
+      "loss": 0.0003,
+      "step": 3973
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1043906810035842e-05,
+      "loss": 0.0294,
+      "step": 3974
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.1021505376344087e-05,
+      "loss": 0.0004,
+      "step": 3975
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.0999103942652331e-05,
+      "loss": 0.0002,
+      "step": 3976
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.0976702508960573e-05,
+      "loss": 0.0002,
+      "step": 3977
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.0954301075268817e-05,
+      "loss": 0.0012,
+      "step": 3978
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.0931899641577063e-05,
+      "loss": 0.0003,
+      "step": 3979
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.0909498207885305e-05,
+      "loss": 0.0002,
+      "step": 3980
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.0887096774193549e-05,
+      "loss": 0.0005,
+      "step": 3981
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.0864695340501792e-05,
+      "loss": 5.137,
+      "step": 3982
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.0842293906810036e-05,
+      "loss": 0.0003,
+      "step": 3983
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.081989247311828e-05,
+      "loss": 0.0003,
+      "step": 3984
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.0797491039426524e-05,
+      "loss": 0.0012,
+      "step": 3985
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.0775089605734768e-05,
+      "loss": 0.0003,
+      "step": 3986
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.0752688172043012e-05,
+      "loss": 0.0003,
+      "step": 3987
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.0730286738351256e-05,
+      "loss": 0.0011,
+      "step": 3988
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.0707885304659498e-05,
+      "loss": 0.0002,
+      "step": 3989
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.0685483870967743e-05,
+      "loss": 0.0002,
+      "step": 3990
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.0663082437275986e-05,
+      "loss": 0.0002,
+      "step": 3991
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.064068100358423e-05,
+      "loss": 0.0002,
+      "step": 3992
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.0618279569892473e-05,
+      "loss": 0.0002,
+      "step": 3993
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.0595878136200717e-05,
+      "loss": 0.0003,
+      "step": 3994
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.0573476702508961e-05,
+      "loss": 0.0002,
+      "step": 3995
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.0551075268817205e-05,
+      "loss": 0.0013,
+      "step": 3996
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.0528673835125449e-05,
+      "loss": 0.0004,
+      "step": 3997
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.0506272401433693e-05,
+      "loss": 0.0005,
+      "step": 3998
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.0483870967741936e-05,
+      "loss": 0.0003,
+      "step": 3999
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.0461469534050179e-05,
+      "loss": 0.0006,
+      "step": 4000
+    },
+    {
+      "epoch": 7.17,
+      "eval_accuracy": 0.7107142806053162,
+      "eval_loss": 1.6864734888076782,
+      "eval_runtime": 73.7775,
+      "eval_samples_per_second": 3.795,
+      "eval_steps_per_second": 3.795,
+      "step": 4000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.0439068100358424e-05,
+      "loss": 0.0003,
+      "step": 4001
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.0416666666666668e-05,
+      "loss": 0.0008,
+      "step": 4002
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.039426523297491e-05,
+      "loss": 0.0003,
+      "step": 4003
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.0371863799283154e-05,
+      "loss": 0.0002,
+      "step": 4004
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.03494623655914e-05,
+      "loss": 0.001,
+      "step": 4005
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.0327060931899642e-05,
+      "loss": 0.0009,
+      "step": 4006
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.0304659498207886e-05,
+      "loss": 0.0005,
+      "step": 4007
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.028225806451613e-05,
+      "loss": 0.0002,
+      "step": 4008
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.0259856630824373e-05,
+      "loss": 0.0009,
+      "step": 4009
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.0237455197132617e-05,
+      "loss": 0.0002,
+      "step": 4010
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.0215053763440861e-05,
+      "loss": 0.0005,
+      "step": 4011
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.0192652329749105e-05,
+      "loss": 0.0004,
+      "step": 4012
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.0170250896057349e-05,
+      "loss": 0.0066,
+      "step": 4013
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.0147849462365591e-05,
+      "loss": 0.0002,
+      "step": 4014
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.0125448028673835e-05,
+      "loss": 0.0116,
+      "step": 4015
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.010304659498208e-05,
+      "loss": 0.0002,
+      "step": 4016
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.0080645161290323e-05,
+      "loss": 0.0004,
+      "step": 4017
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.0058243727598566e-05,
+      "loss": 0.0008,
+      "step": 4018
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.003584229390681e-05,
+      "loss": 0.0003,
+      "step": 4019
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.0013440860215054e-05,
+      "loss": 0.0002,
+      "step": 4020
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.991039426523298e-06,
+      "loss": 0.0003,
+      "step": 4021
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.968637992831542e-06,
+      "loss": 0.0003,
+      "step": 4022
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.946236559139786e-06,
+      "loss": 0.0002,
+      "step": 4023
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.92383512544803e-06,
+      "loss": 0.0011,
+      "step": 4024
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.901433691756273e-06,
+      "loss": 0.0002,
+      "step": 4025
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.879032258064516e-06,
+      "loss": 0.0003,
+      "step": 4026
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.856630824372761e-06,
+      "loss": 0.0002,
+      "step": 4027
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.834229390681005e-06,
+      "loss": 0.0003,
+      "step": 4028
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.811827956989247e-06,
+      "loss": 0.0025,
+      "step": 4029
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.789426523297491e-06,
+      "loss": 0.0004,
+      "step": 4030
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.767025089605737e-06,
+      "loss": 0.0003,
+      "step": 4031
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.744623655913979e-06,
+      "loss": 0.0002,
+      "step": 4032
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 0.0003,
+      "step": 4033
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.699820788530467e-06,
+      "loss": 0.0003,
+      "step": 4034
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.67741935483871e-06,
+      "loss": 0.0067,
+      "step": 4035
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.655017921146954e-06,
+      "loss": 0.0002,
+      "step": 4036
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.632616487455196e-06,
+      "loss": 0.0008,
+      "step": 4037
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.61021505376344e-06,
+      "loss": 0.0002,
+      "step": 4038
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.587813620071686e-06,
+      "loss": 0.0003,
+      "step": 4039
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.565412186379928e-06,
+      "loss": 0.0002,
+      "step": 4040
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.543010752688172e-06,
+      "loss": 0.0003,
+      "step": 4041
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.520609318996416e-06,
+      "loss": 0.0018,
+      "step": 4042
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.49820788530466e-06,
+      "loss": 0.0002,
+      "step": 4043
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.475806451612903e-06,
+      "loss": 0.0002,
+      "step": 4044
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.453405017921147e-06,
+      "loss": 0.0002,
+      "step": 4045
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.431003584229391e-06,
+      "loss": 0.0003,
+      "step": 4046
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.408602150537635e-06,
+      "loss": 0.0003,
+      "step": 4047
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.386200716845879e-06,
+      "loss": 0.0008,
+      "step": 4048
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.363799283154121e-06,
+      "loss": 0.0002,
+      "step": 4049
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.341397849462367e-06,
+      "loss": 0.0004,
+      "step": 4050
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.31899641577061e-06,
+      "loss": 0.0002,
+      "step": 4051
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.296594982078853e-06,
+      "loss": 0.0003,
+      "step": 4052
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.274193548387097e-06,
+      "loss": 0.0007,
+      "step": 4053
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.251792114695342e-06,
+      "loss": 0.0003,
+      "step": 4054
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.229390681003584e-06,
+      "loss": 0.0002,
+      "step": 4055
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.206989247311828e-06,
+      "loss": 0.0003,
+      "step": 4056
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.184587813620072e-06,
+      "loss": 0.0017,
+      "step": 4057
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.162186379928316e-06,
+      "loss": 0.0003,
+      "step": 4058
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.13978494623656e-06,
+      "loss": 0.0002,
+      "step": 4059
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.117383512544804e-06,
+      "loss": 0.0002,
+      "step": 4060
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.094982078853047e-06,
+      "loss": 0.0003,
+      "step": 4061
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.072580645161291e-06,
+      "loss": 0.0002,
+      "step": 4062
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.050179211469534e-06,
+      "loss": 0.0002,
+      "step": 4063
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.027777777777777e-06,
+      "loss": 0.0003,
+      "step": 4064
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.005376344086023e-06,
+      "loss": 0.0002,
+      "step": 4065
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.982974910394265e-06,
+      "loss": 0.0002,
+      "step": 4066
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.960573476702509e-06,
+      "loss": 0.0004,
+      "step": 4067
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.938172043010753e-06,
+      "loss": 0.0004,
+      "step": 4068
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.915770609318997e-06,
+      "loss": 0.0002,
+      "step": 4069
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 8.89336917562724e-06,
+      "loss": 0.0003,
+      "step": 4070
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.870967741935484e-06,
+      "loss": 0.0002,
+      "step": 4071
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.848566308243728e-06,
+      "loss": 0.0083,
+      "step": 4072
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.826164874551972e-06,
+      "loss": 0.0004,
+      "step": 4073
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.803763440860216e-06,
+      "loss": 0.0002,
+      "step": 4074
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.781362007168458e-06,
+      "loss": 0.0002,
+      "step": 4075
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 8.758960573476704e-06,
+      "loss": 0.0009,
+      "step": 4076
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.736559139784948e-06,
+      "loss": 0.0002,
+      "step": 4077
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.71415770609319e-06,
+      "loss": 0.0004,
+      "step": 4078
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.691756272401434e-06,
+      "loss": 0.0002,
+      "step": 4079
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.66935483870968e-06,
+      "loss": 0.0002,
+      "step": 4080
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 8.646953405017921e-06,
+      "loss": 0.0003,
+      "step": 4081
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.624551971326165e-06,
+      "loss": 0.003,
+      "step": 4082
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.602150537634409e-06,
+      "loss": 0.0002,
+      "step": 4083
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.579749103942653e-06,
+      "loss": 0.0003,
+      "step": 4084
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.557347670250897e-06,
+      "loss": 0.0002,
+      "step": 4085
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.534946236559139e-06,
+      "loss": 0.0003,
+      "step": 4086
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 8.512544802867385e-06,
+      "loss": 0.0008,
+      "step": 4087
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.490143369175628e-06,
+      "loss": 0.0006,
+      "step": 4088
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.46774193548387e-06,
+      "loss": 0.0007,
+      "step": 4089
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.445340501792114e-06,
+      "loss": 0.0003,
+      "step": 4090
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.42293906810036e-06,
+      "loss": 0.0003,
+      "step": 4091
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 8.400537634408602e-06,
+      "loss": 0.0002,
+      "step": 4092
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.378136200716846e-06,
+      "loss": 0.0004,
+      "step": 4093
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.35573476702509e-06,
+      "loss": 0.0002,
+      "step": 4094
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.0002,
+      "step": 4095
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.310931899641578e-06,
+      "loss": 0.0002,
+      "step": 4096
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.288530465949821e-06,
+      "loss": 0.0004,
+      "step": 4097
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 8.266129032258065e-06,
+      "loss": 0.0003,
+      "step": 4098
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.24372759856631e-06,
+      "loss": 0.0002,
+      "step": 4099
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.221326164874553e-06,
+      "loss": 0.0006,
+      "step": 4100
+    },
+    {
+      "epoch": 7.35,
+      "eval_accuracy": 0.7142857313156128,
+      "eval_loss": 1.6826765537261963,
+      "eval_runtime": 69.6668,
+      "eval_samples_per_second": 4.019,
+      "eval_steps_per_second": 4.019,
+      "step": 4100
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.198924731182795e-06,
+      "loss": 0.0004,
+      "step": 4101
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.17652329749104e-06,
+      "loss": 0.0002,
+      "step": 4102
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.154121863799285e-06,
+      "loss": 0.0003,
+      "step": 4103
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 8.131720430107527e-06,
+      "loss": 0.0002,
+      "step": 4104
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.10931899641577e-06,
+      "loss": 0.0003,
+      "step": 4105
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.086917562724016e-06,
+      "loss": 0.0002,
+      "step": 4106
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.064516129032258e-06,
+      "loss": 0.0002,
+      "step": 4107
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.042114695340502e-06,
+      "loss": 0.0001,
+      "step": 4108
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 8.019713261648744e-06,
+      "loss": 0.0004,
+      "step": 4109
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.99731182795699e-06,
+      "loss": 0.0002,
+      "step": 4110
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.974910394265234e-06,
+      "loss": 0.0003,
+      "step": 4111
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.952508960573476e-06,
+      "loss": 0.001,
+      "step": 4112
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.93010752688172e-06,
+      "loss": 0.0003,
+      "step": 4113
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.907706093189965e-06,
+      "loss": 0.0002,
+      "step": 4114
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 7.885304659498208e-06,
+      "loss": 0.0002,
+      "step": 4115
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.862903225806451e-06,
+      "loss": 0.0002,
+      "step": 4116
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.840501792114695e-06,
+      "loss": 0.0008,
+      "step": 4117
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.81810035842294e-06,
+      "loss": 0.0002,
+      "step": 4118
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.795698924731183e-06,
+      "loss": 0.0004,
+      "step": 4119
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 7.773297491039427e-06,
+      "loss": 0.0002,
+      "step": 4120
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.75089605734767e-06,
+      "loss": 0.0002,
+      "step": 4121
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.728494623655915e-06,
+      "loss": 0.0002,
+      "step": 4122
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.706093189964159e-06,
+      "loss": 0.0006,
+      "step": 4123
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.6836917562724e-06,
+      "loss": 0.0003,
+      "step": 4124
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.661290322580646e-06,
+      "loss": 0.0003,
+      "step": 4125
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 7.63888888888889e-06,
+      "loss": 0.0002,
+      "step": 4126
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.616487455197132e-06,
+      "loss": 0.0003,
+      "step": 4127
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.594086021505377e-06,
+      "loss": 0.0002,
+      "step": 4128
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.571684587813621e-06,
+      "loss": 0.0003,
+      "step": 4129
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.549283154121864e-06,
+      "loss": 0.0007,
+      "step": 4130
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 7.526881720430108e-06,
+      "loss": 0.0002,
+      "step": 4131
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.504480286738351e-06,
+      "loss": 0.0004,
+      "step": 4132
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.4820788530465954e-06,
+      "loss": 0.0002,
+      "step": 4133
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.459677419354839e-06,
+      "loss": 0.0003,
+      "step": 4134
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.437275985663082e-06,
+      "loss": 0.0002,
+      "step": 4135
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.414874551971326e-06,
+      "loss": 0.0003,
+      "step": 4136
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 7.392473118279571e-06,
+      "loss": 0.0002,
+      "step": 4137
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.370071684587813e-06,
+      "loss": 0.0008,
+      "step": 4138
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.347670250896058e-06,
+      "loss": 0.0002,
+      "step": 4139
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.325268817204302e-06,
+      "loss": 0.0003,
+      "step": 4140
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.302867383512545e-06,
+      "loss": 0.0019,
+      "step": 4141
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.2804659498207885e-06,
+      "loss": 0.0002,
+      "step": 4142
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 7.258064516129033e-06,
+      "loss": 0.0003,
+      "step": 4143
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.235663082437276e-06,
+      "loss": 0.0004,
+      "step": 4144
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.21326164874552e-06,
+      "loss": 0.0017,
+      "step": 4145
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.190860215053764e-06,
+      "loss": 0.4337,
+      "step": 4146
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.168458781362007e-06,
+      "loss": 0.0004,
+      "step": 4147
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 7.146057347670252e-06,
+      "loss": 0.0006,
+      "step": 4148
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.1236559139784956e-06,
+      "loss": 0.0002,
+      "step": 4149
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.101254480286739e-06,
+      "loss": 0.0002,
+      "step": 4150
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.0788530465949824e-06,
+      "loss": 0.0002,
+      "step": 4151
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.056451612903227e-06,
+      "loss": 0.0004,
+      "step": 4152
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.034050179211469e-06,
+      "loss": 0.0001,
+      "step": 4153
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 7.011648745519714e-06,
+      "loss": 0.0002,
+      "step": 4154
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.989247311827957e-06,
+      "loss": 0.2013,
+      "step": 4155
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.966845878136201e-06,
+      "loss": 0.0002,
+      "step": 4156
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.0002,
+      "step": 4157
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.922043010752688e-06,
+      "loss": 0.0005,
+      "step": 4158
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 6.8996415770609325e-06,
+      "loss": 0.0007,
+      "step": 4159
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.877240143369176e-06,
+      "loss": 0.0002,
+      "step": 4160
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.854838709677419e-06,
+      "loss": 0.0002,
+      "step": 4161
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.832437275985663e-06,
+      "loss": 0.0002,
+      "step": 4162
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.810035842293908e-06,
+      "loss": 0.0006,
+      "step": 4163
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.78763440860215e-06,
+      "loss": 0.0005,
+      "step": 4164
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 6.765232974910395e-06,
+      "loss": 0.0004,
+      "step": 4165
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.742831541218639e-06,
+      "loss": 0.0002,
+      "step": 4166
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.720430107526882e-06,
+      "loss": 0.0002,
+      "step": 4167
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.698028673835126e-06,
+      "loss": 0.0003,
+      "step": 4168
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.67562724014337e-06,
+      "loss": 0.0004,
+      "step": 4169
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.6532258064516125e-06,
+      "loss": 0.0003,
+      "step": 4170
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 6.630824372759857e-06,
+      "loss": 0.0002,
+      "step": 4171
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.608422939068101e-06,
+      "loss": 0.0005,
+      "step": 4172
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.586021505376344e-06,
+      "loss": 0.0003,
+      "step": 4173
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.563620071684588e-06,
+      "loss": 0.0003,
+      "step": 4174
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.541218637992833e-06,
+      "loss": 0.0003,
+      "step": 4175
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 6.518817204301076e-06,
+      "loss": 0.0004,
+      "step": 4176
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.4964157706093195e-06,
+      "loss": 0.0002,
+      "step": 4177
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.4740143369175625e-06,
+      "loss": 0.0004,
+      "step": 4178
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.451612903225806e-06,
+      "loss": 0.0002,
+      "step": 4179
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.429211469534051e-06,
+      "loss": 0.0002,
+      "step": 4180
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.406810035842293e-06,
+      "loss": 0.0002,
+      "step": 4181
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 6.384408602150538e-06,
+      "loss": 0.0002,
+      "step": 4182
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.362007168458782e-06,
+      "loss": 0.0002,
+      "step": 4183
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.339605734767025e-06,
+      "loss": 0.0004,
+      "step": 4184
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.317204301075269e-06,
+      "loss": 0.0002,
+      "step": 4185
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.294802867383513e-06,
+      "loss": 0.0003,
+      "step": 4186
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 6.2724014336917564e-06,
+      "loss": 0.0001,
+      "step": 4187
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.25e-06,
+      "loss": 0.0004,
+      "step": 4188
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.227598566308244e-06,
+      "loss": 0.0006,
+      "step": 4189
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.205197132616488e-06,
+      "loss": 0.0004,
+      "step": 4190
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.182795698924732e-06,
+      "loss": 0.0005,
+      "step": 4191
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.160394265232975e-06,
+      "loss": 0.0003,
+      "step": 4192
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 6.13799283154122e-06,
+      "loss": 0.0065,
+      "step": 4193
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.115591397849463e-06,
+      "loss": 0.0003,
+      "step": 4194
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.0931899641577065e-06,
+      "loss": 0.0003,
+      "step": 4195
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.0707885304659495e-06,
+      "loss": 0.0004,
+      "step": 4196
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.048387096774194e-06,
+      "loss": 0.0003,
+      "step": 4197
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 6.025985663082437e-06,
+      "loss": 0.0003,
+      "step": 4198
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 6.003584229390681e-06,
+      "loss": 0.0002,
+      "step": 4199
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.981182795698925e-06,
+      "loss": 0.0002,
+      "step": 4200
+    },
+    {
+      "epoch": 7.53,
+      "eval_accuracy": 0.7071428298950195,
+      "eval_loss": 1.679403305053711,
+      "eval_runtime": 74.0335,
+      "eval_samples_per_second": 3.782,
+      "eval_steps_per_second": 3.782,
+      "step": 4200
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.958781362007169e-06,
+      "loss": 0.0003,
+      "step": 4201
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.936379928315413e-06,
+      "loss": 0.0004,
+      "step": 4202
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.9139784946236566e-06,
+      "loss": 0.0002,
+      "step": 4203
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 5.8915770609319e-06,
+      "loss": 0.0003,
+      "step": 4204
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.8691756272401434e-06,
+      "loss": 0.0003,
+      "step": 4205
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.846774193548387e-06,
+      "loss": 0.0002,
+      "step": 4206
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.824372759856631e-06,
+      "loss": 0.0002,
+      "step": 4207
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.801971326164875e-06,
+      "loss": 0.003,
+      "step": 4208
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.779569892473118e-06,
+      "loss": 0.0002,
+      "step": 4209
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 5.757168458781363e-06,
+      "loss": 0.0002,
+      "step": 4210
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.734767025089606e-06,
+      "loss": 0.0113,
+      "step": 4211
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.71236559139785e-06,
+      "loss": 0.0002,
+      "step": 4212
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.6899641577060935e-06,
+      "loss": 0.0003,
+      "step": 4213
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.667562724014337e-06,
+      "loss": 0.0003,
+      "step": 4214
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 5.64516129032258e-06,
+      "loss": 0.0002,
+      "step": 4215
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.622759856630825e-06,
+      "loss": 0.0003,
+      "step": 4216
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.600358422939068e-06,
+      "loss": 0.0002,
+      "step": 4217
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.577956989247312e-06,
+      "loss": 0.0002,
+      "step": 4218
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.0003,
+      "step": 4219
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.5331541218638e-06,
+      "loss": 0.0002,
+      "step": 4220
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 5.5107526881720435e-06,
+      "loss": 0.0003,
+      "step": 4221
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.4883512544802866e-06,
+      "loss": 0.0003,
+      "step": 4222
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.465949820788531e-06,
+      "loss": 0.0003,
+      "step": 4223
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.443548387096774e-06,
+      "loss": 0.0003,
+      "step": 4224
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.421146953405018e-06,
+      "loss": 0.0003,
+      "step": 4225
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 5.398745519713262e-06,
+      "loss": 0.0005,
+      "step": 4226
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.376344086021506e-06,
+      "loss": 0.0004,
+      "step": 4227
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.353942652329749e-06,
+      "loss": 0.0008,
+      "step": 4228
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.331541218637993e-06,
+      "loss": 0.0002,
+      "step": 4229
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.309139784946237e-06,
+      "loss": 0.0003,
+      "step": 4230
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.2867383512544805e-06,
+      "loss": 0.0004,
+      "step": 4231
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 5.264336917562724e-06,
+      "loss": 0.0002,
+      "step": 4232
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.241935483870968e-06,
+      "loss": 0.0003,
+      "step": 4233
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.219534050179212e-06,
+      "loss": 0.0005,
+      "step": 4234
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.197132616487455e-06,
+      "loss": 0.0003,
+      "step": 4235
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.1747311827957e-06,
+      "loss": 0.0003,
+      "step": 4236
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.152329749103943e-06,
+      "loss": 0.0002,
+      "step": 4237
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 5.129928315412187e-06,
+      "loss": 0.0003,
+      "step": 4238
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.1075268817204305e-06,
+      "loss": 0.0002,
+      "step": 4239
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.085125448028674e-06,
+      "loss": 0.0002,
+      "step": 4240
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.062724014336917e-06,
+      "loss": 3.8829,
+      "step": 4241
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.040322580645161e-06,
+      "loss": 0.0001,
+      "step": 4242
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 5.017921146953405e-06,
+      "loss": 0.0002,
+      "step": 4243
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.995519713261649e-06,
+      "loss": 0.0009,
+      "step": 4244
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.973118279569893e-06,
+      "loss": 0.0002,
+      "step": 4245
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.950716845878137e-06,
+      "loss": 0.0003,
+      "step": 4246
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.928315412186381e-06,
+      "loss": 0.0004,
+      "step": 4247
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.905913978494624e-06,
+      "loss": 0.0002,
+      "step": 4248
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 4.883512544802868e-06,
+      "loss": 0.0003,
+      "step": 4249
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.861111111111111e-06,
+      "loss": 0.0004,
+      "step": 4250
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.838709677419355e-06,
+      "loss": 0.0004,
+      "step": 4251
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.816308243727598e-06,
+      "loss": 0.0002,
+      "step": 4252
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.793906810035843e-06,
+      "loss": 0.0005,
+      "step": 4253
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 4.771505376344086e-06,
+      "loss": 0.0004,
+      "step": 4254
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.74910394265233e-06,
+      "loss": 0.0002,
+      "step": 4255
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.726702508960574e-06,
+      "loss": 0.0003,
+      "step": 4256
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.7043010752688175e-06,
+      "loss": 0.0003,
+      "step": 4257
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.6818996415770606e-06,
+      "loss": 0.0003,
+      "step": 4258
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.659498207885305e-06,
+      "loss": 0.0004,
+      "step": 4259
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 4.637096774193548e-06,
+      "loss": 0.0002,
+      "step": 4260
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.614695340501792e-06,
+      "loss": 0.0002,
+      "step": 4261
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.592293906810036e-06,
+      "loss": 0.0002,
+      "step": 4262
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.56989247311828e-06,
+      "loss": 0.0001,
+      "step": 4263
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.547491039426524e-06,
+      "loss": 0.0003,
+      "step": 4264
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 4.525089605734767e-06,
+      "loss": 0.0006,
+      "step": 4265
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.5026881720430115e-06,
+      "loss": 0.0003,
+      "step": 4266
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.4802867383512545e-06,
+      "loss": 0.0002,
+      "step": 4267
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.457885304659498e-06,
+      "loss": 0.0004,
+      "step": 4268
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.435483870967742e-06,
+      "loss": 2.502,
+      "step": 4269
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.413082437275986e-06,
+      "loss": 0.0003,
+      "step": 4270
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 4.390681003584229e-06,
+      "loss": 0.0002,
+      "step": 4271
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.368279569892474e-06,
+      "loss": 0.0004,
+      "step": 4272
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.345878136200717e-06,
+      "loss": 0.008,
+      "step": 4273
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.323476702508961e-06,
+      "loss": 0.0003,
+      "step": 4274
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.3010752688172045e-06,
+      "loss": 0.0012,
+      "step": 4275
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.278673835125448e-06,
+      "loss": 0.0003,
+      "step": 4276
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 4.256272401433692e-06,
+      "loss": 0.1936,
+      "step": 4277
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.233870967741935e-06,
+      "loss": 0.0003,
+      "step": 4278
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.21146953405018e-06,
+      "loss": 0.0003,
+      "step": 4279
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.189068100358423e-06,
+      "loss": 0.0002,
+      "step": 4280
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.0002,
+      "step": 4281
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 4.144265232974911e-06,
+      "loss": 0.0003,
+      "step": 4282
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.121863799283155e-06,
+      "loss": 0.0004,
+      "step": 4283
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.099462365591398e-06,
+      "loss": 0.0002,
+      "step": 4284
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.077060931899642e-06,
+      "loss": 0.0002,
+      "step": 4285
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.054659498207885e-06,
+      "loss": 0.0002,
+      "step": 4286
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.032258064516129e-06,
+      "loss": 0.0003,
+      "step": 4287
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 4.009856630824372e-06,
+      "loss": 0.0002,
+      "step": 4288
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.987455197132617e-06,
+      "loss": 0.0003,
+      "step": 4289
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.96505376344086e-06,
+      "loss": 0.0002,
+      "step": 4290
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.942652329749104e-06,
+      "loss": 0.0003,
+      "step": 4291
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.920250896057348e-06,
+      "loss": 0.0002,
+      "step": 4292
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 3.8978494623655915e-06,
+      "loss": 0.0041,
+      "step": 4293
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.875448028673835e-06,
+      "loss": 0.0003,
+      "step": 4294
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.853046594982079e-06,
+      "loss": 0.0002,
+      "step": 4295
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.830645161290323e-06,
+      "loss": 0.0004,
+      "step": 4296
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.808243727598566e-06,
+      "loss": 0.0002,
+      "step": 4297
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.7858422939068104e-06,
+      "loss": 0.0004,
+      "step": 4298
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 3.763440860215054e-06,
+      "loss": 0.0002,
+      "step": 4299
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.7410394265232977e-06,
+      "loss": 0.0003,
+      "step": 4300
+    },
+    {
+      "epoch": 7.71,
+      "eval_accuracy": 0.7107142806053162,
+      "eval_loss": 1.6645880937576294,
+      "eval_runtime": 73.802,
+      "eval_samples_per_second": 3.794,
+      "eval_steps_per_second": 3.794,
+      "step": 4300
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.718637992831541e-06,
+      "loss": 0.0006,
+      "step": 4301
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.6962365591397855e-06,
+      "loss": 0.0002,
+      "step": 4302
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.673835125448029e-06,
+      "loss": 0.0003,
+      "step": 4303
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 3.6514336917562723e-06,
+      "loss": 0.0021,
+      "step": 4304
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.6290322580645166e-06,
+      "loss": 0.0002,
+      "step": 4305
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.60663082437276e-06,
+      "loss": 0.0002,
+      "step": 4306
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.5842293906810035e-06,
+      "loss": 0.0003,
+      "step": 4307
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.5618279569892478e-06,
+      "loss": 0.0004,
+      "step": 4308
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.5394265232974912e-06,
+      "loss": 0.0004,
+      "step": 4309
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 3.5170250896057347e-06,
+      "loss": 0.0002,
+      "step": 4310
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.4946236559139785e-06,
+      "loss": 0.0004,
+      "step": 4311
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.4722222222222224e-06,
+      "loss": 0.0002,
+      "step": 4312
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.4498207885304663e-06,
+      "loss": 0.0003,
+      "step": 4313
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.4274193548387097e-06,
+      "loss": 0.0002,
+      "step": 4314
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.405017921146954e-06,
+      "loss": 0.0002,
+      "step": 4315
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 3.3826164874551974e-06,
+      "loss": 0.0002,
+      "step": 4316
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.360215053763441e-06,
+      "loss": 0.0002,
+      "step": 4317
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.337813620071685e-06,
+      "loss": 0.0002,
+      "step": 4318
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.3154121863799286e-06,
+      "loss": 0.0004,
+      "step": 4319
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.293010752688172e-06,
+      "loss": 0.0003,
+      "step": 4320
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 3.2706093189964163e-06,
+      "loss": 0.0003,
+      "step": 4321
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.2482078853046597e-06,
+      "loss": 0.0235,
+      "step": 4322
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.225806451612903e-06,
+      "loss": 0.0003,
+      "step": 4323
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.2034050179211466e-06,
+      "loss": 0.0002,
+      "step": 4324
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.181003584229391e-06,
+      "loss": 0.0003,
+      "step": 4325
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.1586021505376344e-06,
+      "loss": 0.0004,
+      "step": 4326
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 3.1362007168458782e-06,
+      "loss": 0.0003,
+      "step": 4327
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.113799283154122e-06,
+      "loss": 0.0281,
+      "step": 4328
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.091397849462366e-06,
+      "loss": 0.0037,
+      "step": 4329
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.06899641577061e-06,
+      "loss": 0.0003,
+      "step": 4330
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.0465949820788532e-06,
+      "loss": 0.0003,
+      "step": 4331
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 3.024193548387097e-06,
+      "loss": 0.0002,
+      "step": 4332
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 3.0017921146953406e-06,
+      "loss": 0.0003,
+      "step": 4333
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.9793906810035844e-06,
+      "loss": 0.0002,
+      "step": 4334
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.9569892473118283e-06,
+      "loss": 0.0003,
+      "step": 4335
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.9345878136200717e-06,
+      "loss": 0.0003,
+      "step": 4336
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.9121863799283156e-06,
+      "loss": 0.0003,
+      "step": 4337
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.889784946236559e-06,
+      "loss": 0.0003,
+      "step": 4338
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.867383512544803e-06,
+      "loss": 0.0003,
+      "step": 4339
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.8449820788530467e-06,
+      "loss": 0.0002,
+      "step": 4340
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.82258064516129e-06,
+      "loss": 0.0002,
+      "step": 4341
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.800179211469534e-06,
+      "loss": 0.013,
+      "step": 4342
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 0.0003,
+      "step": 4343
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.7553763440860218e-06,
+      "loss": 0.0003,
+      "step": 4344
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.7329749103942656e-06,
+      "loss": 0.0002,
+      "step": 4345
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.710573476702509e-06,
+      "loss": 0.0008,
+      "step": 4346
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.688172043010753e-06,
+      "loss": 0.0002,
+      "step": 4347
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.6657706093189964e-06,
+      "loss": 0.0002,
+      "step": 4348
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.6433691756272402e-06,
+      "loss": 0.0002,
+      "step": 4349
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.620967741935484e-06,
+      "loss": 0.0002,
+      "step": 4350
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.5985663082437275e-06,
+      "loss": 0.0002,
+      "step": 4351
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.5761648745519714e-06,
+      "loss": 0.0002,
+      "step": 4352
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.5537634408602153e-06,
+      "loss": 0.0002,
+      "step": 4353
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.5313620071684587e-06,
+      "loss": 0.0022,
+      "step": 4354
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.5089605734767026e-06,
+      "loss": 0.0026,
+      "step": 4355
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.4865591397849464e-06,
+      "loss": 0.0003,
+      "step": 4356
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.4641577060931903e-06,
+      "loss": 0.0002,
+      "step": 4357
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.441756272401434e-06,
+      "loss": 0.0003,
+      "step": 4358
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.4193548387096776e-06,
+      "loss": 0.0008,
+      "step": 4359
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.3969534050179215e-06,
+      "loss": 0.0002,
+      "step": 4360
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.374551971326165e-06,
+      "loss": 0.0004,
+      "step": 4361
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3521505376344088e-06,
+      "loss": 0.0002,
+      "step": 4362
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3297491039426526e-06,
+      "loss": 0.0004,
+      "step": 4363
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.307347670250896e-06,
+      "loss": 0.0003,
+      "step": 4364
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.28494623655914e-06,
+      "loss": 0.0002,
+      "step": 4365
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.2625448028673834e-06,
+      "loss": 0.0003,
+      "step": 4366
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.2401433691756272e-06,
+      "loss": 0.0003,
+      "step": 4367
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.217741935483871e-06,
+      "loss": 0.0004,
+      "step": 4368
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.1953405017921145e-06,
+      "loss": 0.0004,
+      "step": 4369
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.1729390681003584e-06,
+      "loss": 1.0802,
+      "step": 4370
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.1505376344086023e-06,
+      "loss": 0.0002,
+      "step": 4371
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.128136200716846e-06,
+      "loss": 0.0004,
+      "step": 4372
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.10573476702509e-06,
+      "loss": 0.0002,
+      "step": 4373
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.0833333333333334e-06,
+      "loss": 0.0019,
+      "step": 4374
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.0609318996415773e-06,
+      "loss": 0.0002,
+      "step": 4375
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.038530465949821e-06,
+      "loss": 0.0004,
+      "step": 4376
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.0161290322580646e-06,
+      "loss": 0.0004,
+      "step": 4377
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.9937275985663085e-06,
+      "loss": 0.0003,
+      "step": 4378
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.971326164874552e-06,
+      "loss": 0.0002,
+      "step": 4379
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.9489247311827958e-06,
+      "loss": 0.0003,
+      "step": 4380
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.9265232974910396e-06,
+      "loss": 0.0002,
+      "step": 4381
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.904121863799283e-06,
+      "loss": 0.0003,
+      "step": 4382
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.881720430107527e-06,
+      "loss": 0.0003,
+      "step": 4383
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.8593189964157706e-06,
+      "loss": 0.0012,
+      "step": 4384
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.8369175627240144e-06,
+      "loss": 0.0002,
+      "step": 4385
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.8145161290322583e-06,
+      "loss": 0.0001,
+      "step": 4386
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.7921146953405017e-06,
+      "loss": 0.0003,
+      "step": 4387
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.7697132616487456e-06,
+      "loss": 0.0002,
+      "step": 4388
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.7473118279569893e-06,
+      "loss": 0.5562,
+      "step": 4389
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.7249103942652331e-06,
+      "loss": 0.0004,
+      "step": 4390
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.702508960573477e-06,
+      "loss": 0.0012,
+      "step": 4391
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.6801075268817204e-06,
+      "loss": 2.1194,
+      "step": 4392
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.6577060931899643e-06,
+      "loss": 0.0003,
+      "step": 4393
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.6353046594982082e-06,
+      "loss": 0.0003,
+      "step": 4394
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.6129032258064516e-06,
+      "loss": 0.0002,
+      "step": 4395
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.5905017921146955e-06,
+      "loss": 0.0003,
+      "step": 4396
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.5681003584229391e-06,
+      "loss": 0.0002,
+      "step": 4397
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.545698924731183e-06,
+      "loss": 0.0002,
+      "step": 4398
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.5232974910394266e-06,
+      "loss": 0.0036,
+      "step": 4399
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.5008960573476703e-06,
+      "loss": 0.0037,
+      "step": 4400
+    },
+    {
+      "epoch": 7.89,
+      "eval_accuracy": 0.7250000238418579,
+      "eval_loss": 1.5962432622909546,
+      "eval_runtime": 73.6249,
+      "eval_samples_per_second": 3.803,
+      "eval_steps_per_second": 3.803,
+      "step": 4400
+    }
+  ],
+  "max_steps": 4464,
+  "num_train_epochs": 8,
+  "total_flos": 1.6224600141321426e+18,
+  "trial_name": null,
+  "trial_params": null
+}